Fix some typos in the german manpage, correct the encoding of "ß".
[maxima/cygwin.git] / src / numth.lisp
blobc8dc56b6209932f6cfcca34df8b7913a4d984fa5
1 ;;; -*- Mode: Lisp; Package: Maxima; Syntax: Common-Lisp; Base: 10 -*- ;;;;
2 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3 ;;; The data in this file contains enhancments. ;;;;;
4 ;;; ;;;;;
5 ;;; Copyright (c) 1984,1987 by William Schelter,University of Texas ;;;;;
6 ;;; All rights reserved ;;;;;
7 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
8 ;;; (c) Copyright 1981 Massachusetts Institute of Technology ;;;
9 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
11 (in-package :maxima)
13 ;;; *****************************************************************
14 ;;; ***** NUMTH ******* VARIOUS NUMBER THEORY FUNCTIONS *************
15 ;;; *****************************************************************
17 (macsyma-module numth)
19 (declare-top (special $intfaclim modulus))
21 (load-macsyma-macros rzmac)
23 ;;; Sum of divisors and Totient functions
25 ;; compute the sum of the k'th powers of the divisors of the absolute
26 ;; value of n
28 (defmfun $divsum (n &optional (k 1))
29 (if (and (integerp k) (integerp n))
30 (let ((n (abs n)))
31 (if (minusp k)
32 (list '(rat) (divsum n (- k)) (expt n (- k)))
33 (divsum n k)))
34 (list '($divsum) n k)))
36 ;; divsum assumes its arguments to be non-negative integers.
38 (defun divsum (n k)
39 (declare (type (integer 0) n k) (optimize (speed 3)))
40 (let (($intfaclim nil)) ;get-factor-list returns list
41 ;of (p_rime e_xponent) pairs
42 ;e is a fixnum
43 (cond ((<= n 1) 1)
44 ((zerop k) (reduce #'* (get-factor-list n) ; product over e+1
45 :key #'(lambda (pe) (1+ (the fixnum (cadr pe))))))
46 (t (reduce #'* (get-factor-list n) ; product over ((p^k)^(e+1)-1)/(p^k-1)
47 :key #'(lambda (pe)
48 (let ((p (car pe)) (e (cadr pe)))
49 (declare (type (integer 2) p)
50 (type (integer 1 (#.most-positive-fixnum)) e))
51 (let ((tmp (expt p k)))
52 (truncate (1- (expt tmp (1+ e)))
53 (1- tmp))))))))))
55 ;; totient computes the euler totient function
56 ;; i.e. the count of numbers relatively prime to n between 1 and n.
58 (defmfun $totient (n)
59 (if (integerp n)
60 (totient (abs n))
61 (list '($totient) n)))
63 ;; totient assumes its argument to be an integer.
64 ;; the exponents in the prime factorization of n are assumed to be
65 ;; fixnums (anything else would exceed memory).
67 (defun totient (n)
68 (declare (type (integer 0) n) (optimize (speed 3)))
69 (let (($intfaclim nil))
70 (if (< n 2)
72 (reduce #'* (get-factor-list n)
73 :key #'(lambda (pe)
74 (let ((p (car pe)) (e (cadr pe)))
75 (declare (type (integer 2) p)
76 (type (integer 1 (#.most-positive-fixnum)) e))
77 (* (1- p) (expt p (1- e)))))))))
79 ;;; JACOBI symbol and Gaussian factoring
81 ;; $jacobi/jacobi implement the Kronecker-Jacobi symbol, a light
82 ;; generalization of the Legendre/Jacobi symbols.
83 ;; (for a prime b, $jacobi(a b) is the Legendre symbol).
85 ;; The implementation follows algorithm 1.4.10 in 'A Course in
86 ;; Computational Algebraic Number Theory' by H. Cohen
88 (defmfun $jacobi (a b)
89 (if (and (integerp a) (integerp b))
90 (jacobi a b)
91 `(($jacobi) ,a ,b)))
93 (defun jacobi (a b)
94 (declare (integer a b) (optimize (speed 3)))
95 (cond ((zerop b)
96 (if (= 1 (abs a)) 1 0))
97 ((and (evenp a) (evenp b))
100 (let ((k 1)
101 (tab2 (make-array 8 :element-type '(integer -1 1)
102 :initial-contents #(0 1 0 -1 0 -1 0 1))))
103 (declare (type (integer -1 1) k))
105 (loop for v fixnum = 0 then (1+ v) ;remove 2's from b
106 while (evenp b) do (setf b (ash b -1))
107 finally (when (oddp v)
108 (setf k (aref tab2 (logand a 7))))) ; (-1)^(a^2-1)/8
110 (when (minusp b)
111 (setf b (- b))
112 (when (minusp a)
113 (setf k (- k))))
115 (loop
116 (when (zerop a)
117 (return-from jacobi (if (> b 1) 0 k)))
119 (loop for v fixnum = 0 then (1+ v)
120 while (evenp a) do (setf a (ash a -1))
121 finally (when (oddp v)
122 (when (minusp (aref tab2 (logand b 7))); (-1)^(b^2-1)/8
123 (setf k (- k)))))
125 (when (plusp (logand a b 2)) ;compute (-1)^(a-1)(b-1)/4
126 (setf k (- k)))
128 (let ((r (abs a)))
129 (setf a (rem b r))
130 (setf b r)))))))
133 ;; factor over the gaussian primes
135 (declaim (inline gctimes gctime1))
137 (defmfun $gcfactor (n)
138 (let ((n (cdr ($totaldisrep ($bothcoef ($rat ($rectform n) '$%i) '$%i)))))
139 (if (not (and (integerp (first n)) (integerp (second n))))
140 (gcdisp (nreverse n))
141 (loop for (term exp) on (gcfactor (second n) (first n)) by #'cddr
142 with res = () do
143 (push (if (= exp 1)
144 (gcdisp term)
145 (list '(mexpt simp) (gcdisp term) exp))
146 res)
147 finally
148 (return (cond ((null res)
150 ((null (cdr res))
151 (if (mexptp (car res))
152 `((mexpt simp gcfactored) ,@(cdar res))
153 (car res)))
155 `((mtimes simp gcfactored) ,@(nreverse res)))))))))
157 (defun imodp (p)
158 (declare (integer p) (optimize (speed 3)))
159 (cond ((not (= (rem p 4) 1)) nil)
160 ((= (rem p 8) 5) (power-mod 2 (ash (1- p) -2) p))
161 ((= (rem p 24) 17) (power-mod 3 (ash (1- p) -2) p)) ;p=2(mod 3)
162 (t (do ((i 5 (+ i j)) ;p=1(mod 24)
163 (j 2 (- 6 j)))
164 ((= (jacobi i p) -1) (power-mod i (ash (1- p) -2) p))
165 (declare (integer i) (fixnum j))))))
167 (defun psumsq (p)
168 (declare (integer p) (optimize (speed 3)))
169 (if (= p 2)
170 (list 1 1)
171 (let ((x (imodp p)))
172 (if (null x)
174 (do ((sp (isqrt p))
175 (r1 p r2)
176 (r2 x (rem r1 r2)))
177 ((not (> r1 sp)) (list r1 r2))
178 (declare (integer r1 r2)))))))
180 (defun gctimes (a b c d)
181 (declare (integer a b c d) (optimize (speed 3)))
182 (list (- (* a c) (* b d))
183 (+ (* a d) (* b c))))
185 (defun gcdisp (term)
186 (if (atom term)
187 term
188 (let ((rp (car term))
189 (ip (cadr term)))
190 (setq ip (if (= ip 1) '$%i `((mtimes) ,ip $%i)))
191 (if (eql 0 rp)
193 `((mplus) ,rp ,ip)))))
195 (defun gcfactor (a b)
196 (declare (integer a b) (optimize (speed 3)))
197 (when (and (zerop a) (zerop b))
198 (return-from gcfactor (list 0 1)))
199 (let* (($intfaclim nil)
200 (e1 0)
201 (e2 0)
202 (econt 0)
203 (g (gcd a b))
204 (a (truncate a g))
205 (b (truncate b g))
206 (p 0)
207 (nl (cfactorw (+ (* a a) (* b b))))
208 (gl (cfactorw g))
209 (cd nil)
210 (dc nil)
211 (ans nil)
212 (plis nil))
213 (declare (integer e1 e2 econt g a b))
214 (when (= 1 (the integer (car gl))) (setq gl nil))
215 (when (= 1 (the integer (car nl))) (setq nl nil))
216 (tagbody
217 loop
218 (cond ((null gl)
219 (if (null nl)
220 (go ret)
221 (setq p (car nl))))
222 ((null nl)
223 (setq p (car gl)))
225 (setq p (max (the integer (car gl))
226 (the integer (car nl))))))
227 (setq cd (psumsq p))
228 (cond ((null cd)
229 (setq plis (list* p (cadr gl) plis))
230 (setq gl (cddr gl))
231 (go loop))
232 ((equal p (car nl))
233 (cond ((zerop (rem (+ (* a (the integer (car cd))) ; gcremainder
234 (* b (the integer (cadr cd))))
235 (the integer p))) ; remainder(real((a+bi)cd~),p)
236 ; z~ is complex conjugate
237 (setq e1 (cadr nl))
238 (setq dc cd))
240 (setq e2 (cadr nl))
241 (setq dc (reverse cd))))
242 (setq dc (gcexpt dc (the integer (cadr nl)))
243 dc (gctimes a b (the integer (car dc)) (- (the integer (cadr dc))))
244 a (truncate (the integer (car dc)) (the integer p))
245 b (truncate (the integer (cadr dc)) (the integer p))
246 nl (cddr nl))))
247 (when (equal p (car gl))
248 (incf econt (the integer (cadr gl)))
249 (cond ((equal p 2)
250 (incf e1 (* 2 (the integer (cadr gl)))))
252 (incf e1 (the integer (cadr gl)))
253 (incf e2 (the integer (cadr gl)))))
254 (setq gl (cddr gl)))
255 (unless (zerop e1)
256 (setq ans (list* cd e1 ans))
257 (setq e1 0))
258 (unless (zerop e2)
259 (setq ans (list* (reverse cd) e2 ans))
260 (setq e2 0))
261 (go loop)
263 ret
264 (let* ((cd (gcexpt (list 0 -1) (rem econt 4)))
265 (rcd (first cd))
266 (icd (second cd))
267 (l (mapcar #'signum (gctimes a b rcd icd)))
268 (r (first l))
269 (i (second l)))
270 (declare (integer r i rcd icd))
271 (when (or (= r -1) (= i -1))
272 (setq plis (list* -1 1 plis)))
273 (when (zerop r)
274 (setq ans (list* '(0 1) 1 ans)))
275 (return-from gcfactor (append plis ans))))))
277 (defun gcsqr (a)
278 (declare (optimize (speed 3)))
279 (let ((r (first a))
280 (i (second a)))
281 (declare (integer r i))
282 (list (+ (* r r) (* i i)) (ash (* r i) 1))))
284 (defun gctime1 (a b)
285 (gctimes (car a) (cadr a) (car b) (cadr b)))
287 (defun gcexpt (a n)
288 (cond ((zerop n) '(1 0))
289 ((= n 1) a)
290 ((evenp n) (gcexpt (gcsqr a) (ash n -1)))
291 (t (gctime1 a (gcexpt (gcsqr a) (ash n -1))))))
293 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
295 ;; Maxima functions in (Z/nZ)*
297 ;; zn_order, zn_primroot_p, zn_primroot, zn_log,
298 ;; chinese, zn_characteristic_factors, zn_factor_generators, zn_nth_root,
299 ;; zn_add_table, zn_mult_table, zn_power_table
301 ;; 2012 - 2020, Volker van Nek
304 ;; Maxima option variables:
305 (defmvar $zn_primroot_limit 1000 "Upper bound for `zn_primroot'." fixnum)
306 (defmvar $zn_primroot_verbose nil "Print message when `zn_primroot_limit' is reached." boolean)
307 (defmvar $zn_primroot_pretest nil "`zn_primroot' pretests whether (Z/nZ)* is cyclic." boolean)
309 (declaim (inline zn-quo))
310 (defun zn-quo (n d p)
311 (declare (integer n d p) (optimize (speed 3)))
312 (mod (* n (inv-mod d p)) p) )
314 ;; compute the order of x in (Z/nZ)*
316 ;; optional argument: ifactors of totient(n) as returned in Maxima by
317 ;; block([factors_only:false], ifactors(totient(n)))
318 ;; e.g. [[2, 3], [3, 1], ... ]
320 (defmfun $zn_order (x n &optional fs-phi)
321 (unless (and (integerp x) (integerp n))
322 (return-from $zn_order
323 (if fs-phi
324 (list '($zn_order) x n fs-phi)
325 (list '($zn_order) x n) )))
326 (setq x (mod x n))
327 (cond
328 ((= 0 x) nil)
329 ((= 1 x) (if (= n 1) nil 1))
330 ((/= 1 (gcd x n)) nil)
332 (if fs-phi
333 (if (and ($listp fs-phi) ($listp (cadr fs-phi)))
334 (progn
335 (setq fs-phi (mapcar #'cdr (cdr fs-phi))) ; Lispify fs-phi
336 (setq fs-phi (cons (totient-from-factors fs-phi) fs-phi)) )
337 (gf-merror (intl:gettext
338 "Third argument to `zn_order' must be of the form [[p1, e1], ..., [pk, ek]]." )))
339 (setq fs-phi (totient-with-factors n)) )
340 (zn-order x
342 (car fs-phi) ;; phi
343 (cdr fs-phi)) ))) ;; factors of phi with multiplicity
345 (defun zn_order (x n phi fs-phi)
346 (format t "`zn_order' is deprecated. ~%Please use `zn-order' instead.~%" )
347 (zn-order x n phi fs-phi) )
349 ;; compute order of x as a divisor of the known group order
351 (defun zn-order (x n ord fs-ord)
352 (let (p e z)
353 (dolist (f fs-ord ord)
354 (setq p (car f) e (cadr f)
355 ord (truncate ord (expt p e))
356 z (power-mod x ord n) )
357 ;; ord(z) = p^i, i from [0,e]
358 ;; replace p^e in ord by p^i : x^(ord*p^i/p^e) = 1
359 (do ()
360 ((= z 1))
361 (setq ord (* ord p))
362 (when (= e 1) (return))
363 (decf e)
364 (setq z (power-mod z p n)) ))))
367 ;; compute totient (euler-phi) of n and its factors in one function
369 ;; returns a list of the form (phi ((p1 e1) ... (pk ek)))
371 (defun totient-with-factors (n)
372 (let (($intfaclim) (phi 1) fs-n (fs) p e (fs-phi) g)
373 (setq fs-n (get-factor-list n))
374 (dolist (f fs-n fs)
375 (setq p (car f) e (cadr f))
376 (setq phi (* phi (1- p) (expt p (1- e))))
377 (when (> e 1) (setq fs (cons `(,p ,(1- e)) fs)))
378 (setq fs (append (get-factor-list (1- p)) fs)) )
379 (setq fs (copy-tree fs)) ;; this deep copy is a workaround to avoid references
380 ;; to the list returned by ifactor.lisp/get-factor-list.
381 ;; see bug 3510983
382 (setq fs (sort fs #'< :key #'car))
383 (setq g (car fs))
384 (dolist (f (cdr fs) (cons phi (reverse (cons g fs-phi))))
385 (if (= (car f) (car g))
386 (incf (cadr g) (cadr f)) ;; assignment
387 (progn
388 (setq fs-phi (cons g fs-phi))
389 (setq g f) ))) ))
391 ;; recompute totient from given factors
393 ;; fs-phi: factors of totient with multiplicity: ((p1 e1) ... (pk ek))
395 (defun totient-from-factors (fs-phi)
396 (let ((phi 1) p e)
397 (dolist (f fs-phi phi)
398 (setq p (car f) e (cadr f))
399 (setq phi (* phi (expt p e))) )))
402 ;; for n > 2 is x a primitive root modulo n
403 ;; when n does not divide x
404 ;; and for all prime factors p of phi = totient(n)
405 ;; x^(phi/p) mod n # 1
407 ;; optional argument: ifactors of totient(n)
409 (defmfun $zn_primroot_p (x n &optional fs-phi)
410 (unless (and (integerp x) (integerp n))
411 (return-from $zn_primroot_p
412 (if fs-phi
413 (list '($zn_primroot_p) x n fs-phi)
414 (list '($zn_primroot_p) x n) )))
415 (setq x (mod x n))
416 (cond
417 ((= 0 x) nil)
418 ((= 1 x) (if (= n 2) t nil))
419 ((<= n 2) nil)
421 (if fs-phi
422 (if (and ($listp fs-phi) ($listp (cadr fs-phi)))
423 (progn
424 (setq fs-phi (mapcar #'cdr (cdr fs-phi))) ; Lispify fs-phi
425 (setq fs-phi (cons (totient-from-factors fs-phi) fs-phi)) )
426 (gf-merror (intl:gettext
427 "Third argument to `zn_primroot_p' must be of the form [[p1, e1], ..., [pk, ek]]." )))
428 (setq fs-phi (totient-with-factors n)) )
429 (zn-primroot-p x n
430 (car fs-phi) ;; phi
431 (mapcar #'car (cdr fs-phi))) ))) ;; factors only (omitting multiplicity)
433 (defun zn-primroot-p (x n phi fs-phi)
434 (unless (= 1 (gcd x n))
435 (return-from zn-primroot-p nil) )
436 (dolist (p fs-phi t)
437 (when (= 1 (power-mod x (truncate phi p) n))
438 (return-from zn-primroot-p nil) )))
441 ;; find the smallest primitive root modulo n
443 ;; optional argument: ifactors of totient(n)
445 (defmfun $zn_primroot (n &optional fs-phi)
446 (unless (integerp n)
447 (return-from $zn_primroot
448 (if fs-phi
449 (list '($zn_primroot) n fs-phi)
450 (list '($zn_primroot) n) )))
451 (cond
452 ((<= n 1) nil)
453 ((= n 2) 1)
455 (when $zn_primroot_pretest
456 (unless (cyclic-p n)
457 (return-from $zn_primroot nil) ))
458 (if fs-phi
459 (if (and ($listp fs-phi) ($listp (cadr fs-phi)))
460 (progn
461 (setq fs-phi (mapcar #'cdr (cdr fs-phi))) ; Lispify fs-phi
462 (setq fs-phi (cons (totient-from-factors fs-phi) fs-phi)) )
463 (gf-merror (intl:gettext
464 "Second argument to `zn_primroot' must be of the form [[p1, e1], ..., [pk, ek]]." )))
465 (setq fs-phi (totient-with-factors n)) )
466 (zn-primroot n
467 (car fs-phi) ;; phi
468 (mapcar #'car (cdr fs-phi))) ))) ;; factors only (omitting multiplicity)
470 ;; (Z/nZ)* is cyclic if n = 2, 4, p^k or 2*p^k where p prime > 2
471 (defun cyclic-p (n)
472 (prog ()
473 (when (< n 2) (return))
474 (when (< n 8) (return t)) ;; 2,3,4,5,2*3,7
475 (when (evenp n) ;; 2*p^k
476 (setq n (ash n -1)) ;; -> p^k
477 (when (evenp n) (return)) )
478 (let (($intfaclim) fs (len 0))
479 (multiple-value-setq (n fs) (get-small-factors n))
480 (when fs (setq len (length fs)))
481 (when (= 1 n) (return (= 1 len)))
482 (when (> len 0) (return))
483 (when (primep n) (return t))
484 (setq fs (convert-list (get-large-factors n)))
485 (return (= 1 (length fs))) )))
487 (defun zn-primroot (n phi fs-phi)
488 (do ((i 2 (1+ i)))
489 ((= i n) nil)
490 (when (zn-primroot-p i n phi fs-phi)
491 (return i) )
492 (when (= i $zn_primroot_limit)
493 (when $zn_primroot_verbose
494 (format t "`zn_primroot' stopped at zn_primroot_limit = ~A~%" $zn_primroot_limit) )
495 (return nil) )))
498 ;; Chinese Remainder Theorem
500 (defmfun $chinese (rems mods &optional (return-lcm? nil))
501 (cond
502 ((not (and ($listp rems) ($listp mods)))
503 (list '($chinese) rems mods) )
504 ((let ((lr ($length rems)) (lm ($length mods)))
505 (or (= 0 lr) (= 0 lm) (/= lr lm)) )
506 (gf-merror (intl:gettext
507 "Unsuitable arguments to `chinese': ~m ~m" ) rems mods ))
508 ((notevery #'integerp (setq rems (cdr rems)))
509 (list '($chinese) (cons '(mlist simp) rems) mods) )
510 ((notevery #'integerp (setq mods (cdr mods)))
511 (list '($chinese) (cons '(mlist simp) rems) (cons '(mlist simp) mods)) )
512 ((eql return-lcm? '$lcm)
513 (cons '(mlist simp) (chinese rems mods)) )
515 (car (chinese rems mods)) )))
517 (defun chinese (rems mods)
518 (if (= 1 (length mods))
519 (list (car rems) (car mods))
520 (let ((rp (car rems))
521 (p (car mods))
522 (rq-q (chinese (cdr rems) (cdr mods))) )
523 (when rq-q
524 (let* ((rq (car rq-q))
525 (q (cadr rq-q))
526 (gc (zn-gcdex2 q p))
527 (g (car gc)) ;; gcd
528 (c (cadr gc)) ) ;; CRT-coefficient
529 (cond
530 ((= 1 g) ;; coprime moduli
531 (let* ((h (mod (* (- rp rq) c) p))
532 (x (+ (* h q) rq)) )
533 (list x (* p q)) ))
534 ((= 0 (mod (- rp rq) g)) ;; ensures unique solution
535 (let* ((h (* (- rp rq) c))
536 (q/g (truncate q g))
537 (lcm-pq (* p q/g)) )
538 (list (mod (+ rq (* h q/g)) lcm-pq) lcm-pq) ))))))))
540 ;; (zn-gcdex2 x y) returns `(,g ,c) where c*x + d*y = g = gcd(x,y)
542 (defun zn-gcdex2 (x y)
543 (let ((x1 1) (y1 0) q r)
544 (do ()((= 0 y) (list x x1))
545 (multiple-value-setq (q r) (truncate x y))
546 (psetf x y y r)
547 (psetf x1 y1 y1 (- x1 (* q y1))) )))
550 ;; discrete logarithm:
551 ;; solve g^x = a mod n, where g is a generator of (Z/nZ)* or of a subgroup containing a
553 ;; see: lecture notes 'Grundbegriffe der Kryptographie' - Eike Best
554 ;; http://theoretica.informatik.uni-oldenburg.de/~best/publications/kry-Mai2005.pdf
556 ;; optional argument: ifactors of zn_order(g,n)
558 (defmfun $zn_log (a g n &optional fs-ord)
559 (unless (and (integerp a) (integerp g) (integerp n))
560 (return-from $zn_log
561 (if fs-ord
562 (list '($zn_log) a g n fs-ord)
563 (list '($zn_log) a g n) )))
564 (when (minusp a) (setq a (mod a n)))
565 (cond
566 ((or (= 0 a) (>= a n)) nil)
567 ((= 1 a) 0)
568 ((= g a) 1)
569 ((> (gcd a n) 1) nil)
571 (if fs-ord
572 (if (and ($listp fs-ord) ($listp (cadr fs-ord)))
573 (progn
574 (setq fs-ord (mapcar #'cdr (cdr fs-ord))) ; Lispify fs-ord
575 (setq fs-ord (cons (totient-from-factors fs-ord) fs-ord)) ) ; totient resp. order in general
576 (gf-merror (intl:gettext
577 "Fourth argument to `zn_log' must be of the form [[p1, e1], ..., [pk, ek]]." )))
578 (let (($intfaclim) (ord ($zn_order g n)))
579 (setq fs-ord (cons ord (get-factor-list ord))) ))
580 (cond
581 ((= 0 (mod (- a (* g g)) n))
583 ((= 1 (mod (* a g) n))
584 (mod -1 (car fs-ord)) )
586 (zn-dlog a g n
587 (car fs-ord) ;; order
588 (cdr fs-ord) ) ))))) ;; factors with multiplicity
590 ;; Pohlig-Hellman-reduction:
592 ;; Solve g^x = a mod n.
593 ;; Assume, that a is an element of (Z/nZ)*
594 ;; and g is a generator of (Z/nZ)* or of a subgroup containing a.
596 (defun zn-dlog (a g n ord fs-ord)
597 (let (p e ord/p om xp xk mods dlogs (g-inv (inv-mod g n)))
598 (dolist (f fs-ord)
599 (setq p (car f) e (cadr f)
600 ord/p (truncate ord p)
601 om (power-mod g ord/p n) ;; om is a generator of prime order p
602 xp 0 )
603 ;; Let op = ord/p^e, gp = g^op (mod n), ap = a^op (mod n) and
604 ;; xp = x (mod p^e).
605 ;; gp is of order p^e and therefore
606 ;; (*) gp^xp = ap (mod n).
607 (do ((b a) (k 0) (pk 1) (acc g-inv) (e1 (1- e))) (()) ;; Solve (*) by solving e logs ..
608 (setq xk (dlog-rho (power-mod b ord/p n) om p n)) ;; .. in subgroups of order p.
609 (incf xp (* xk pk))
610 (incf k)
611 (when (= k e) (return)) ;; => xp = x_0+x_1*p+x_2*p^2+...+x_{e-1}*p^{e-1} < p^e
612 (setq ord/p (truncate ord/p p)
613 pk (* pk p) )
614 (when (/= xk 0) (setq b (mod (* b (power-mod acc xk n)) n)))
615 (when (/= k e1) (setq acc (power-mod acc p n))) )
616 (push (expt p e) mods)
617 (push xp dlogs) )
618 (car (chinese dlogs mods)) )) ;; Find x (mod ord) with x = xp (mod p^e) for all p,e.
620 ;; baby-steps-giant-steps:
622 (defun dlog-baby-giant (a g p n) ;; g is generator of order p mod n
623 (let* ((m (1+ (isqrt p)))
624 (s (floor (* 1.3 m)))
625 (gi (inv-mod g n))
626 g^m babies )
627 (setf babies
628 (make-hash-table :size s :test #'eql :rehash-threshold 0.9) )
629 (do ((r 0) (b a))
630 (())
631 (when (= 1 b)
632 (clrhash babies)
633 (return-from dlog-baby-giant r) )
634 (setf (gethash b babies) r)
635 (incf r)
636 (when (= r m) (return))
637 (setq b (mod (* gi b) n)) )
638 (setq g^m (power-mod g m n))
639 (do ((rr 0 (1+ rr))
640 (bb 1 (mod (* g^m bb) n))
641 r ) (())
642 (when (setq r (gethash bb babies))
643 (clrhash babies)
644 (return (+ (* rr m) r)) )) ))
646 ;; brute-force:
648 (defun dlog-naive (a g n)
649 (do ((i 0 (1+ i)) (gi 1 (mod (* gi g) n)))
650 ((= gi a) i) ))
652 ;; Pollard rho for dlog computation (Brents variant of collision detection)
654 (defun dlog-rho (a g p n) ;; g is generator of prime order p mod n
655 (cond
656 ((= 1 a) 0)
657 ((= g a) 1)
658 ((= 0 (mod (- a (* g g)) n)) 2)
659 ((= 1 (mod (* a g) n)) (1- p))
660 ((< p 512) (dlog-naive a g n))
661 ((< p 65536) (dlog-baby-giant a g p n))
663 (prog ((b 1) (y 0) (z 0) ;; b = g^y * a^z
664 (bb 1) (yy 0) (zz 0) ;; bb = g^yy * a^zz
665 dy dz )
667 (do ((i 0)(j 1)) (()) (declare (fixnum i j))
668 (multiple-value-setq (b y z) (dlog-f b y z a g p n))
669 (when (equal b bb) (return)) ;; g^y * a^z = g^yy * a^zz
670 (incf i)
671 (when (= i j)
672 (setq j (1+ (ash j 1)))
673 (setq bb b yy y zz z) ))
674 (setq dy (mod (- y yy) p) dz (mod (- zz z) p)) ;; g^dy = a^dz = g^(x*dz)
675 (when (= 1 (gcd dz p))
676 (return (zn-quo dy dz p)) ) ;; x = dy/dz mod p (since g is generator of order p)
677 (setq y 0
680 yy (1+ (random (1- p)))
681 zz (1+ (random (1- p)))
682 bb (mod (* (power-mod g yy n) (power-mod a zz n)) n) )
683 (go rho) ))))
685 ;; iteration for Pollard rho: b = g^y * a^z in each step
687 (defun dlog-f (b y z a g ord n)
688 (let ((m (mod b 3)))
689 (cond
690 ((= 0 m)
691 (values (mod (* b b) n) (mod (ash y 1) ord) (mod (ash z 1) ord)) )
692 ((= 1 m) ;; avoid stationary case b=1 => b*b=1
693 (values (mod (* a b) n) y (mod (+ z 1) ord) ) )
695 (values (mod (* g b) n) (mod (+ y 1) ord) z ) ) )))
698 ;; characteristic factors:
700 (defmfun $zn_characteristic_factors (m)
701 (unless (and (integerp m) (> m 1)) ;; according to Shanks no result for m = 1
702 (gf-merror (intl:gettext
703 "`zn_characteristic_factors': Argument must be an integer greater than 1." )))
704 (cons '(mlist simp) (zn-characteristic-factors m)) )
706 (defmfun $zn_carmichael_lambda (m)
707 (cond
708 ((integerp m)
709 (if (= m 1) 1 (zn-characteristic-factors m t)) )
710 (t (gf-merror (intl:gettext
711 "`zn_carmichael_lambda': Argument must be a positive integer." )))))
713 ;; D. Shanks - Solved and unsolved problems in number theory, 2. ed
714 ;; definition 29 and 30 (p. 92 - 94)
716 ;; (zn-characteristic-factors 104) --> (2 2 12)
717 ;; => Z104* is isomorphic to M2 x M2 x M12
718 ;; the direct product of modulo multiplication groups of order 2 resp. 12
720 (defun zn-characteristic-factors (m &optional lambda-only) ;; m > 1
721 (let* (($intfaclim)
722 (pe-list (get-factor-list m)) ;; def. 29 part A
723 (shanks-phi ;; part D
724 (sort
725 (apply #'nconc (mapcar #'zn-shanks-phi-step-bc pe-list))
726 #'zn-pe> )))
727 ;; def. 30 :
728 (do ((todo shanks-phi (nreverse left))
729 (p 0 0) (f 1 1) (left nil nil)
730 fs q d )
731 ((null todo) fs)
732 (dolist (qd todo)
733 (setq q (car qd) d (cadr qd))
734 (if (= q p)
735 (push qd left)
736 (setq p q f (* f (expt q d))) ))
737 (when lambda-only (return-from zn-characteristic-factors f))
738 (push f fs) )))
740 ;; definition 29 parts B,C (p. 92)
741 (defun zn-shanks-phi-step-bc (pe)
742 (let ((p (car pe)) (e (cadr pe)) qd)
743 (cond
744 ((= 2 p)
745 (setq qd (list (if (> e 1) '(2 1) '(1 1))))
746 (when (> e 2) (setq qd (nconc qd (list `(2 ,(- e 2)))))) )
748 (setq qd (let (($intfaclim)) (get-factor-list (1- p))))
749 (when (> e 1)
750 (setq qd (nconc qd (list `(,p ,(1- e))))) )))
751 qd ))
753 (defun zn-pe> (a b)
754 (cond ((> (car a) (car b)) t)
755 ((< (car a) (car b)) nil)
756 (t (> (cadr a) (cadr b))) ))
759 ;; factor generators:
761 (defmfun $zn_factor_generators (m)
762 (unless (and (integerp m) (> m 1))
763 (gf-merror (intl:gettext
764 "`zn_factor_generators': Argument must be an integer greater or equal 2." )))
765 (cons '(mlist simp) (zn-factor-generators m)) )
767 ;; D. Shanks - Solved and unsolved problems in number theory, 2. ed
768 ;; Theorem 44, page 94
770 ;; zn_factor_generators(104); --> [79,27,89]
771 ;; zn_characteristic_factors(104); --> [2,2,12]
772 ;; map(lambda([g], zn_order(g,104)), [79,27,89]); --> [2,2,12]
774 ;; Every element in Z104* can be expressed as
775 ;; 79^i * 27^j * 89^k (mod m) where 0 <= i,j < 2 and 0 <= k < 12
777 ;; The proof of theorem 44 contains the construction of the factor generators.
779 (defun zn-factor-generators (m) ;; m > 1
780 (let* (($intfaclim)
781 (fs (sort (get-factor-list m) #'< :key #'car))
782 (pe (car fs))
783 (p (car pe)) (e (cadr pe))
784 (a (expt p e))
785 phi fs-phi ga gs ords fs-ords pegs )
786 ;; lemma 1, page 98 :
787 ;; (Z/mZ)* is cyclic when m =
788 (when (= m 2) ;; 2
789 (return-from zn-factor-generators (list 1)) )
790 (when (or (< m 8) ;; 3,4,5,6,7
791 (and (> p 2) (null (cdr fs))) ;; p^k, p#2
792 (and (= 2 p) (= 1 e) (null (cddr fs))) ) ;; 2*p^k, p#2
793 (setq phi (totient-by-fs-n fs)
794 fs-phi (sort (mapcar #'car (get-factor-list phi)) #'<)
795 ga (zn-primroot m phi fs-phi) )
796 (return-from zn-factor-generators (list ga)) )
797 (setq fs (cdr fs))
798 (cond
799 ((= 2 p)
800 (unless (and (= e 1) (cdr fs)) ;; phi(2*m) = phi(m) if m#1 is odd
801 (push (1- a) gs) ) ;; a = 2^e
802 (when (> e 1) (setq ords (list 2) fs-ords (list '((2 1)))))
803 (when (> e 2)
804 (push 3 gs) (push (expt 2 (- e 2)) ords) (push `((2 ,(- e 2))) fs-ords) ))
805 ;; lemma 2, page 100 :
807 (setq phi (* (1- p) (expt p (1- e)))
808 fs-phi (sort (get-factor-list (1- p)) #'< :key #'car) )
809 (when (> e 1) (setq fs-phi (nconc fs-phi (list `(,p ,(1- e))))))
810 (setq ga (zn-primroot a phi (mapcar #'car fs-phi)) ;; factors only
811 gs (list ga)
812 ords (list phi)
813 fs-ords (list fs-phi) )))
815 (do (b gb c h ia)
816 ((null fs))
817 (setq pe (car fs) fs (cdr fs)
818 p (car pe) e (cadr pe)
819 phi (* (1- p) (expt p (1- e)))
820 fs-phi (sort (get-factor-list (1- p)) #'< :key #'car) )
821 (when (> e 1) (setq fs-phi (nconc fs-phi (list `(,p ,(1- e))))))
822 (setq b (expt p e)
823 gb (zn-primroot b phi (mapcar #'car fs-phi))
824 c (mod (* (inv-mod b a) (- 1 gb)) a) ;; CRT: h = gb mod b
825 h (+ (* b c) gb) ;; CRT: h = 1 mod a
826 ia (inv-mod a b)
827 gs (mapcar #'(lambda (g) (+ (* a (mod (* ia (- 1 g)) b)) g)) gs)
828 gs (cons h gs)
829 ords (cons phi ords)
830 fs-ords (cons fs-phi fs-ords)
831 a (* a b) ))
832 ;; lemma 3, page 101 :
833 (setq pegs
834 (mapcar #'(lambda (g ord f)
835 (mapcar #'(lambda (pe)
836 (append pe
837 (list (power-mod g (truncate ord (apply #'expt pe)) m)) ))
838 f ))
839 gs ords fs-ords ))
840 (setq pegs (sort (apply #'append pegs) #'zn-pe>))
841 (do ((todo pegs (nreverse left))
842 (q 0 0) (fg 1 1) (left nil nil)
843 g fgs )
844 ((null todo) fgs)
845 (dolist (peg todo)
846 (setq p (car peg) g (caddr peg))
847 (if (= p q)
848 (push peg left)
849 (setq q p fg (mod (* fg g) m)) ))
850 (push fg fgs) )))
853 ;; r-th roots --------------------------------------------------------------- ;;
855 ;; If the residue class a is an r-th power modulo n and contained in a multiplication
856 ;; subgroup of (Z/nZ), return all r-th roots from this subgroup and false otherwise.
858 (defmfun $zn_nth_root (a r n &optional fs-n)
859 (unless (and (integerp a) (integerp r) (integerp n))
860 (gf-merror (intl:gettext
861 "`zn_nth_root' needs three integer arguments. Found ~m, ~m, ~m." ) a r n))
862 (unless (and (> r 0) (> n 0))
863 (gf-merror (intl:gettext
864 "`zn_nth_root': Second and third arg must be pos integers. Found ~m, ~m." ) r n))
865 (when fs-n
866 (if (and ($listp fs-n) ($listp (cadr fs-n)))
867 (setq fs-n (mapcar #'cdr (cdr fs-n))) ;; Lispify fs-n
868 (gf-merror (intl:gettext
869 "`zn_nth_root': The opt fourth arg must be of the form [[p1, e1], ..., [pk, ek]]." ))))
870 (let ((rts (zn-nrt a r n fs-n)))
871 (when rts (cons '(mlist simp) rts)) ))
873 (defun zn-nrt (a r n &optional fs-n)
874 (let (g n/g c p q aq ro ord qs rt rts rems res)
875 (unless fs-n (setq fs-n (let (($intfaclim)) (get-factor-list n))))
876 (setq a (mod a n))
877 (cond
878 ((every #'onep (mapcar #'second fs-n)) ;; RSA-like case (n is squarefree)
879 (when (= a 0) (return-from zn-nrt (list 0))) ) ;; n = 1: exit here
880 ((/= (gcd a n) 1)
881 ;; Handle residue classes not coprime to n (n is not squarefree):
882 ;; Use Theorems 49 and 50 from
883 ;; Shanks - Solved and Unsolved Problems in Number Theory
884 (setq g (gcd a n) n/g (truncate n g))
885 (when (/= (gcd g n/g) 1) ;; a is not contained in any mult. subgroup (Th. 50):
886 (return-from zn-nrt nil) ) ;; exit here
887 (when (= a 0) (return-from zn-nrt (list 0)))
888 ;; g = gcd(a,n). Now assume gcd(g,n/g) = 1.
889 ;; There are totient(n/g) multiples of g, i*g, with gcd(i,n/g) = 1,
890 ;; which form a modulo multiplication subgroup of (Z/nZ),
891 ;; isomorphic to (Z/mZ)*, where m = n/g.
892 ;; a is one of these multiples of g.
893 ;; Find the r-th roots of a resp. mod(a,m) in (Z/mZ)* and then
894 ;; by using CRT all corresponding r-th roots of a in (Z/nZ).
895 (setq a (mod a n/g)
896 rts (zn-nrt a r n/g)
897 c (inv-mod g n/g) ;; CRT-coeff
898 ;; isomorphic mapping (Th. 49):
899 ;; (use CRT with x = 0 mod g and x = rt mod n/g)
900 res (mapcar #'(lambda (rt) (* g (mod (* c rt) n/g))) rts) )
901 (return-from zn-nrt (sort res #'<)) ))
903 ;; for every prime power factor of n
904 ;; reduce a and r if possible and call zq-nrt:
905 (dolist (pe fs-n)
906 (setq p (car pe)
907 q (apply #'expt pe)
908 aq (mod a q)
909 ord (* (1- p) (truncate q p)) )
910 (cond
911 ((> r ord)
912 (setq ro (mod r ord))
913 (when (= ro 0) (setq ro ord)) )
914 (t (setq ro r)) )
915 (cond
916 ((= aq 0)
917 (if (or (= p q) (= ro 1))
918 (setq rt (list 0))
919 (return-from zn-nrt nil) ))
920 ((= ro 1)
921 (setq rt (list aq)) )
923 (setq rt (zq-nrt aq ro p q))
924 (unless rt (return-from zn-nrt nil)) ))
925 (push q qs)
926 (push rt rts) )
927 ;; CRT in case n splits into more than one factor:
928 (if (= 1 (length fs-n))
929 (setq res rt) ;; n is a prime power
930 (setq qs (nreverse qs)
931 rems (zn-distrib-lists (nreverse rts))
932 res (mapcar #'(lambda (rs) (car (chinese rs qs))) rems) ))
933 (sort res #'<) ))
935 ;; return all possible combinations containing one entry per list:
936 ;; (zn-distrib-lists '((1 2 3) (4) (5 6)))
937 ;; --> ((1 4 5) (1 4 6) (2 4 5) (2 4 6) (3 4 5) (3 4 6))
939 (defun zn-distrib-lists (ll)
940 (let ((res (car ll)) tmp e)
941 (dolist (l (cdr ll) res)
942 (setq tmp nil)
943 (dolist (r res)
944 (dolist (n l)
945 (setq e (if (listp r) (copy-list r) (list r)))
946 (push (nconc e (list n)) tmp) ))
947 (setq res (nreverse tmp)) )))
949 ;; handle composite r (which are not coprime to totient(q)):
950 ;; e.g. r=x*x*y*z, then a^(1/r) = (((a^(1/x))^(1/x))^(1/y))^(1/z)
952 (defun zq-nrt (a r p q) ;; prime power q = p^e
953 ;; assume a < q, r <= q
954 (let (rts)
955 (cond
956 ((or (= 1 r) (primep r))
957 (setq rts (zq-amm a r p q)) )
958 ((and (= (gcd r (1- p)) 1) (or (= p q) (= (gcd r p) 1))) ;; r is coprime to totient(q):
959 (let ((ord (* (1- p) (truncate q p))))
960 (setq rts (list (power-mod a (inv-mod r ord) q))) )) ;; unique solution
962 (let* (($intfaclim) (rs (get-factor-list r)))
963 (setq rs (sort rs #'< :key #'car))
964 (setq rs
965 (apply #'nconc
966 (mapcar
967 #'(lambda (pe) (apply #'(lambda (p e) (make-list e :initial-element p)) pe))
968 rs )))
969 (setq rts (zq-amm a (car rs) p q))
970 (dolist (r (cdr rs))
971 (setq rts (apply #'nconc (mapcar #'(lambda (a) (zq-amm a r p q)) rts))) ))))
972 (if (and (= p 2) (> q 2) (evenp r)) ;; this case needs a postprocess (see below)
973 (nconc (mapcar #'(lambda (rt) (- q rt)) rts) rts) ;; add negative solutions
974 rts )))
976 ;; Computing r-th roots modulo a prime power p^n, where r is a prime
978 ;; inspired by
979 ;; Bach,Shallit - Algorithmic Number Theory, Theorem 7.3.2
980 ;; and
981 ;; Shanks - Solved and Unsolved Problems in Number Theory, Th. 46, Lemma 1 to Th. 44
983 ;; The algorithm AMM (Adleman, Manders, Miller) is essentially based on
984 ;; properties of cyclic groups and with the exception of q = 2^n, n > 2
985 ;; it can be applied to any multiplicative group (Z/qZ)* where q = p^n.
987 ;; Doing so, the order q-1 of Fq* in Th. 7.3.2 has to be replaced by the
988 ;; group order totient(q) of (Z/qZ)*.
990 ;; But how to include q = 8,16,32,... ?
991 ;; r > 2: r is prime. There exists a unique solution for all a in (Z/qZ)*.
992 ;; r = 2 (the square root case):
993 ;; - (Z/qZ)* has k = 2 characteristic factors [2,q/4] with [-1,3] as possible
994 ;; factor generators (see Shanks, Lemma 1 to Th. 44).
995 ;; I.e. 3 is of order q/4 and 3^2 = 9 of order q/8.
996 ;; - (Z/qZ)* has totient/2^k = q/8 quadratic residues with 2^k = 4 roots each
997 ;; (see Shanks, Th. 46).
998 ;; - It follows that the subgroup <3> generated by 3 contains all quadratic
999 ;; residues of (Z/qZ)* (which must be all the powers of 9 located in <3>).
1000 ;; - We apply the algorithm AMM for cyclic groups to <3> and compute two
1001 ;; square roots x,y.
1002 ;; - The numbers -x and -y, obviously roots as well, both lie in (-1)*<3>
1003 ;; and therefore they differ from x,y and complete the set of 4 roots.
1005 (defun zq-amm (a r p q) ;; r,p prime, q = p^n
1006 ;; assume a < q, r <= q
1007 (cond
1008 ((= 1 r) (list a))
1009 ((= 2 q) (when (= 1 a) (list 1)))
1010 ((= 4 q) (when (or (= 1 a) (and (= 3 a) (oddp r))) (list a)))
1012 (let ((ord (* (1- p) (truncate q p))) ;; group order: totient(q)
1013 rt s m e u )
1014 (when (= 2 r)
1015 (if (= 2 p)
1016 (when (/= 1 (mod a 8)) (return-from zq-amm nil)) ;; a must be a power of 9
1017 (cond
1018 ((/= 1 ($jacobi (mod a p) p))
1019 (return-from zq-amm nil) )
1020 ((= 2 (mod ord 4))
1021 (setq rt (power-mod a (ash (+ ord 2) -2) q))
1022 (return-from zq-amm `(,rt ,(- q rt))) )
1023 ((and (= p q) (= 5 (mod p 8)))
1024 (let* ((x (ash a 1))
1025 (y (power-mod x (ash (- p 5) -3) p))
1026 (i (mod (* x y y) p))
1027 (rt (mod (* a y (1- i)) p)) )
1028 (return-from zq-amm `(,rt ,(- p rt))) )))))
1029 (when (= 2 p) ;; q = 8,16,32,..
1030 (setq ord (ash ord -1)) ) ;; factor generator 3 is of order ord/2
1031 (multiple-value-setq (s m) (truncate ord r))
1032 (when (and (= 0 m) (/= 1 (power-mod a s q))) (return-from zq-amm nil))
1033 ;; r = 3, first 2 cases:
1034 (when (= 3 r)
1035 (cond
1036 ((= 1 (setq m (mod ord 3))) ;; unique solution
1037 (return-from zq-amm
1038 `(,(power-mod a (truncate (1+ (ash ord 1)) 3) q)) ))
1039 ((= 2 m) ;; unique solution
1040 (return-from zq-amm
1041 `(,(power-mod a (truncate (1+ ord) 3) q)) ))))
1042 ;; compute u,e with ord = u*r^e and r,u coprime:
1043 (setq u ord e 0)
1044 (do ((u1 u)) (())
1045 (multiple-value-setq (u1 m) (truncate u1 r))
1046 (when (/= 0 m) (return))
1047 (setq u u1 e (1+ e)) )
1048 (cond
1049 ((= 0 e)
1050 (setq rt (power-mod a (inv-mod r u) q)) ;; unique solution, see Bach,Shallit 7.3.1
1051 (list rt) )
1052 (t ;; a is an r-th power
1053 (let (g re om)
1054 ;; find generator of order r^e:
1055 (if (= p 2) ;; p = 2: then r = 2 (other r: e = 0)
1056 (setq g 3)
1057 (do ((n 2 ($next_prime n)))
1058 ((and (= 1 (gcd n q)) (/= 1 (power-mod n s q))) ;; n is no r-th power
1059 (setq g (power-mod n u q)) )))
1060 (setq re (expt r e)
1061 om (power-mod g (truncate re r) q) ) ;; r-th root of unity
1062 (cond
1063 ((or (/= 3 r) (= 0 (setq m (mod ord 9))))
1064 (let (ar au br bu k ab alpha beta)
1065 ;; map a from Zq* to C_{r^e} x C_u:
1066 (setq ar (power-mod a u q) ;; in C_{r^e}
1067 au (power-mod a re q) ) ;; in C_u
1068 ;; compute direct factors of rt:
1069 ;; (the loop in algorithm AMM is effectively a Pohlig-Hellman-reduction, equivalent to zn-dlog)
1070 (setq k (zn-dlog ar g q re `((,r ,e))) ;; g^k = ar, where r|k
1071 br (power-mod g (truncate k r) q) ;; br^r = g^k (factor of rt in C_{r^e})
1072 bu (power-mod au (inv-mod r u) q) ) ;; bu^r = au (factor of rt in C_u)
1073 ;; mapping from C_{r^e} x C_u back to Zq*:
1074 (setq ab (cdr (zn-gcdex1 u re))
1075 alpha (car ab)
1076 beta (cadr ab) )
1077 (if (< alpha 0) (incf alpha ord) (incf beta ord))
1078 (setq rt (mod (* (power-mod br alpha q) (power-mod bu beta q)) q)) ))
1079 ;; r = 3, remaining cases:
1080 ((= 3 m)
1081 (setq rt (power-mod a (truncate (+ (ash ord 1) 3) 9) q)) )
1082 ((= 6 m)
1083 (setq rt (power-mod a (truncate (+ ord 3) 9) q)) ))
1084 ;; mult with r-th roots of unity:
1085 (do ((i 1 (1+ i)) (j 1) (res (list rt)))
1086 ((= i r) res)
1087 (setq j (mod (* j om) q))
1088 (push (mod (* rt j) q) res) ))))))))
1090 ;; -------------------------------------------------------------------------- ;;
1093 ;; Two variants of gcdex:
1095 ;; returns gcd as first entry:
1096 ;; (zn-gcdex1 12 45) --> (3 4 -1), so 4*12 + -1*45 = 3
1097 (defun zn-gcdex1 (x y)
1098 (let ((x1 1) (x2 0) (y1 0) (y2 1) q r)
1099 (do ()((= 0 y) (list x x1 x2))
1100 (multiple-value-setq (q r) (truncate x y))
1101 (psetf x y y r)
1102 (psetf x1 y1 y1 (- x1 (* q y1)))
1103 (psetf x2 y2 y2 (- x2 (* q y2))) )))
1105 ;; returns gcd as last entry:
1106 ;; (zn-gcdex 12 45 21) --> (4 -1 0 3), so 4*12 + -1*45 + 0*21 = 3
1107 (defun zn-gcdex (&rest args)
1108 (let* ((ex (zn-gcdex1 (car args) (cadr args)))
1109 (g (car ex))
1110 (cs (cdr ex)) c1 )
1111 (dolist (a (cddr args) (nconc cs (list g)))
1112 (setq ex (zn-gcdex1 g a)
1113 g (car ex)
1114 ex (cdr ex)
1115 c1 (car ex)
1116 cs (nconc (mapcar #'(lambda (c) (* c c1)) cs) (cdr ex)) ))))
1119 ;; for educational puposes: tables of small residue class rings
1121 (defun zn-table-errchk (n fun)
1122 (unless (and (fixnump n) (< 1 n))
1123 (gf-merror (intl:gettext
1124 "Argument to `~m' must be a small fixnum greater than 1." ) fun )))
1126 (defmfun $zn_add_table (n)
1127 (zn-table-errchk n "zn_add_table")
1128 (do ((i 0 (1+ i)) res)
1129 ((= i n)
1130 (cons '($matrix simp) (nreverse res)) )
1131 (push (mfuncall '$makelist `((mod) (+ ,i $j) ,n) '$j 0 (1- n)) res) ))
1133 ;; multiplication table modulo n
1135 ;; The optional g allows to choose subsets of (Z/nZ). Show i with gcd(i,n) = g resp. all i#0.
1136 ;; If n # 1 add row and column headings for better readability.
1138 (defmfun $zn_mult_table (n &optional g)
1139 (zn-table-errchk n "zn_mult_table")
1140 (let ((i0 1) all header choice res)
1141 (cond
1142 ((not g) (setq g 1))
1143 ((equal g '$all) (setq all t))
1144 ((not (fixnump g))
1145 (gf-merror (intl:gettext
1146 "`zn_mult_table': The opt second arg must be `all' or a small fixnum." )))
1148 (when (= n g) (setq i0 0))
1149 (push 1 choice) ;; creates the headers
1150 (setq header t) ))
1151 (do ((i i0 (1+ i)))
1152 ((= i n)
1153 (setq choice (cons '(mlist simp) (nreverse choice))) )
1154 (when (or all (= g (gcd i n))) (push i choice)) )
1155 (when (and header (= (length choice) 2))
1156 (return-from $zn_mult_table) )
1157 (dolist (i (cdr choice))
1158 (push (mfuncall '$makelist `((mod) (* ,i $j) ,n) '$j choice) res) )
1159 (setq res (nreverse res))
1160 (when header (rplaca (cdar res) "*"))
1161 (cons '($matrix simp) res) ))
1163 ;; power table modulo n
1165 ;; The optional g allows to choose subsets of (Z/nZ). Show i with gcd(i,n) = g resp. all i.
1167 (defmfun $zn_power_table (n &optional g e)
1168 (zn-table-errchk n "zn_power_table")
1169 (let (all)
1170 (cond
1171 ((not g) (setq g 1))
1172 ((equal g '$all) (setq all t))
1173 ((not (fixnump g))
1174 (gf-merror (intl:gettext
1175 "`zn_power_table': The opt second arg must be `all' or a small fixnum." ))))
1176 (cond
1177 ((not e) (setq e (zn-characteristic-factors n t)))
1178 ((not (fixnump e))
1179 (gf-merror (intl:gettext
1180 "`zn_power_table': The opt third arg must be a small fixnum." ))))
1181 (do ((i 0 (1+ i)) res)
1182 ((= i n)
1183 (when res (cons '($matrix simp) (nreverse res))) )
1184 (when (or all (= g (gcd i n)))
1185 (push (mfuncall '$makelist `((power-mod) ,i $j ,n) '$j 1 e) res) ))))
1188 ;; $zn_invert_by_lu (m p)
1189 ;; $zn_determinant (m p)
1190 ;; see below: --> galois fields--> interfaces to linearalgebra/lu.lisp
1193 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1196 ;; -----------------------------------------------------------------------------
1197 ;; *** GALOIS FIELDS ***
1199 ;; The following is a revision and improvement of the first part of share/
1200 ;; contrib/gf/gf.mac by Alasdair McAndrew, Fabrizio Caruso and Jacopo D'Aurizio
1201 ;; released under terms of the GPLv2 in 2007.
1203 ;; I would like to thank the original authors for their contribution to Maxima
1204 ;; which allowed me to study, improve and extend the source code.
1206 ;; I would also like to thank Camm Maguire who helped me coding compiler macros
1207 ;; for GCL.
1209 ;; 2012 - 2014, Volker van Nek
1211 (declare-top (special *gf-char* *gf-exp* *ef-arith?*)) ;; modulus $intfaclim see above
1213 (defvar *gf-rat-header* nil "header of internal CRE representation")
1215 (defvar *ef-arith?* nil "Should extension field arithmetic be used?")
1217 ;; base field:
1218 (defvar *gf-char* 0 "characteristic p")
1219 (defvar *gf-exp* 0 "exponent n, degree of the reduction polynomial")
1220 (defvar *gf-ord* 0 "group order, number of units")
1221 (defvar *gf-card* 0 "cardinality, ring order")
1222 (defvar *gf-red* nil "reduction polynomial")
1223 (defvar *gf-prim* nil "primitive element")
1224 (defvar *gf-fs-ord* nil "ifactors of *gf-ord*")
1225 (defvar *gf-fsx* nil "extended factors of *gf-ord*")
1226 (defvar *gf-fsx-base-p* nil "*gf-fsx* in base p")
1227 (defvar *gf-x^p-powers* nil "x^p^i, i=0,..,n-1")
1229 (defvar *f2-red* 0 "reduction polynomial") ;; used by ef coeff arith over binary fields
1231 (declaim (fixnum *gf-exp* *ef-exp*))
1233 ;; extension:
1234 (defvar *ef-exp* 0 "exponent m, degree of the reduction polynomial")
1235 (defvar *ef-ord* 0 "group order, number of units")
1236 (defvar *ef-card* 0 "cardinality, ring order")
1237 (defvar *ef-red* nil "reduction polynomial")
1238 (defvar *ef-prim* nil "primitive element")
1239 (defvar *ef-fs-ord* nil "ifactors of *ef-ord*")
1240 (defvar *ef-fsx* nil "extended factors of *ef-ord*")
1241 (defvar *ef-fsx-base-q* nil "*ef-fsx* in base q = p^n")
1242 (defvar *ef-x^q-powers* nil "x^q^i, i=0,..,m-1")
1244 (defvar *gf-char?* nil "Was the characteristic defined?")
1245 (defvar *gf-red?* nil "Was the reduction polynomial defined?")
1246 (defvar *gf-irred?* nil "Is the reduction polynomial irreducible?")
1247 (defvar *gf-data?* nil "gf_set_data called?")
1249 (defvar *ef-red?* nil "Was the reduction polynomial defined?")
1250 (defvar *ef-irred?* nil "Is the reduction polynomial irreducible?")
1251 (defvar *ef-data?* nil "ef_set_data called?")
1253 (defmvar $gf_rat nil "Return values are rational expressions?" boolean)
1255 (defmvar $gf_symmetric nil "A symmetric modulus should be used?" boolean) ;; deprecated
1256 (defmvar $gf_balanced nil "A balanced modulus should be used?" boolean) ;; there is ec_balanced in elliptic_curves.lisp
1258 (putprop '$gf_symmetric 'gf-balanced-info 'assign)
1260 (defun gf-balanced-info (assign-var arg)
1261 (declare (ignore assign-var))
1262 (setq $gf_balanced arg)
1263 (format t "`gf_symmetric' is deprecated and replaced by `gf_balanced'.~%The value is bound to `gf_balanced'.") )
1264 ;; temporarily print this message
1267 (defmvar $gf_coeff_limit 256
1268 "`gf_coeff_limit' limits the coeffs when searching for irreducible and primitive polynomials." fixnum)
1270 (putprop '$gf_coeff_limit 'gf-coeff-check 'assign)
1272 (defun gf-coeff-check (assign-var arg)
1273 (declare (ignore assign-var))
1274 (unless (and (integerp arg) (> arg 1))
1275 (gf-merror (intl:gettext
1276 "`gf_coeff_limit': Assignment ignored. Value must be an integer greater than 1.~%" ))))
1278 (defmvar $gf_cantor_zassenhaus t "Should the Cantor-Zassenhaus algorithm be used?" boolean)
1280 (defmvar $ef_coeff_mult nil)
1281 (defmvar $ef_coeff_add nil)
1282 (defmvar $ef_coeff_inv nil)
1283 (defmvar $ef_coeff_exp nil)
1285 (defmvar $gf_powers nil)
1286 (defmvar $gf_logs nil)
1287 (defmvar $gf_zech_logs nil)
1288 (defvar *gf-powers* nil "alpha^i, i=0,..,ord-1 where alpha is a primitive element")
1289 (defvar *gf-logs?* nil "Were the power and log tables calculated?")
1292 ;; contains parts of merror.lisp/merror but avoids "To debug this ...".
1294 (defun gf-merror (sstring &rest l)
1295 (setq $error `((mlist) ,sstring ,@ l))
1296 (and $errormsg ($errormsg))
1297 (fresh-line *standard-output*)
1298 (format t (intl:gettext "~& -- an error.~%"))
1299 (throw 'macsyma-quit 'maxima-error) )
1302 (defun gf-char? (fun)
1303 (if *gf-char?* t
1304 (gf-merror (intl:gettext "`~m': The characteristic is not defined yet.") fun) ))
1306 (defun gf-red? (fun)
1307 (if *gf-red?* t
1308 (gf-merror (intl:gettext "`~m': The reduction polynomial is not defined yet.") fun) ))
1310 (defun gf-data? (fun)
1311 (if *gf-data?* t
1312 (gf-merror (intl:gettext "`~m': gf_set_data called?") fun) ))
1314 (defun gf-field? (fun)
1315 (if (and (gf-data? fun) *gf-irred?*) t
1316 (gf-merror (intl:gettext "`~m': The reduction polynomial is not irreducible.") fun) ))
1319 (defun ef-gf-field? (fun)
1320 (if (and *gf-data?* *gf-irred?*) t
1321 (gf-merror (intl:gettext "`~m': The base field is not defined yet.") fun) ))
1323 (defun ef-red? (fun)
1324 (if (and (ef-gf-field? fun) *ef-red?*) t
1325 (gf-merror (intl:gettext "`~m': The reduction polynomial is not defined yet.") fun) ))
1327 (defun ef-data? (fun)
1328 (if (and (ef-gf-field? fun) *ef-data?*) t
1329 (gf-merror (intl:gettext "`~m': ef_set_data called?") fun) ))
1331 (defun ef-field? (fun)
1332 (if (and (ef-data? fun) *ef-irred?*) t
1333 (gf-merror (intl:gettext "`~m': The extension is no field.") fun) ))
1335 ;; -----------------------------------------------------------------------------
1338 ;; basic coefficient arithmetic ------------------------------------------------
1341 ;; optimize the fixnum cases
1343 (defmacro maybe-char-is-fixnum-let (binds &body body)
1344 `(if (or (and (not *ef-arith?*) (typep *gf-char* 'fixnum))
1345 (and *ef-arith?* (typep *gf-card* 'fixnum)) )
1346 (let ,binds
1347 (declare (fixnum ,@(mapcar #'(lambda (x) (car x)) binds)))
1348 ,@body)
1349 (let ,binds
1350 (declare (integer ,@(mapcar #'(lambda (x) (car x)) binds)))
1351 ,@body )))
1353 ;; basic coefficient functions and compiler macros
1355 ;; gf coefficient arith :
1357 ;; *ef-arith?* controls coefficient arithmetic. If *ef-arith?* is false,
1358 ;; coeffs are elements of Zp, where p is the defined characteristic *gf-char*.
1359 ;; If *ef-arith?* is true, coeffs are interpreted as the integer representation
1360 ;; of a polynomial over Zp[x] reduced by the irreducible polynomial *gf-red*.
1362 (defun gf-cinv (c)
1363 (if *ef-arith?*
1364 (ef-cinv c)
1365 (maybe-char-is-fixnum-let ((c c))
1366 (cond
1367 ((= 0 c) (gf-merror (intl:gettext "gf coefficient inversion: Quotient by zero")))
1368 (t (inv-mod c *gf-char*)) )))) ; *gf-char* is prime
1370 (defun gf-cpow (c n)
1371 (if *ef-arith?*
1372 (ef-cpow c n)
1373 (maybe-char-is-fixnum-let ((c c))
1374 (power-mod c n *gf-char*) )))
1376 (defun gf-cmod (c)
1377 (if *ef-arith?*
1378 (ef-cmod c)
1379 (maybe-char-is-fixnum-let ((c c))
1380 (mod c *gf-char*) )))
1382 (defun gf-ctimes (a b)
1383 (if *ef-arith?*
1384 (ef-ctimes a b)
1385 (maybe-char-is-fixnum-let ((a a)(b b))
1386 (mod (* a b) *gf-char*) )))
1388 (defun gf-cplus-b (a b) ;; assumes that both 0 <= a,b < *gf-char*
1389 (cond
1390 (*ef-arith?* (ef-cplus-b a b))
1391 (t (maybe-char-is-fixnum-let ((a a)(b b))
1392 (let ((s (+ a b)))
1393 (if (< (the integer s) *gf-char*)
1395 (- (the integer s) *gf-char*) ))))))
1397 (defun gf-cminus-b (c) ;; assumes that 0 <= c < *gf-char*
1398 (cond
1399 ((= 0 c) 0)
1400 ((= 2 *gf-char*) c)
1401 (*ef-arith?* (ef-cminus-b c))
1402 (t (maybe-char-is-fixnum-let ((c c))
1403 (- *gf-char* c) ))))
1405 ;; ef coefficient arith :
1407 (defun ef-cinv (c)
1408 (declare (integer c))
1409 (cond
1410 ((= 0 c) (gf-merror (intl:gettext "ef coefficient inversion: Quotient by zero")))
1411 ($ef_coeff_inv (mfuncall '$ef_coeff_inv c))
1412 (*gf-logs?* (ef-cinv-by-table c))
1413 ((= 2 *gf-char*) (f2-inv c))
1414 (t (let ((*ef-arith?*))
1415 (gf-x2n (gf-inv (gf-n2x c) *gf-red*)) ))))
1417 (defun ef-cpow (c n)
1418 (cond
1419 ($ef_coeff_exp (mfuncall '$ef_coeff_exp c n))
1420 (*gf-logs?* (ef-cpow-by-table c n))
1421 ((= 2 *gf-char*) (f2-pow c n))
1422 (t (let ((*ef-arith?*))
1423 (gf-x2n (gf-pow (gf-n2x c) n *gf-red*)) ))))
1425 (defun ef-cmod (c)
1426 (declare (integer c))
1427 (cond
1428 ((plusp c)
1429 (cond
1430 ((< c *gf-ord*) c)
1431 ((= 2 *gf-char*) (f2-red c))
1432 (t (let ((*ef-arith?*))
1433 (gf-x2n (gf-nred (gf-n2x c) *gf-red*)) ))))
1435 (setq c (ef-cmod (abs c)))
1436 (let ((*ef-arith?* t)) (gf-ctimes (1- *gf-char*) c)) )))
1438 (defun ef-ctimes (a b)
1439 (cond
1440 ($ef_coeff_mult (mfuncall '$ef_coeff_mult a b))
1441 (*gf-logs?* (ef-ctimes-by-table a b))
1442 ((= 2 *gf-char*) (f2-times a b))
1443 (t (let ((*ef-arith?*))
1444 (gf-x2n (gf-times (gf-n2x a) (gf-n2x b) *gf-red*)) ))))
1446 (defun ef-cplus-b (a b)
1447 (cond
1448 ((= 2 *gf-char*) (logxor a b))
1449 ($ef_coeff_add (mfuncall '$ef_coeff_add a b))
1450 (*gf-logs?* (ef-cplus-by-table a b))
1451 (t (let ((*ef-arith?*))
1452 (gf-x2n (gf-nplus (gf-n2x a) (gf-n2x b))) ))))
1454 (defun ef-cminus-b (a)
1455 (cond
1456 ((= 0 a) 0)
1457 ((= 2 *gf-char*) a)
1458 ($ef_coeff_mult (mfuncall '$ef_coeff_mult (1- *gf-char*) a))
1459 (*gf-logs?* (ef-cminus-by-table a))
1460 (t (let ((*ef-arith?*))
1461 (gf-x2n (gf-nminus (gf-n2x a))) ))))
1463 ;; ef coefficient arith by lookup:
1465 (defun ef-ctimes-by-table (c d)
1466 (declare (fixnum c d))
1467 (cond
1468 ((or (= 0 c) (= 0 d)) 0)
1469 (t (let ((cd (+ (the fixnum (svref $gf_logs c))
1470 (the fixnum (svref $gf_logs d)) )))
1471 (svref $gf_powers (if (< (the integer cd) *gf-ord*) cd (- cd *gf-ord*))) ))))
1473 (defun ef-cminus-by-table (c)
1474 (declare (fixnum c))
1475 (cond
1476 ((= 0 c) 0)
1477 ((= 2 *gf-char*) c)
1478 (t (let ((e (ash *gf-ord* -1))) (declare (fixnum e))
1479 (setq c (svref $gf_logs c))
1480 (svref $gf_powers (the fixnum (if (< c e) (+ c e) (- c e)))) ))))
1482 (defun ef-cinv-by-table (c)
1483 (declare (fixnum c))
1484 (cond
1485 ((= 0 c) (gf-merror (intl:gettext "ef coefficient inversion: Quotient by zero")))
1486 (t (svref $gf_powers (- *gf-ord* (the fixnum (svref $gf_logs c))))) ))
1488 (defun ef-cplus-by-table (c d)
1489 (declare (fixnum c d))
1490 (cond
1491 ((= 0 c) d)
1492 ((= 0 d) c)
1493 (t (setq c (svref $gf_logs c) d (aref $gf_logs d))
1494 (let ((z (svref $gf_zech_logs (the fixnum (if (< d c) (+ *gf-ord* (- d c)) (- d c))))))
1495 (cond
1496 (z (incf z c)
1497 (svref $gf_powers (the fixnum (if (> z *gf-ord*) (- z *gf-ord*) z))) )
1498 (t 0) )))))
1500 (defun ef-cpow-by-table (c n)
1501 (declare (fixnum c n))
1502 (cond
1503 ((= 0 n) 1)
1504 ((= 0 c) 0)
1505 (t (svref $gf_powers
1506 (mod (* n (the fixnum (svref $gf_logs c))) *gf-ord*) )) ))
1509 (defun gf-pow-by-table (x n) ;; table lookup uses current *gf-red* for reduction
1510 (declare (fixnum n))
1511 (cond
1512 ((= 0 n) (list 0 1))
1513 ((null x) nil)
1514 (t (svref *gf-powers*
1515 (mod (* n (the fixnum (svref $gf_logs (gf-x2n x)))) *gf-ord*) )) ))
1518 ;; ef coefficient arith for binary base fields:
1520 (defun f2-red (a)
1521 (declare (optimize (speed 3) (safety 0)))
1522 (let* ((red *f2-red*)
1523 (ilen (integer-length red))
1524 (e 0) )
1525 (declare (fixnum e ilen))
1526 (do () ((= a 0) 0)
1527 (setq e (- (integer-length a) ilen))
1528 (when (< e 0) (return a))
1529 (setq a (logxor a (ash red e))) )))
1531 (defun f2-times (a b)
1532 (declare (optimize (speed 3) (safety 0)))
1533 (let* ((ilen (integer-length b))
1534 (a1 (ash a (1- ilen)))
1535 (ab a1) )
1536 (do ((i (- ilen 2) (1- i)) (k 0))
1537 ((< i 0) (f2-red ab))
1538 (declare (fixnum i k))
1539 (decf k)
1540 (when (logbitp i b)
1541 (setq a1 (ash a1 k)
1542 ab (logxor ab a1)
1543 k 0 )))))
1545 (defun f2-pow (a n) ;; assume n >= 0
1546 (declare (optimize (speed 3) (safety 0))
1547 (integer n) )
1548 (cond
1549 ((= n 0) 1)
1550 ((= a 0) 0)
1551 (t (do (res) (())
1552 (when (oddp n)
1553 (setq res (if res (f2-times a res) a))
1554 (when (= 1 n)
1555 (return-from f2-pow res) ))
1556 (setq n (ash n -1)
1557 a (f2-times a a)) ))))
1559 (defun f2-inv (b)
1560 (declare (optimize (speed 3) (safety 0)))
1561 (when (= b 0)
1562 (gf-merror (intl:gettext "f2 arithmetic: Quotient by zero")) )
1563 (let ((b1 1) (a *f2-red*) (a1 0) q r)
1564 (do ()
1565 ((= b 0) a1)
1566 (multiple-value-setq (q r) (f2-divide a b))
1567 (psetf a b b r)
1568 (psetf a1 b1 b1 (logxor (f2-times q b1) a1)) )))
1570 (defun f2-divide (a b)
1571 (declare (optimize (speed 3) (safety 0)))
1572 (cond
1573 ((= b 0)
1574 (gf-merror (intl:gettext "f2 arithmetic: Quotient by zero")) )
1575 ((= a 0) (values 0 0))
1577 (let ((ilen (integer-length b))
1578 (e 0)
1579 (q 0) )
1580 (declare (fixnum e ilen))
1581 (do () ((= a 0) (values q 0))
1582 (setq e (- (integer-length a) ilen))
1583 (when (< e 0) (return (values q a)))
1584 (setq q (logxor q (ash 1 e)))
1585 (setq a (logxor a (ash b e))) )))))
1587 ;; -------------------------------------------------------------------------- ;;
1590 #-gcl (eval-when (:compile-toplevel :load-toplevel :execute)
1591 (progn
1592 (define-compiler-macro gf-cmod (a)
1593 `(cond
1594 (*ef-arith?*
1595 (ef-cmod ,a) )
1596 ((and (typep *gf-char* 'fixnum) (typep ,a 'fixnum)) ;; maybe a > *gf-char*
1597 (let ((x ,a) (z *gf-char*)) (declare (fixnum x z))
1598 (the fixnum (mod x z)) ))
1600 (mod (the integer ,a) *gf-char*) )))
1602 (define-compiler-macro gf-ctimes (a b)
1603 `(cond
1604 (*ef-arith?*
1605 (ef-ctimes ,a ,b) )
1606 ((typep *gf-char* 'fixnum)
1607 (let ((x ,a) (y ,b) (z *gf-char*)) (declare (fixnum x y z))
1608 (the fixnum (mod (* x y) z)) ))
1610 (mod (* (the integer ,a) (the integer ,b)) *gf-char*) )))
1612 (define-compiler-macro gf-cplus-b (a b) ;; assumes that both 0 <= a,b < *gf-char*
1613 `(cond
1614 (*ef-arith?*
1615 (ef-cplus-b ,a ,b) )
1616 ((typep *gf-char* 'fixnum)
1617 (let ((x ,a) (y ,b) (z *gf-char*) (s 0)) (declare (fixnum x y z) (integer s))
1618 (setq s (the integer (+ x y)))
1619 (if (< s z) s (- s z)) ))
1621 (let ((x (+ (the integer ,a) (the integer ,b)))) (declare (integer x))
1622 (if (< x *gf-char*) x (- x *gf-char*)) ))))
1624 (define-compiler-macro gf-cminus-b (a) ;; assumes that 0 <= a < *gf-char*
1625 `(cond
1626 ((= 0 ,a) 0)
1627 (*ef-arith?*
1628 (ef-cminus-b ,a) )
1629 ((typep *gf-char* 'fixnum)
1630 (let ((x ,a) (z *gf-char*)) (declare (fixnum x z))
1631 (the fixnum (- z x)) ))
1633 (- *gf-char* (the integer ,a)) )))
1636 #+gcl (eval-when (compile load eval)
1637 (progn
1638 (push '((fixnum fixnum) fixnum #.(compiler::flags compiler::rfa)
1639 "(fixnum)(((long long)(#0))%((long long)(#1)))" )
1640 (get 'i% 'compiler::inline-always) )
1641 (push '((fixnum fixnum fixnum) fixnum #.(compiler::flags compiler::rfa)
1642 "(fixnum)((((long long)(#0))*((long long)(#1)))%((long long)(#2)))" )
1643 (get '*% 'compiler::inline-always) )
1644 (push '((fixnum fixnum fixnum) fixnum #.(compiler::flags compiler::rfa compiler::set)
1645 "@02;({long long _t=((long long)(#0))+((long long)(#1)),_w=((long long)(#2));_t<_w ? (fixnum)_t : (fixnum)(_t - _w);})" )
1646 (get '+%b 'compiler::inline-always) )
1647 (push '((fixnum fixnum) fixnum #.(compiler::flags compiler::rfa)
1648 "(fixnum)(((long long)(#1))-((long long)(#0)))" )
1649 (get 'neg%b 'compiler::inline-always) )
1651 (setf (get 'i% 'compiler::return-type) t)
1652 (setf (get '*% 'compiler::return-type) t)
1653 (setf (get '+%b 'compiler::return-type) t)
1654 (setf (get 'neg%b 'compiler::return-type) t)
1656 (si::define-compiler-macro gf-cmod (a)
1657 `(cond
1658 (*ef-arith?*
1659 (ef-cmod ,a) )
1660 ((and (typep *gf-char* 'fixnum) (typep ,a 'fixnum) (plusp ,a)) ;; maybe a > *gf-char*
1661 (let ((x ,a) (z *gf-char*)) (declare (fixnum x z))
1662 (i% x z) ))
1664 (mod (the integer ,a) *gf-char*) )))
1666 (si::define-compiler-macro gf-ctimes (a b) ;; assume that 0 <= a,b :
1667 `(cond
1668 (*ef-arith?*
1669 (ef-ctimes ,a ,b) )
1670 ((typep *gf-char*
1671 ',(if (< (integer-length most-positive-fixnum) 32) `fixnum `(signed-byte 32)) )
1672 (let ((x ,a) (y ,b) (z *gf-char*)) (declare (fixnum x y z))
1673 (*% x y z) ))
1675 (mod (* (the integer ,a) (the integer ,b)) *gf-char*) )))
1677 (si::define-compiler-macro gf-cplus-b (a b) ;; assume that both 0 <= a,b < *gf-char* :
1678 `(cond
1679 (*ef-arith?*
1680 (ef-cplus-b ,a ,b) )
1681 ((typep *gf-char*
1682 ',(if (< (integer-length most-positive-fixnum) 63) `fixnum `(signed-byte 63)) )
1683 (let ((x ,a) (y ,b) (z *gf-char*)) (declare (fixnum x y z))
1684 (+%b x y z) ))
1686 (let ((x (+ (the integer ,a) (the integer ,b)))) (declare (integer x))
1687 (if (< x *gf-char*) x (- x *gf-char*)) ))))
1689 (si::define-compiler-macro gf-cminus-b (a) ;; assume that 0 <= a < *gf-char* :
1690 `(cond
1691 ((= 0 ,a) 0)
1692 (*ef-arith?*
1693 (ef-cminus-b ,a) )
1694 ((typep *gf-char* 'fixnum)
1695 (let ((x ,a) (z *gf-char*)) (declare (fixnum x z))
1696 (neg%b x z) ))
1698 (- *gf-char* (the integer ,a)) )))
1701 ;; -----------------------------------------------------------------------------
1704 ;; setting the finite field and retrieving basic informations ------------------
1707 (defmfun $gf_set (p &optional a1 a2 a3) ;; deprecated
1708 (format t "`gf_set' is deprecated. ~%~\
1709 The user is asked to use `gf_set_data' instead.~%" )
1710 (when a2
1711 (format t "In future versions `gf_set_data' will only accept two arguments.~%") )
1712 ($gf_set_data p a1 a2 a3) )
1715 (defmfun $gf_set_data (p &optional a1 a2 a3) ;; opt: *gf-exp*, *gf-red*, *gf-fs-ord*
1716 (declare (ignore a2 a3)) ;; remove a2 a3 in next versions
1717 (let ((*ef-arith?*))
1718 (unless (and (integerp p) (primep p))
1719 (gf-merror (intl:gettext "`gf_set_data': Field characteristic must be a prime number.")) )
1720 ($gf_unset)
1721 (setq *gf-char* p)
1723 (when a1 ;; exponent or reduction poly
1724 (cond
1725 ((integerp a1)
1726 (unless (and (fixnump a1) (plusp a1))
1727 (gf-merror (intl:gettext "`gf_set_data': The exponent must be a positive fixnum.")) )
1728 (setq *gf-exp* a1) )
1730 (setq *gf-red* (gf-p2x-red a1 "gf_set_data")
1731 *gf-exp* (car *gf-red*)
1732 *gf-irred?* (gf-irr-p *gf-red* *gf-char* *gf-exp*) )) ))
1734 (gf-set-rat-header) ;; CRE-headers
1736 (unless *gf-red* ;; find irreducible reduction poly:
1737 (setq *gf-red* (if (= 1 *gf-exp*) (list 1 1) (gf-irr p *gf-exp*))
1738 *gf-irred?* t ))
1740 (when (= *gf-char* 2) (setq *f2-red* (gf-x2n *gf-red*)))
1742 (setq *gf-card* (expt p *gf-exp*)) ;; cardinality #(F)
1744 (setq *gf-ord* ;; group order #(F*)
1745 (cond
1746 ((= 1 *gf-exp*) (1- p))
1747 ((not *gf-irred?*) (gf-group-order *gf-char* *gf-red*))
1748 (t (1- (expt p *gf-exp*))) ))
1749 (let* (($intfaclim)
1750 (fs (get-factor-list *gf-ord*)) )
1751 (setq *gf-fs-ord* (sort fs #'< :key #'car)) ) ;; .. [pi, ei] ..
1753 (when *gf-irred?* (gf-precomp))
1755 (setq *gf-prim* ;; primitive element
1756 (cond
1757 ((= 1 *gf-exp*)
1758 (if (= 2 *gf-char*) (list 0 1)
1759 (list 0 (zn-primroot p *gf-ord* (mapcar #'car *gf-fs-ord*))) )) ;; .. pi .. (factors_only:true)
1761 (if *gf-irred?* (gf-prim) '$unknown) )))
1763 (setq *gf-char?* t *gf-red?* t *gf-data?* t) ;; global flags
1764 ($gf_get_data) )) ;; data structure
1767 (defun gf-set-rat-header ()
1768 (let ((modulus))
1769 (setq *gf-rat-header* (car ($rat '$x))) ))
1771 (defun gf-p2x-red (p fun)
1772 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
1773 (let* ((modulus) (x (car (prep1 p))))
1774 (unless (and (listp x)
1775 (every #'numberp (setq x (cdr x))) )
1776 (gf-merror (intl:gettext "`~m': Not suitable as reduction polynomial: ~m") fun p) )
1777 (setq x (gf-mod x))
1778 (unless (and (typep (car x) 'fixnum) (plusp (car x)))
1779 (gf-merror (intl:gettext "`~m': The exponent must be a positive fixnum.") fun) )
1780 (unless (eql 1 (cadr x))
1781 (gf-merror (intl:gettext "`~m': A monic reduction polynomial is assumed.") fun) )
1782 x ))
1785 (defmfun $ef_set_data (red)
1786 (ef-gf-field? "ef_set_data")
1787 ($ef_unset)
1788 (let ((*ef-arith?* t))
1789 (setq *ef-red* (gf-p2x-red red "ef_set_data")
1790 *ef-exp* (car *ef-red*)
1791 *ef-card* (expt *gf-card* *ef-exp*)
1792 *ef-irred?* (gf-irr-p *ef-red* *gf-card* *ef-exp*)
1793 *ef-ord* (if *ef-irred?*
1794 (1- *ef-card*)
1795 (gf-group-order *gf-card* *ef-red*) ))
1796 (let* (($intfaclim)
1797 (fs (get-factor-list *ef-ord*)) )
1798 (setq *ef-fs-ord* (sort fs #'< :key #'car)) )
1799 (when *ef-irred?* (ef-precomp))
1800 (setq *ef-data?* t
1801 *ef-red?* t
1802 *ef-prim* (if (= 1 *ef-exp*)
1803 (list 0 (let ((*ef-arith?*)) (gf-x2n *gf-prim*)))
1804 (if *ef-irred?* (ef-prim) '$unknown) )))
1805 ($ef_get_data) )
1808 (defstruct (gf-data (:print-function gf-data-short-print))
1809 char exp red prim card
1810 ord fs-ord fsx fsx-base-p x^p-powers )
1812 (defun gf-data-short-print (struct stream i)
1813 (declare (ignore struct i))
1814 (format stream "Structure [GF-DATA]") ) ;; wxMaxima returns this
1815 ;; terminal should return this too
1817 ;; returns a struct containing all data necessary to use gf_set_again (see below)
1818 (defmfun $gf_get_data ()
1819 (gf-data? "gf_get_data")
1820 (make-gf-data
1821 :char *gf-char* ; characteristic
1822 :exp *gf-exp* ; exponent
1823 :red *gf-red* ; reduction
1824 :prim *gf-prim* ; primitive
1825 :card *gf-card* ; cardinality
1826 :ord *gf-ord* ; order
1827 :fs-ord *gf-fs-ord* ; factors of order
1828 :fsx *gf-fsx* ; extended factors of order
1829 :fsx-base-p *gf-fsx-base-p* ; extended factors in base p
1830 :x^p-powers *gf-x^p-powers* )) ; pre-calculated powers
1832 (defstruct (ef-data (:print-function ef-data-short-print))
1833 exp red prim card
1834 ord fs-ord fsx fsx-base-q x^q-powers )
1836 (defun ef-data-short-print (struct stream i)
1837 (declare (ignore struct i))
1838 (format stream "Structure [EF-DATA]") )
1840 (defmfun $ef_get_data ()
1841 (ef-data? "ef_get_data")
1842 (make-ef-data
1843 :exp *ef-exp* ; exponent
1844 :red *ef-red* ; reduction
1845 :prim *ef-prim* ; primitive
1846 :card *ef-card* ; cardinality
1847 :ord *ef-ord* ; order
1848 :fs-ord *ef-fs-ord* ; factors of order
1849 :fsx *ef-fsx* ; extended factors of order
1850 :fsx-base-q *ef-fsx-base-q* ; extended factors in base q
1851 :x^q-powers *ef-x^q-powers* )) ; pre-calculated powers
1853 (defmfun $gf_info (&optional (t? t))
1854 (gf-data? "gf_info")
1855 (let ((no-prim (or (null *gf-prim*) (equal *gf-prim* '$unknown)))
1856 (*ef-arith?*) )
1857 (format t?
1858 "characteristic = ~a~:[, ~;~%~]~\
1859 reduction polynomial = ~a~:[, ~;~%~]~\
1860 primitive element = ~a~:[, ~;~%~]~\
1861 nr of elements = ~a~:[, ~;~%~]~\
1862 nr of units = ~a~:[, ~;~]~\
1863 ~:[~;~%nr of primitive elements = ~a~] ~%"
1864 *gf-char* t?
1865 (mfuncall '$string (gf-x2p *gf-red*)) t?
1866 (mfuncall '$string
1867 (if no-prim
1868 *gf-prim*
1869 (gf-x2p *gf-prim*) )) t?
1870 *gf-card* t?
1871 *gf-ord* (or t? no-prim) (not no-prim)
1872 (totient-by-fs-n *gf-fs-ord*) )))
1874 (defun totient-by-fs-n (fs-n)
1875 (let ((phi 1) p e)
1876 (dolist (f fs-n phi)
1877 (setq p (car f) e (cadr f))
1878 (setq phi (* phi (1- p) (expt p (1- e)))) )))
1880 (defmfun $gf_infolist () ;; enables testing gf_set_data in rtest
1881 (gf-data? "gf_infolist")
1882 (let ((*ef-arith?*))
1883 `((mlist simp)
1884 ,*gf-char*
1885 ,(gf-x2p *gf-red*)
1886 ,(if (or (null *gf-prim*) (equal *gf-prim* '$unknown))
1887 *gf-prim*
1888 (gf-x2p *gf-prim*) )
1889 ,*gf-card*
1890 ,*gf-ord* )))
1892 (defmfun $ef_info (&optional (t? t))
1893 (ef-data? "ef_info")
1894 (let ((no-prim (or (null *ef-prim*) (equal *ef-prim* '$unknown)))
1895 (*ef-arith?* t) )
1896 (format t?
1897 "reduction polynomial = ~a~:[, ~;~%~]~\
1898 primitive element = ~a~:[, ~;~%~]~\
1899 nr of elements = ~a~:[, ~;~%~]~\
1900 nr of units = ~a~:[, ~;~]~\
1901 ~:[~;~%nr of primitive elements = ~a~] ~%"
1902 (mfuncall '$string (gf-x2p *ef-red*)) t?
1903 (mfuncall '$string
1904 (if no-prim
1905 *ef-prim*
1906 (gf-x2p *ef-prim*) )) t?
1907 *ef-card* t?
1908 *ef-ord* (or t? no-prim) (not no-prim)
1909 (totient-by-fs-n *ef-fs-ord*) )))
1911 (defmfun $ef_infolist () ;; enables testing ef_set_data in rtest
1912 (ef-data? "ef_infolist")
1913 (let ((*ef-arith?* t))
1914 `((mlist simp)
1915 ,(gf-x2p *ef-red*)
1916 ,(if (or (null *ef-prim*) (equal *ef-prim* '$unknown))
1917 *ef-prim*
1918 (gf-x2p *ef-prim*) )
1919 ,*ef-card*
1920 ,*ef-ord* )))
1923 (defmfun $gf_unset ()
1924 (setq $gf_powers nil $gf_logs nil $gf_zech_logs nil *gf-powers* nil *gf-logs?* nil
1925 $gf_rat nil
1926 $ef_coeff_mult nil $ef_coeff_add nil $ef_coeff_inv nil $ef_coeff_exp nil
1927 *gf-rat-header* nil *gf-char* 0
1928 *gf-exp* 1 *gf-ord* 0 *gf-card* 0 ;; *gf-exp* = 1 when gf_set_data has no optional arg
1929 *gf-red* nil *f2-red* 0 *gf-prim* nil
1930 *gf-fs-ord* nil *gf-fsx* nil *gf-fsx-base-p* nil *gf-x^p-powers* nil
1931 *gf-char?* nil *gf-red?* nil *gf-irred?* nil *gf-data?* nil )
1934 (defmfun $ef_unset ()
1935 (setq *ef-exp* 0 *ef-ord* 0 *ef-card* 0
1936 *ef-red* nil *ef-prim* nil
1937 *ef-fs-ord* nil *ef-fsx* nil *ef-fsx-base-q* nil *ef-x^q-powers* nil
1938 *ef-red?* nil *ef-irred?* nil *ef-data?* nil )
1942 ;; Minimal set
1943 ;; Just set characteristic and reduction poly to allow basic arithmetics on the fly.
1944 (defmfun $gf_minimal_set (p &optional (red))
1945 (unless (and (integerp p) (primep p))
1946 (gf-merror (intl:gettext "First argument to `gf_minimal_set' must be a prime number.")) )
1947 ($gf_unset)
1948 (setq *gf-char* p
1949 *gf-char?* t )
1950 (gf-set-rat-header)
1951 (let ((*ef-arith?*))
1952 (when red
1953 (setq *gf-red* (gf-p2x-red red "gf_minimal_set")
1954 *gf-red?* t
1955 *gf-exp* (car *gf-red*) ))
1956 (format nil "characteristic = ~a, reduction polynomial = ~a"
1957 *gf-char*
1958 (if red (mfuncall '$string (gf-x2p *gf-red*)) "false") )))
1961 (defmfun $ef_minimal_set (red)
1962 (ef-gf-field? "ef_minimal_set")
1963 ($ef_unset)
1964 (let ((*ef-arith?* t))
1965 (when red
1966 (setq *ef-red* (gf-p2x-red red "ef_minimal_set")
1967 *ef-exp* (car *ef-red*)
1968 *ef-red?* t ))
1969 (format nil "reduction polynomial = ~a"
1970 (if red (mfuncall '$string (gf-x2p *ef-red*)) "false") )))
1973 (defmfun $gf_characteristic ()
1974 (gf-char? "gf_characteristic")
1975 *gf-char* )
1977 (defmfun $gf_exponent ()
1978 (gf-red? "gf_exponent")
1979 *gf-exp* )
1981 (defmfun $gf_reduction ()
1982 (gf-red? "gf_reduction")
1983 (when *gf-red* (let ((*ef-arith?*)) (gf-x2p *gf-red*))) )
1985 (defmfun $gf_cardinality ()
1986 (gf-data? "gf_cardinality")
1987 *gf-card* )
1990 (defmfun $ef_exponent ()
1991 (ef-red? "ef_exponent")
1992 *ef-exp* )
1994 (defmfun $ef_reduction ()
1995 (ef-red? "ef_reduction")
1996 (when *ef-red* (let ((*ef-arith?* t)) (gf-x2p *ef-red*))) )
1998 (defmfun $ef_cardinality ()
1999 (ef-data? "ef_cardinality")
2000 *ef-card* )
2003 ;; Reuse data and results from a previous gf_set_data
2004 (defmfun $gf_set_again (data)
2005 (unless (gf-data-p data)
2006 (gf-merror (intl:gettext
2007 "Argument to `gf_set_again' must be a return value of `gf_set_data'." )))
2008 ($gf_unset)
2009 (gf-set-rat-header)
2010 (setq *gf-char* (gf-data-char data)
2011 *gf-exp* (gf-data-exp data)
2012 *gf-red* (gf-data-red data)
2013 *gf-prim* (gf-data-prim data)
2014 *gf-card* (gf-data-card data)
2015 *gf-ord* (gf-data-ord data)
2016 *gf-fs-ord* (gf-data-fs-ord data)
2017 *gf-fsx* (gf-data-fsx data)
2018 *gf-fsx-base-p* (gf-data-fsx-base-p data)
2019 *gf-x^p-powers* (gf-data-x^p-powers data)
2020 *gf-irred?* (= *gf-ord* (1- *gf-card*))
2021 *gf-char?* t
2022 *gf-red?* t
2023 *gf-data?* t ))
2025 (defmfun $ef_set_again (data)
2026 (ef-gf-field? "ef_set_again")
2027 (unless (ef-data-p data)
2028 (gf-merror (intl:gettext
2029 "Argument to `ef_set_again' must be a return value of `ef_set_data'." )))
2030 ($ef_unset)
2031 (setq *ef-exp* (ef-data-exp data)
2032 *ef-red* (ef-data-red data)
2033 *ef-prim* (ef-data-prim data)
2034 *ef-card* (ef-data-card data)
2035 *ef-ord* (ef-data-ord data)
2036 *ef-fs-ord* (ef-data-fs-ord data)
2037 *ef-fsx* (ef-data-fsx data)
2038 *ef-fsx-base-q* (ef-data-fsx-base-q data)
2039 *ef-x^q-powers* (ef-data-x^q-powers data)
2040 *ef-irred?* (= *ef-ord* (1- *ef-card*))
2041 *ef-red?* t
2042 *ef-data?* t ))
2044 ;; -----------------------------------------------------------------------------
2047 ;; lookup tables ---------------------------------------------------------------
2050 (defmfun $gf_make_arrays ()
2051 (format t "`gf_make_arrays' is deprecated. ~%~\
2052 The user is asked to use `gf_make_logs' instead.~%" )
2053 ($gf_make_logs) )
2055 (defmfun $gf_make_logs () ;; also zech-logs and antilogs
2056 (gf-field? "gf_make_logs")
2057 (let ((*ef-arith?*)) (gf-make-logs)) )
2059 (defun gf-make-logs ()
2060 (unless (typep *gf-ord* 'fixnum)
2061 (gf-merror (intl:gettext "`gf_make_logs': group order must be a fixnum.")) )
2062 (let ((x (list 0 1)) (ord *gf-ord*) (primx *gf-prim*) (red *gf-red*))
2063 (declare (fixnum ord))
2065 ;; power table of the field, where the i-th element is (the numerical
2066 ;; equivalent of) the field element e^i, where e is a primitive element
2068 (setq $gf_powers (make-array (1+ ord) :element-type 'integer)
2069 *gf-powers* (make-array (1+ ord) :element-type 'list :initial-element nil) )
2070 (setf (svref $gf_powers 0) 1
2071 (svref *gf-powers* 0) (list 0 1) )
2072 (do ((i 1 (1+ i)))
2073 ((> i ord))
2074 (declare (fixnum i))
2075 (setq x (gf-times x primx red))
2076 (setf (svref $gf_powers i) (gf-x2n x)
2077 (svref *gf-powers* i) x ))
2079 ;; log table: the inverse lookup of the power table
2081 (setq $gf_logs (make-array (1+ ord) :initial-element nil))
2082 (do ((i 0 (1+ i)))
2083 ((= i ord))
2084 (declare (fixnum i))
2085 (setf (svref $gf_logs (svref $gf_powers i)) i) )
2087 ;; zech-log table: lookup table for efficient addition
2089 (setq $gf_zech_logs (make-array (1+ ord) :initial-element nil))
2090 (do ((i 0 (1+ i)) (one (list 0 1)))
2091 ((> i ord))
2092 (declare (fixnum i))
2093 (setf (svref $gf_zech_logs i)
2094 (svref $gf_logs (gf-x2n (gf-plus (svref *gf-powers* i) one))) ))
2096 (setq *gf-logs?* t)
2097 `((mlist simp) ,$gf_powers ,$gf_logs ,$gf_zech_logs) ))
2099 (defun gf-clear-tables ()
2100 (setq $gf_powers nil
2101 $gf_logs nil
2102 $gf_zech_logs nil
2103 *gf-logs?* nil ))
2105 ;; -----------------------------------------------------------------------------
2108 ;; converting to/from internal representation ----------------------------------
2110 ;; user level <---> internal
2111 ;; 0 nil
2112 ;; integer # 0 (0 integer') where integer' = mod(integer, *gf-char*)
2113 ;; x (1 1)
2114 ;; x^4 + 3*x^2 + 4 (4 1 2 3 0 4)
2116 ;; This representation uses the term part of the internal CRE representation.
2117 ;; The coeffcients are exclusively positive: 1, 2, ..., (*gf-char* -1)
2118 ;; Header informations are stored in *gf-rat-header*.
2120 ;; gf_set_data(5, 4)$
2121 ;; :lisp `(,*gf-char* ,*gf-exp*)
2122 ;; (5 4)
2123 ;; p : x^4 + 3*x^2 - 1$
2124 ;; :lisp ($rat $p)
2125 ;; ((MRAT SIMP ($X) (X33303)) (X33303 4 1 2 3 0 -1) . 1)
2126 ;; :lisp (gf-p2x $p)
2127 ;; (4 1 2 3 0 4)
2128 ;; :lisp *gf-rat-header*
2129 ;; (MRAT SIMP ($X) (X33303))
2131 ;; Remark: I compared the timing results of the arithmetic functions using this
2132 ;; data structure to arithmetics using an array implementation and in case of
2133 ;; modulus 2 to an implementation using bit-arithmetics over integers.
2134 ;; It turns out that in all cases the timing advantages of other data structures
2135 ;; were consumed by conversions from/to the top-level.
2136 ;; So for sparse polynomials the CRE representation seems to fit best.
2139 (defun gf-p2x (p)
2140 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2141 (setq p (car (let ((modulus)) (prep1 p))))
2142 (cond
2143 ((integerp p)
2144 (cond
2145 ((= p 0) nil)
2146 (t (setq p (gf-cmod p))
2147 (if (= p 0) nil (list 0 p)) )))
2149 (setq p (gf-mod (cdr p)))
2150 (if (typep (car p) 'fixnum)
2152 (gf-merror (intl:gettext "Exponents are limited to fixnums.")) ))))
2155 ;; version of gf-p2x that doesn't apply mod reduction
2157 (defun gf-p2x-raw (p)
2158 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2159 (setq p (car (let ((modulus)) (prep1 p))))
2160 (cond
2161 ((integerp p) (if (= 0 p) nil (list 0 p)))
2162 (t (setq p (cdr p))
2163 (unless (every #'numberp p)
2164 (gf-merror (intl:gettext "gf: polynomials must be univariate.")) )
2165 p )))
2168 (defun gf-x2p (x)
2169 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2170 (setq x
2171 (cond
2172 ((null x) 0)
2173 ((= 0 (the fixnum (car x))) (gf-cp2smod (cadr x)))
2174 (t (gf-np2smod x)) ))
2175 (if (eql $gf_rat t)
2176 (gf-x2cre x)
2177 (gf-disrep x) ))
2179 ;; depending on $gf_rat gf-x2p returns a CRE or a ratdisrepped expression
2181 (defun gf-x2cre (x)
2182 #+ (or ccl ecl gcl sbcl) (declare (optimize (speed 3) (safety 0)))
2183 (if (integerp x)
2184 `(,*gf-rat-header* ,x . 1)
2185 `(,*gf-rat-header* ,(cons (caar (cdddr *gf-rat-header*)) x) . 1) ))
2187 (defun gf-disrep (x &optional (var '$x))
2188 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2189 (if (integerp x) x
2190 (maybe-char-is-fixnum-let ((c 0))
2191 (do ((not-plus? (null (cddr x))) p (e 0))
2192 ((null x) (if not-plus? (car p) (cons '(mplus simp) p)))
2193 (declare (fixnum e))
2194 (setq e (car x) c (cadr x) x (cddr x)
2195 p (cond
2196 ((= 0 e)
2197 (cons c p) )
2198 ((= 1 e)
2199 (if (= 1 c)
2200 (cons var p)
2201 (cons `((mtimes simp) ,c ,var) p) ))
2202 ((= 1 c)
2203 (cons `((mexpt simp) ,var ,e) p) )
2205 (cons `((mtimes simp) ,c ((mexpt simp) ,var ,e)) p) )))))))
2207 ;; -----------------------------------------------------------------------------
2210 ;; evaluation and adjustment ---------------------------------------------------
2213 ;; an arbitrary polynomial is evaluated in a given field
2215 (defmfun $gf_eval (a)
2216 (gf-char? "gf_eval")
2217 (let ((*ef-arith?*)) (gf-eval a *gf-red* "gf_eval")) )
2219 (defmfun $ef_eval (a)
2220 (ef-gf-field? "ef_eval")
2221 (let ((*ef-arith?* t))
2222 (unless (equal a ($expand a))
2223 (gf-merror (intl:gettext "`ef_eval': The argument must be an expanded polynomial.")) )
2224 (gf-eval a *ef-red* "ef_eval") ))
2226 (defun gf-eval (a red fun)
2227 (setq a (let ((modulus)) (car (prep1 a))))
2228 (cond
2229 ((integerp a) (gf-cmod a))
2231 (setq a (gf-mod (cdr a)))
2232 (and a (not (typep (car a) 'fixnum))
2233 (gf-merror (intl:gettext "`~m': The exponent is expected to be a fixnum.") fun) )
2234 (gf-x2p (gf-nred a red)) )))
2237 ;; gf-mod adjusts arbitrary integer coefficients (pos, neg or unbounded)
2239 (defun gf-mod (x)
2240 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2241 (if (null x) nil
2242 (maybe-char-is-fixnum-let ((c 0))
2243 (do ((r x (cddr r)) res)
2244 ((null r) (nreverse res))
2245 (unless (numberp (cadr r))
2246 (gf-merror (intl:gettext "gf: polynomials must be univariate.")) )
2247 (setq c (gf-cmod (cadr r)))
2248 (unless (= c 0) (setq res (cons c (cons (car r) res)))) ))))
2250 ;; positive 2 symmetric mod:
2252 (defun gf-np2smod (x) ;; modifies x
2253 (cond
2254 ((null x) nil)
2255 ((not $gf_balanced) x)
2256 (*ef-arith?*
2257 (*f-np2smod x *gf-card* #'(lambda (c) (neg (gf-ctimes (1- *gf-char*) c)))) )
2259 (*f-np2smod x *gf-char* #'(lambda (c) (- (the integer c) *gf-char*))) )))
2261 (defun *f-np2smod (x p cp2smod-fn)
2262 (if (null x) x
2263 (maybe-char-is-fixnum-let ((p2 (ash p -1)))
2264 (do ((r (cdr x) (cddr r))) (())
2265 (when (> (the integer (car r)) p2)
2266 (rplaca r (funcall cp2smod-fn (car r))) )
2267 (when (null (cdr r)) (return x)) ))))
2269 ;; adjust a coefficient to a symmetric modulus:
2270 (defun gf-cp2smod (c)
2271 (cond
2272 ((not $gf_balanced) c)
2273 (*ef-arith?*
2274 (if (> c (ash *gf-card* -1)) (neg (gf-ctimes c (1- *gf-char*))) c) )
2276 (if (> c (ash *gf-char* -1)) (- (the integer c) *gf-char*) c) )))
2278 ;; -----------------------------------------------------------------------------
2281 ;; arithmetic in Galois Fields - Maxima level functions ------------------------
2284 ;; gf:
2286 (defmfun $gf_neg (a)
2287 (gf-char? "gf_neg")
2288 (let ((*ef-arith?*))
2289 (gf-x2p (gf-nminus (gf-p2x a))) ))
2291 (defmfun $gf_add (&rest args)
2292 (gf-char? "gf_add")
2293 (let ((*ef-arith?*))
2294 (setq args (mapcar #'gf-p2x args))
2295 (gf-x2p (reduce #'gf-plus args)) ))
2297 (defmfun $gf_sub (&rest args)
2298 (gf-char? "gf_sub")
2299 (let ((*ef-arith?*))
2300 (setq args (mapcar #'gf-p2x args))
2301 (gf-x2p (gf-plus (car args) (gf-minus (reduce #'gf-plus (cdr args))))) ))
2303 (defmfun $gf_mult (&rest args)
2304 (gf-char? "gf_mult")
2305 (let ((*ef-arith?*))
2306 (setq args (mapcar #'gf-p2x args))
2307 (and (not *gf-red*)
2308 (not (some #'null args))
2309 (not (typep (apply #'+ (mapcar #'car args)) 'fixnum))
2310 (gf-merror (intl:gettext "`gf_mult': Resulting exponent won't be a fixnum.")) )
2311 (gf-x2p (reduce #'(lambda (x y) (gf-times x y *gf-red*)) args)) ))
2313 (defmfun $gf_reduce (a b)
2314 (gf-char? "gf_reduce")
2315 (let ((*ef-arith?*))
2316 (gf-x2p (gf-nrem (gf-p2x a) (gf-p2x b))) ))
2318 (defmfun $gf_inv (a)
2319 (gf-red? "gf_inv")
2320 (let ((*ef-arith?*))
2321 (setq a (gf-inv (gf-p2x a) *gf-red*))
2322 (when a (gf-x2p a)) )) ;; a is nil in case the inverse does not exist
2324 (defmfun $gf_div (&rest args)
2325 (gf-red? "gf_div")
2326 (unless (cadr args)
2327 (gf-merror (intl:gettext "`gf_div' needs at least two arguments." )) )
2328 (let* ((*ef-arith?*)
2329 (a2 (mapcar #'gf-p2x args))
2330 (a2 (cons (car a2) (mapcar #'(lambda (x) (gf-inv x *gf-red*)) (cdr a2)))) )
2331 (cond
2332 ((some #'null (cdr a2)) ;; but check if exact division is possible ..
2333 (let ((q (gf-p2x (car args))) r)
2334 (setq args (cdr args))
2335 (do ((d (car args) (car args)))
2336 ((null d) (gf-x2p q))
2337 (multiple-value-setq (q r) (gf-divide q (gf-p2x d)))
2338 (when r (return)) ;; .. in case it is not return false
2339 (setq args (cdr args)) )))
2340 (t ;; a / b = a * b^-1 :
2341 (gf-x2p (reduce #'(lambda (x y) (gf-times x y *gf-red*)) a2)) ))))
2343 (defmfun $gf_exp (a n)
2344 (gf-char? "gf_exp")
2345 (let ((*ef-arith?*))
2346 (cond
2347 ((not n)
2348 (gf-merror (intl:gettext "`gf_exp' needs two arguments.")) )
2349 ((not (integerp n))
2350 (gf-merror (intl:gettext "Second argument to `gf_exp' must be an integer.")) )
2351 ((< (the integer n) 0)
2352 (unless *gf-red*
2353 (gf-merror (intl:gettext "`gf_exp': Unknown reduction polynomial.")) )
2354 (setq a (gf-inv (gf-p2x a) *gf-red*))
2355 (when a ($gf_exp (gf-x2p a) (neg n))) ) ;; a is nil in case the inverse does not exist
2356 (*gf-logs?*
2357 (gf-x2p (gf-pow-by-table (gf-p2x a) n)) )
2358 ((and *gf-irred?* *gf-x^p-powers*)
2359 (gf-x2p (gf-pow$ (gf-p2x a) n *gf-red*)) )
2361 (setq a (gf-p2x a))
2362 (and (not *gf-red*)
2363 (not (null a))
2364 (not (typep (* n (car a)) 'fixnum))
2365 (gf-merror (intl:gettext "`gf_exp': Resulting exponent won't be a fixnum.")) )
2366 (gf-x2p (gf-pow a n *gf-red*)) ))))
2368 ;; ef:
2370 (defmfun $ef_neg (a)
2371 (ef-gf-field? "ef_neg")
2372 (let ((*ef-arith?* t))
2373 (gf-x2p (gf-nminus (gf-p2x a))) ))
2375 (defmfun $ef_add (&rest args)
2376 (ef-gf-field? "ef_add")
2377 (let ((*ef-arith?* t))
2378 (setq args (mapcar #'gf-p2x args))
2379 (gf-x2p (reduce #'gf-plus args)) ))
2381 (defmfun $ef_sub (&rest args)
2382 (ef-gf-field? "ef_sub")
2383 (let ((*ef-arith?* t))
2384 (setq args (mapcar #'gf-p2x args))
2385 (gf-x2p (gf-plus (car args) (gf-minus (reduce #'gf-plus (cdr args))))) ))
2387 (defmfun $ef_mult (&rest args)
2388 (ef-gf-field? "ef_mult")
2389 (let ((*ef-arith?* t)
2390 (red *ef-red*) )
2391 (setq args (mapcar #'gf-p2x args))
2392 (and (not red)
2393 (not (some #'null args))
2394 (not (typep (apply #'+ (mapcar #'car args)) 'fixnum))
2395 (gf-merror (intl:gettext "`ef_mult': Resulting exponent won't be a fixnum.")) )
2396 (gf-x2p (reduce #'(lambda (x y) (gf-times x y red)) args)) ))
2398 (defmfun $ef_reduce (a b)
2399 (ef-gf-field? "ef_reduce")
2400 (let ((*ef-arith?* t))
2401 (gf-x2p (gf-nrem (gf-p2x a) (gf-p2x b))) ))
2403 (defmfun $ef_inv (a)
2404 (ef-red? "ef_inv")
2405 (let ((*ef-arith?* t))
2406 (setq a (gf-inv (gf-p2x a) *ef-red*))
2407 (when a (gf-x2p a)) ))
2409 (defmfun $ef_div (&rest args)
2410 (ef-red? "ef_div")
2411 (unless (cadr args)
2412 (gf-merror (intl:gettext "`ef_div' needs at least two arguments." )) )
2413 (let ((*ef-arith?* t)
2414 (red *ef-red*) )
2415 (setq args (mapcar #'gf-p2x args))
2416 (setq args
2417 (cons (car args) (mapcar #'(lambda (x) (gf-inv x red)) (cdr args))) )
2418 (cond
2419 ((null (car args)) 0)
2420 ((some #'null (cdr args)) nil)
2421 (t (gf-x2p (reduce #'(lambda (x y) (gf-times x y red)) args))) )))
2423 (defmfun $ef_exp (a n)
2424 (ef-gf-field? "ef_exp")
2425 (let ((*ef-arith?* t))
2426 (cond
2427 ((< (the integer n) 0)
2428 (unless *ef-red*
2429 (gf-merror (intl:gettext "`ef_exp': Unknown reduction polynomial.")) )
2430 (setq a (gf-inv (gf-p2x a) *ef-red*))
2431 (when a ($ef_exp (gf-x2p a) (neg n))) )
2432 ((and *ef-irred?* *ef-x^q-powers*)
2433 (gf-x2p (gf-pow$ (gf-p2x a) n *ef-red*)) )
2435 (setq a (gf-p2x a))
2436 (and (not *ef-red*)
2437 (not (null a))
2438 (not (typep (* n (car a)) 'fixnum))
2439 (gf-merror (intl:gettext "`ef_exp': Resulting exponent won't be a fixnum.")) )
2440 (gf-x2p (gf-pow a n *ef-red*)) ))))
2442 ;; -----------------------------------------------------------------------------
2445 ;; arithmetic in Galois Fields - Lisp level functions --------------------------
2448 ;; Both gf (base field) and ef (extension field) Maxima level functions use
2449 ;; this Lisp level functions. The switch *ef-arith?* controls how the coefficients
2450 ;; were treated. The coefficient functions gf-ctimes and friends behave
2451 ;; differently depending on *ef-arith?*. See above definitions.
2453 ;; Remark: A prefixed character 'n' indicates a destructive function.
2455 ;; c * x
2457 (defun gf-xctimes (x c)
2458 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2459 (maybe-char-is-fixnum-let ((c c))
2460 (if (or (= 0 c) (null x)) nil
2461 (do* ((res (list (car x) (gf-ctimes c (cadr x))))
2462 (r (cdr res) (cddr r))
2463 (rx (cddr x) (cddr rx)) )
2464 ((null rx) res)
2465 (rplacd r (list (car rx) (gf-ctimes c (cadr rx)))) ))))
2467 (defun gf-nxctimes (x c) ;; modifies x
2468 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2469 (maybe-char-is-fixnum-let ((c c))
2470 (if (or (= 0 c) (null x)) nil
2471 (do ((r (cdr x) (cddr r)))
2472 ((null r) x)
2473 (rplaca r (gf-ctimes c (car r))) ))))
2475 ;; c*v^e * x
2477 (defun gf-xectimes (x e c)
2478 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2479 (declare (fixnum e))
2480 (maybe-char-is-fixnum-let ((c c))
2481 (if (or (= 0 c) (null x)) nil
2482 (do* ((res (list (+ e (the fixnum (car x))) (gf-ctimes c (cadr x))))
2483 (r (cdr res) (cddr r))
2484 (rx (cddr x) (cddr rx)) )
2485 ((null rx) res)
2486 (rplacd r (list (+ e (the fixnum (car rx))) (gf-ctimes c (cadr rx)))) ))))
2488 ;; v^e * x
2490 (defun gf-nxetimes (x e) ;; modifies x
2491 (if (null x) x
2492 (do ((r x (cddr r)))
2493 ((null r) x)
2494 (rplaca r (+ e (car r))) )))
2496 ;; - x
2498 (defun gf-minus (x)
2499 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2500 (if (or (null x) (= 2 *gf-char*)) x
2501 (do* ((res (list (car x) (gf-cminus-b (cadr x))))
2502 (r (cdr res) (cddr r))
2503 (rx (cddr x) (cddr rx)) )
2504 ((null rx) res)
2505 (rplacd r (list (car rx) (gf-cminus-b (cadr rx)))) )))
2507 (defun gf-nminus (x) ;; modifies x
2508 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2509 (if (or (null x) (= 2 *gf-char*)) x
2510 (do ((r (cdr x) (cddr r))) (())
2511 (rplaca r (gf-cminus-b (car r)))
2512 (when (null (cdr r)) (return x)) )))
2514 ;; x + y
2516 (defun gf-plus (x y)
2517 (cond
2518 ((null x) y)
2519 ((null y) x)
2520 (t (gf-nplus (copy-list x) y)) ))
2522 ;; merge y into x
2524 (defun gf-nplus (x y) ;; modifies x
2525 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2526 (cond
2527 ((null x) y)
2528 ((null y) x)
2530 (maybe-char-is-fixnum-let ((cy 0)(c 0))
2531 (prog ((ex 0)(ey 0) r) (declare (fixnum ex ey))
2533 (setq ex (car x) ey (car y) cy (cadr y))
2534 (cond
2535 ((> ey ex)
2536 (setq x (cons ey (cons cy x)) y (cddr y)) )
2537 ((= ey ex)
2538 (setq c (gf-cplus-b (cadr x) cy) y (cddr y))
2539 (cond
2540 ((= 0 c)
2541 (when (null (setq x (cddr x))) (return y))
2542 (when (null y) (return x))
2543 (go a1) )
2544 (t (rplaca (cdr x) c)) ))
2545 (t (setq r (cdr x)) (go b)) )
2546 (setq r (cdr x))
2548 (when (null y) (return x))
2549 (setq ey (car y) cy (cadr y))
2551 (while (and (cdr r) (> (the fixnum (cadr r)) ey))
2552 (setq r (cddr r)) )
2553 (cond
2554 ((null (cdr r)) (rplacd r y) (return x))
2555 ((> ey (the fixnum (cadr r)))
2556 (rplacd r (cons ey (cons cy (cdr r))))
2557 (setq r (cddr r) y (cddr y)) )
2559 (setq c (gf-cplus-b (caddr r) cy) y (cddr y))
2560 (if (= 0 c)
2561 (rplacd r (cdddr r))
2562 (rplaca (setq r (cddr r)) c) )) )
2563 (go a) )))))
2565 ;; x + c*v^e*y
2567 (defun gf-xyecplus (x y e c)
2568 (cond
2569 ((null y) x)
2570 ((null x) (gf-xectimes y e c))
2571 (t (gf-nxyecplus (copy-list x) y e c) )))
2573 ;; merge c*v^e*y into x
2575 (defun gf-nxyecplus (x y e c) ;; modifies x
2576 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2577 (cond
2578 ((null y) x)
2579 ((null x) (gf-xectimes y e c))
2581 (maybe-char-is-fixnum-let ((cy 0) (cc 0))
2582 (prog ((e e) (ex 0) (ey 0) r) (declare (fixnum e ex ey))
2584 (setq ey (+ (the fixnum (car y)) e)
2585 cy (gf-ctimes c (cadr y))
2586 ex (car x) )
2587 (cond
2588 ((> ey ex)
2589 (setq x (cons ey (cons cy x)) y (cddr y)) )
2590 ((= ey ex)
2591 (setq cc (gf-cplus-b (cadr x) cy) y (cddr y))
2592 (cond
2593 ((= 0 cc)
2594 (when (null (setq x (cddr x))) (return (gf-xectimes y e c)))
2595 (when (null y) (return x))
2596 (go a1) )
2597 (t (rplaca (cdr x) cc)) ))
2598 (t (setq r (cdr x)) (go b)) )
2599 (setq r (cdr x))
2601 (when (null y) (return x))
2602 (setq ey (+ (the fixnum (car y)) e)
2603 cy (gf-ctimes c (cadr y)) )
2605 (when (null (cdr r)) (go d))
2606 (setq ex (cadr r))
2607 (cond
2608 ((> ey ex)
2609 (rplacd r (cons ey (cons cy (cdr r))))
2610 (setq r (cddr r) y (cddr y))
2611 (go a) )
2612 ((= ey ex)
2613 (setq cc (gf-cplus-b (caddr r) cy))
2614 (if (= 0 cc)
2615 (rplacd r (cdddr r))
2616 (rplaca (setq r (cddr r)) cc) )
2617 (setq y (cddr y))
2618 (go a) )
2619 (t (setq r (cddr r)) (go b)) )
2621 (do () ((null y))
2622 (setq x (nconc x (list (+ (the fixnum (car y)) e) (gf-ctimes c (cadr y))))
2623 y (cddr y) ))
2624 (return x) ) ))))
2626 ;; x * y
2628 ;; For sparse polynomials (in Galois Fields) with not too high degrees
2629 ;; simple school multiplication is faster than Karatsuba.
2631 ;; x * y = (x1 + x2 + ... + xk) * (y1 + y2 + ... + yn)
2632 ;; = x1 * (y1 + y2 + ... + yn) + x2 * (y1 + y2 + ... + yn) + ...
2634 ;; where e.g. xi = ci*v^ei
2636 (defun gf-times (x y red)
2637 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2638 (if (or (null x) (null y)) nil
2639 (maybe-char-is-fixnum-let ((c 0) (cx 0))
2640 (do* ((res (gf-xectimes y (car x) (cadr x))) ;; x1 * (y1 + y2 + ... + yn). summands in res are sorted. res is a new list.
2641 (r1 (cdr res)) ;; r1 marks the place of xi*y1 in res. x[i+1]*y1 will be smaller.
2642 ry ;; ry iterates over y
2643 (x (cddr x) (cddr x)) ;; each loop: res += xi * (y1 + y2 + ... + yn)
2644 (e 0)(ex 0) )
2645 ((or (null x)(null y)) (gf-nred res red))
2646 (declare (fixnum e ex))
2647 (setq ry y ;; start with y1 again
2648 ex (car x) cx (cadr x) ;; xi = ci*v^ei
2649 e (+ ex (the fixnum (car ry))) ;; c*v^e = xi*y1
2650 c (gf-ctimes (cadr ry) cx) ) ;; zero divisor free mult in Fp^n
2652 (while (and (cdr r1) (< e (the fixnum (cadr r1))))
2653 (setq r1 (cddr r1)) ) ;; mark the position of xi*y1
2655 (do ((r r1)) (()) ;; merge xi*y1 into res and then xi*y2, etc...
2656 (cond
2657 ((or (null (cdr r)) (> e (the fixnum (cadr r))))
2658 (rplacd r (cons e (cons c (cdr r))))
2659 (setq r (cddr r)) )
2660 ((= 0 (setq c (gf-cplus-b (caddr r) c)))
2661 (rplacd r (cdddr r)) )
2662 (t (rplaca (setq r (cddr r)) c)) )
2664 (when (null (setq ry (cddr ry))) (return))
2665 (setq e (+ ex (the fixnum (car ry)))
2666 c (gf-ctimes (cadr ry) cx) )
2668 (while (and (cdr r) (< e (the fixnum (cadr r))))
2669 (setq r (cddr r)) ) )) )))
2671 ;; x^2
2673 ;; x * x = (x_1 + x_2 + ... + x_k) * (x_1 + x_2 + ... + x_k)
2675 ;; = x_1^2 + 2*x_1*x_2 + 2*x_1*x_3 + ... + x_2^2 + 2*x_2*x_3 + 2*x_2*x_4 + ...
2677 ;; = x_k^2 + x_{k-1}^2 + 2*x_{k-1}*xk + x_{k-2}^2 + 2*x_{k-2}*x_{k-1} + 2*x_{k-2}*xk + ...
2679 ;; The reverse needs some additional consing but is slightly faster.
2681 (defun gf-sq (x red)
2682 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2683 (cond
2684 ((null x) x)
2685 ((and (not *ef-arith?*) (eql *gf-char* 2)) ;; the mod 2 case degrades to x_1^2 + x_2^2 + ... + x_k^2
2686 (do (res)
2687 ((null x) (gf-nred (nreverse res) red))
2688 (setq res (cons 1 (cons (ash (car x) 1) res))
2689 x (cddr x) ) ))
2691 (maybe-char-is-fixnum-let ((ci 0)(*2ci 0)(c 0))
2692 (setq x (reverse x)) ;; start with x_k
2693 (prog (res ;; result
2694 r ;; insertion marker in res
2695 acc ;; acc accumulates previous x_i
2696 r1 ;; r1 iterates in each loop over acc
2697 (e 0) (ei 0) ) (declare (fixnum e ei))
2699 (setq ci (car x) ei (cadr x) ;; x_i = ci*v^ei
2700 *2ci (gf-cplus-b ci ci) ;; 2*ci (2*ci # 0 when *gf-char* # 2)
2701 res (cons (+ ei ei) (cons (gf-ctimes ci ci) res)) ;; res += x_i^2 (ci^2 # 0, no zero divisors)
2702 r (cdr res) ;; place insertion marker behind x_i^2
2703 r1 acc )
2705 (when (or (null r1) (= 0 *2ci)) ;; in ef *2ci might be 0 !
2706 (when (null (setq x (cddr x))) (return (gf-nred res red)))
2707 (setq acc (cons ei (cons ci acc))) ;; cons previous x_i to acc ..
2708 (go a1) ) ;; .. and start next loop
2710 (setq e (+ ei (the fixnum (car r1)))
2711 c (gf-ctimes *2ci (cadr r1))
2712 r1 (cddr r1) )
2714 (while (< e (the fixnum (cadr r)))
2715 (setq r (cddr r)) )
2716 (cond
2717 ((> e (the fixnum (cadr r)))
2718 (rplacd r (cons e (cons c (cdr r))))
2719 (setq r (cddr r)) )
2721 (setq c (gf-cplus-b c (caddr r)))
2722 (if (= 0 c)
2723 (rplacd r (cdddr r))
2724 (rplaca (setq r (cddr r)) c) ) ))
2725 (go a) ))) ))
2727 ;; x^n mod y
2729 (defun gf-pow (x n red) ;; assume 0 <= n
2730 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2731 (declare (integer n))
2732 (cond
2733 ((= 0 n) (list 0 1))
2734 ((null x) nil)
2735 (t (do (res)(())
2736 (when (oddp n)
2737 (setq res (if res (gf-times x res red) x))
2738 (when (= 1 n)
2739 (return-from gf-pow res) ))
2740 (setq n (ash n -1)
2741 x (gf-sq x red)) ))))
2743 ;; in a field use precomputed *gf-x^p-powers* resp. *ef-x^q-powers*
2745 (defun gf-pow$ (x n red)
2746 (if *ef-arith?*
2747 (if *ef-irred?*
2748 (*f-pow$ x n red *gf-card* *ef-card* *ef-x^q-powers*)
2749 (gf-pow x n red) )
2750 (if *gf-irred?*
2751 (*f-pow$ x n red *gf-char* *gf-card* *gf-x^p-powers*)
2752 (gf-pow x n red) )))
2754 (defun *f-pow$ (x n red p card x^p-powers)
2755 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2756 (declare (integer n p card))
2757 (cond
2758 ((= 0 n) (list 0 1))
2759 ((null x) nil)
2760 ((>= n card) (gf-pow x n red))
2762 (let ((prod (list 0 1))
2763 (j 0) n-base-p y )
2764 (do (quo r) ((= 0 n))
2765 (multiple-value-setq (quo r) (truncate n p))
2766 (push r n-base-p)
2767 (setq n quo) )
2768 (dolist (ni (nreverse n-base-p))
2769 (setq y (gf-compose (svref x^p-powers j) x red)
2770 y (gf-pow y ni red)
2771 prod (gf-times prod y red)
2772 j (1+ j) ))
2773 prod ))))
2775 ;; remainder:
2776 ;; x - quotient(x, y) * y
2778 (defun gf-rem (x y)
2779 (when (null y)
2780 (gf-merror (intl:gettext "~m arithmetic: Quotient by zero") (if *ef-arith?* "ef" "gf")) )
2781 (if (null x) x
2782 (gf-nrem (copy-list x) y) ))
2784 (defun gf-nrem (x y) ;; modifies x
2785 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2786 (when (null y)
2787 (gf-merror (intl:gettext "~m arithmetic: Quotient by zero") (if *ef-arith?* "ef" "gf")) )
2788 (if (null x) x
2789 (maybe-char-is-fixnum-let ((c 0) (lcx 0) (lcy-inv (gf-cinv (cadr y))))
2790 (let ((e 0) (ley (car y)))
2791 (declare (fixnum e ley))
2792 (setq lcy-inv (gf-cminus-b lcy-inv))
2793 (do ((y (cddr y)))
2794 ((null x) x)
2795 (setq e (- (the fixnum (car x)) ley))
2796 (when (< e 0) (return x))
2797 (setq lcx (cadr x)
2798 c (gf-ctimes lcx lcy-inv)
2799 x (gf-nxyecplus (cddr x) y e c)) )))))
2801 ;; reduce x by red
2803 ;; assume lc(red) = 1, reduction poly is monic
2805 (defun gf-nred (x red) ;; modifies x
2806 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2807 (if (or (null x) (null red)) x
2808 (let ((e 0) (le-red (car red)))
2809 (declare (fixnum e le-red))
2810 (setq red (cddr red))
2811 (do () ((null x) x)
2812 (setq e (- (the fixnum (car x)) le-red))
2813 (when (< e 0) (return x))
2814 (setq x (gf-nxyecplus (cddr x) red e (gf-cminus-b (cadr x)))) ))))
2816 ;; (monic) gcd
2818 (defun gf-gcd (x y)
2819 #+ (or ccl ecl gcl sbcl) (declare (optimize (speed 3) (safety 0)))
2820 (cond
2821 ((null x) y)
2822 ((null y) x)
2823 (t (let ((r nil))
2824 (do ()((null y)
2825 (if (eql 0 (car x)) (list 0 1)
2826 (gf-xctimes x (gf-cinv (cadr x))) ))
2827 (setq r (gf-rem x y))
2828 (psetf x y y r) )))))
2830 ;; (monic) extended gcd
2832 (defun gf-gcdex (x y red)
2833 #+ (or ccl ecl gcl sbcl) (declare (optimize (speed 3) (safety 0)))
2834 (let ((x1 (list 0 1)) x2 y1 (y2 (list 0 1)) q r)
2835 (do ()((null y)
2836 (let ((inv (gf-cinv (cadr x))))
2837 (mapcar #'(lambda (a) (gf-xctimes a inv)) (list x1 x2 x)) ))
2838 (multiple-value-setq (q r) (gf-divide x y))
2839 (psetf x y y r)
2840 (psetf
2841 y1 (gf-nplus (gf-nminus (gf-times q y1 red)) x1)
2842 x1 y1 )
2843 (psetf
2844 y2 (gf-nplus (gf-nminus (gf-times q y2 red)) x2)
2845 x2 y2 ) )))
2847 ;; inversion: y^-1
2849 ;; in case the inverse does not exist it returns nil
2850 ;; (might happen when reduction poly isn't irreducible)
2852 (defun gf-inv (y red)
2853 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2854 (when (null y)
2855 (gf-merror (intl:gettext "~m arithmetic: Quotient by zero") (if *ef-arith?* "ef" "gf")) )
2856 (let ((y1 (list 0 1)) (x red) x1 q r)
2857 (setq y (copy-list y))
2858 (do ()((null y)
2859 (when (= 0 (car x)) ;; gcd = 1 (const)
2860 (gf-nxctimes x1 (gf-cinv (cadr x))) ))
2861 (multiple-value-setq (q r) (gf-divide x y))
2862 (psetf x y y r)
2863 (psetf
2864 x1 y1
2865 y1 (gf-nplus (gf-nminus (gf-times q y1 red)) x1) )) ))
2867 ;; quotient and remainder
2869 (defun gf-divide (x y)
2870 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2871 (cond
2872 ((null y)
2873 (gf-merror (intl:gettext "~m arithmetic: Quotient by zero") (if *ef-arith?* "ef" "gf")) )
2874 ((null x) (values nil nil))
2876 (maybe-char-is-fixnum-let ((c 0) (lcx 0) (lcyi (gf-cinv (cadr y))))
2877 (let ((e 0) (ley (car y)))
2878 (declare (fixnum e ley))
2879 (setq x (copy-list x))
2880 (do (q (y (cddr y)))
2881 ((null x) (values (nreverse q) x))
2882 (setq e (- (the fixnum (car x)) ley))
2883 (when (< e 0)
2884 (return (values (nreverse q) x)) )
2885 (setq lcx (cadr x)
2886 x (cddr x)
2887 c (gf-ctimes lcx lcyi) )
2888 (unless (null y) (setq x (gf-nxyecplus x y e (gf-cminus-b c))))
2889 (setq q (cons c (cons e q))) ))))))
2891 ;; -----------------------------------------------------------------------------
2894 ;; polynomial/number/list - conversions ----------------------------------------
2897 ;; poly 2 number:
2899 (defmfun $ef_p2n (p)
2900 (gf-data? "ef_p2n")
2901 (let ((*ef-arith?* t)) (gf-x2n (gf-p2x p))) )
2903 (defmfun $gf_p2n (p &optional gf-char)
2904 (let ((*ef-arith?*))
2905 (cond
2906 (gf-char
2907 (let ((*gf-char* gf-char)) (gf-x2n (gf-p2x p))) )
2908 (*gf-char?*
2909 (gf-x2n (gf-p2x p)) )
2911 (gf-merror (intl:gettext "`gf_p2n': missing modulus.")) ))))
2913 (defun gf-x2n (x)
2914 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2915 (if (null x) 0
2916 (maybe-char-is-fixnum-let ((m *gf-char*))
2917 (when *ef-arith?* (setq m *gf-card*))
2918 (do ((n 0))(())
2919 (incf n (cadr x))
2920 (if (null (cddr x))
2921 (return (* n (expt m (the fixnum (car x)))))
2922 (setq n (* n (expt m (- (the fixnum (car x)) (the fixnum (caddr x)))))) )
2923 (setq x (cddr x)) ))))
2925 ;; number 2 poly:
2927 (defun gf-n2p-errchk (fun n)
2928 (unless (integerp n)
2929 (gf-merror (intl:gettext "`~m': Not an integer: ~m") fun n) ))
2931 (defmfun $gf_n2p (n &optional gf-char)
2932 (gf-n2p-errchk "gf_n2p" n)
2933 (let ((*ef-arith?*))
2934 (cond
2935 (gf-char
2936 (gf-set-rat-header)
2937 (let ((*gf-char* gf-char)) (gf-x2p (gf-n2x n))) )
2938 (*gf-char?*
2939 (gf-x2p (gf-n2x n)) )
2941 (gf-merror (intl:gettext "`gf_n2p': missing modulus.")) ))))
2943 (defmfun $ef_n2p (n)
2944 (gf-data? "ef_n2p")
2945 (gf-n2p-errchk "ef_n2p" n)
2946 (let ((*ef-arith?* t)) (gf-x2p (gf-n2x n))) )
2948 (defun gf-n2x (n)
2949 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
2950 (declare (integer n))
2951 (maybe-char-is-fixnum-let ((r 0) (m *gf-char*))
2952 (let ((e 0)) (declare (fixnum e))
2953 (when *ef-arith?* (setq m *gf-card*))
2954 (do (x)
2955 ((= 0 n) x)
2956 (multiple-value-setq (n r) (truncate n m))
2957 (unless (= 0 r)
2958 (setq x (cons e (cons r x))) )
2959 (incf e) ))))
2961 ;; poly 2 list:
2963 (defmfun $ef_p2l (p &optional (len 0))
2964 (declare (fixnum len))
2965 (let ((*ef-arith?* t))
2966 (cons '(mlist simp) (gf-x2l (gf-p2x-raw p) len)) )) ;; more flexibility ...
2968 (defmfun $gf_p2l (p &optional (len 0)) ;; len = 0 : no padding
2969 (declare (fixnum len))
2970 (let ((*ef-arith?*))
2971 (cons '(mlist simp) (gf-x2l (gf-p2x-raw p) len)) )) ;; ... by omitting mod reduction
2973 (defun gf-x2l (x len)
2974 #+ (or ccl ecl gcl sbcl) (declare (optimize (speed 3) (safety 0)))
2975 (declare (fixnum len))
2976 (do* ((e (if x (the fixnum (car x)) 0))
2977 (k (max e (1- len)) (1- k))
2978 l )
2979 ((< k 0) (nreverse l))
2980 (declare (fixnum e k))
2981 (cond
2982 ((or (null x) (> k e))
2983 (push 0 l) )
2984 ((= k e)
2985 (push (cadr x) l)
2986 (setq x (cddr x))
2987 (unless (null x) (setq e (the fixnum (car x)))) ))))
2989 ;; list 2 poly:
2991 (defmfun $ef_l2p (l)
2992 (gf-l2p-errchk l "ef_l2p")
2993 (let ((*ef-arith?* t)) (gf-x2p (gf-l2x (cdr l)))) )
2995 (defmfun $gf_l2p (l)
2996 (gf-l2p-errchk l "gf_l2p")
2997 (let ((*ef-arith?*)) (gf-x2p (gf-l2x (cdr l)))) )
2999 (defun gf-l2p-errchk (l fun)
3000 (unless ($listp l)
3001 (gf-merror (intl:gettext "`~m': Argument must be a list of integers.") fun) ))
3003 (defun gf-l2x (l)
3004 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3005 (setq l (reverse l))
3006 (maybe-char-is-fixnum-let ((c 0))
3007 (let ((e 0)) (declare (fixnum e))
3008 (do (x)
3009 ((null l) x)
3010 (unless (= 0 (setq c (car l)))
3011 (setq x (cons e (cons c x))) )
3012 (setq l (cdr l))
3013 (incf e) ))))
3015 ;; list 2 number:
3017 (defmfun $gf_l2n (l)
3018 (gf-char? "gf_l2n")
3019 (gf-l2p-errchk l "gf_l2n")
3020 (let ((*ef-arith?*)) (gf-l2n (cdr l))) )
3022 (defmfun $ef_l2n (l)
3023 (gf-data? "ef_l2n")
3024 (gf-l2p-errchk l "ef_l2n")
3025 (let ((*ef-arith?* t)) (gf-l2n (cdr l))) )
3027 (defun gf-l2n (l)
3028 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3029 (maybe-char-is-fixnum-let ((m *gf-char*) (c1 (car l)) (c 0))
3030 (when *ef-arith?* (setq m *gf-card*))
3031 (setq l (reverse (cdr l)))
3032 (do ((n 0) (b 1))
3033 ((null l) (+ (* c1 b) n))
3034 (declare (integer n b))
3035 (unless (= 0 (setq c (car l))) (incf n (* c b)))
3036 (setq b (* b m) l (cdr l)) )))
3038 ;; number 2 list:
3040 (defmfun $gf_n2l (n &optional (len 0)) ;; in case of len = 0 the list isn't padded or truncated
3041 (declare (integer n) (fixnum len))
3042 (gf-char? "gf_n2l")
3043 (gf-n2p-errchk "gf_n2l" n)
3044 (cons '(mlist simp)
3045 (let ((*ef-arith?*))
3046 (if (= 0 len) (gf-n2l n) (gf-n2l-twoargs n len)) )))
3048 (defmfun $ef_n2l (n &optional (len 0)) ;; in case of len = 0 the list isn't padded or truncated
3049 (declare (integer n) (fixnum len))
3050 (gf-data? "ef_n2l")
3051 (gf-n2p-errchk "ef_n2l" n)
3052 (cons '(mlist simp)
3053 (let ((*ef-arith?* t))
3054 (if (= 0 len) (gf-n2l n) (gf-n2l-twoargs n len)) )))
3056 (defun gf-n2l (n) ;; this version is frequently called by gf-precomp, keep it simple
3057 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3058 (declare (integer n))
3059 (maybe-char-is-fixnum-let ((m *gf-char*) (r 0))
3060 (when *ef-arith?* (setq m *gf-card*))
3061 (do (l) ((= 0 n) l)
3062 (multiple-value-setq (n r) (truncate n m))
3063 (setq l (cons r l)) )))
3065 (defun gf-n2l-twoargs (n len)
3066 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3067 (declare (integer n) (fixnum len))
3068 (maybe-char-is-fixnum-let ((m *gf-char*) (r 0))
3069 (when *ef-arith?* (setq m *gf-card*))
3070 (do (l) ((= 0 len) l)
3071 (multiple-value-setq (n r) (truncate n m))
3072 (setq l (cons r l))
3073 (decf len) )))
3075 ;; -----------------------------------------------------------------------------
3078 ;; irreducibility (Ben-Or algorithm) -------------------------------------------
3081 (defmfun $gf_irreducible_p (a &optional p)
3082 (cond
3083 (p (unless (and (integerp p) (primep p))
3084 (gf-merror (intl:gettext
3085 "`gf_irreducible_p': Second argument must be a prime number." )) ))
3086 (t (gf-char? "gf_irreducible_p")
3087 (setq p *gf-char*) ))
3088 (let* ((*ef-arith?*)
3089 (*gf-char* p)
3090 (x (gf-p2x a)) n) ;; gf-p2x depends on *gf-char*
3091 (cond
3092 ((null x) nil)
3093 ((= 0 (setq n (car x))) nil)
3094 ((= 1 n) t)
3095 (t (gf-irr-p x p (car x))) )))
3097 (defmfun $ef_irreducible_p (a)
3098 (ef-gf-field? "ef_irreducible_p")
3099 (let ((*ef-arith?* t))
3100 (setq a (gf-p2x a))
3101 (gf-irr-p a *gf-card* (car a)) ))
3103 ;; is y irreducible of degree n over Fq[x] ?
3105 ;; q,n > 1 !
3106 (defun gf-irr-p (y q n) ;; gf-irr-p is independent from any settings
3107 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3108 (declare (integer q) (fixnum n))
3109 (let* ((*gf-char* (car (cfactorw q)))
3110 (x (list 1 1))
3111 (mx (gf-minus x)) ;; gf-minus needs *gf-char*
3112 (lc (cadr y)) )
3113 (unless (= 1 lc)
3114 (setq y (gf-xctimes y (gf-cinv lc))) ) ;; monicize y
3115 (do ((i 1 (1+ i)) (xq x) (n2 (ash n -1)))
3116 ((> i n2) t)
3117 (declare (fixnum i n2))
3118 (setq xq (gf-pow xq q y))
3119 (unless (= 0 (car (gf-gcd y (gf-plus xq mx))))
3120 (return) ) )))
3122 ;; find an irreducible element
3124 ;; gf_irreducible is independent from any settings
3126 (defmfun $gf_irreducible (p n) ;; p,n : desired characteristic and degree
3127 (unless (and (integerp p) (primep p) (integerp n))
3128 (gf-merror (intl:gettext "`gf_irreducible' needs a prime number and an integer.")) )
3129 (gf-set-rat-header)
3130 (let* ((*gf-char* p)
3131 (*ef-arith?*)
3132 (irr (gf-irr p n)) )
3133 (when irr (gf-x2p irr)) ))
3135 (defmfun $ef_irreducible (n) ;; n : desired degree
3136 (ef-gf-field? "ef_irreducible")
3137 (unless (integerp n)
3138 (gf-merror (intl:gettext "`ef_irreducible' needs an integer.")) )
3139 (let* ((*ef-arith?* t)
3140 (irr (ef-irr n)) )
3141 (when irr (gf-x2p irr)) ))
3144 (defun gf-irr (p n)
3145 (*f-irr p n) )
3147 (defun ef-irr (n)
3148 (*f-irr *gf-card* n) )
3150 (defun *f-irr (q n)
3151 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3152 (when (= 1 n)
3153 (return-from *f-irr (list 1 1)) )
3154 (let* ((inc (min $gf_coeff_limit q))
3155 (i-lim (expt inc n))
3157 (do ((i 1 (1+ i)))
3158 ((>= i i-lim)
3159 (gf-merror (intl:gettext "No irreducible polynomial found.~%~\
3160 `gf_coeff_limit' might be too small.~%" )))
3161 (setq x (let ((*gf-char* inc)) (gf-n2x i)))
3162 (when (= 0 (car (last x 2)))
3163 (setq x (cons n (cons 1 x)))
3164 (when (gf-irr-p x q n) (return-from *f-irr x)) ))))
3166 ;; -----------------------------------------------------------------------------
3169 ;; Primitive elements ----------------------------------------------------------
3172 ;; Tests if an element is primitive in the field
3174 (defmfun $gf_primitive_p (a)
3175 (gf-data? "gf_primitive_p") ;; --> precomputations are performed
3176 (let* ((*ef-arith?*)
3177 (x (gf-p2x a))
3178 (n (gf-x2n x)) )
3179 (cond
3180 ((or (= 0 n) (>= n *gf-card*)) nil)
3181 ((= 1 *gf-exp*)
3182 (zn-primroot-p n *gf-char* *gf-ord* (mapcar #'car *gf-fs-ord*)) )
3184 (gf-prim-p x) ))))
3186 (defmfun $ef_primitive_p (a)
3187 (ef-data? "ef_primitive_p") ;; --> precomputations are performed
3188 (let ((*ef-arith?* t))
3189 (setq a (gf-p2x a))
3190 (cond
3191 ((null a) nil)
3192 ((>= (car a) *ef-exp*) nil)
3193 ((= (car a) 0)
3194 (if (= 1 *ef-exp*)
3195 (let ((*ef-arith?*)) (gf-prim-p (gf-n2x (cadr a))))
3196 nil ))
3197 (t (ef-prim-p a)) )))
3200 ;; Testing primitivity in (Fq^n)*:
3202 ;; We check f(x)^ei # 1 (ei = ord/pi) for all prime factors pi of ord.
3204 ;; With ei = sum(aij*q^j, j,0,m) in base q and using f(x)^q = f(x^q) we get
3206 ;; f(x)^ei = f(x)^sum(aij*q^j, j,0,m) = prod(f(x^q^j)^aij, j,0,m).
3209 ;; Special case: red is irreducible, f(x) = x+c and pi|ord and pi|q-1.
3211 ;; Then ei = (q^n-1)/(q-1) * (q-1)/pi = sum(q^j, j,0,n-1) * (q-1)/pi.
3213 ;; With ai = (q-1)/pi and using red(z) = prod(z - x^q^j, j,0,n-1) we get
3215 ;; f(x)^ei = f(x)^sum(ai*q^j, j,0,n-1) = (prod(f(x)^q^j, j,0,n-1))^ai
3217 ;; = (prod(x^q^j + c, j,0,n-1))^ai = ((-1)^n * prod(-c - x^q^j, j,0,n-1))^ai
3219 ;; = ((-1)^n * red(-c))^ai
3222 (defun gf-prim-p (x)
3223 (cond
3224 (*gf-irred?*
3225 (*f-prim-p-2 x *gf-char* *gf-red* *gf-fsx* *gf-fsx-base-p* *gf-x^p-powers*) )
3226 ((gf-unit-p x *gf-red*)
3227 (*f-prim-p-1 x *gf-red* *gf-ord* *gf-fs-ord*) )
3228 (t nil) ))
3230 (defun ef-prim-p (x)
3231 (cond
3232 (*ef-irred?*
3233 (*f-prim-p-2 x *gf-card* *ef-red* *ef-fsx* *ef-fsx-base-q* *ef-x^q-powers*) )
3234 ((gf-unit-p x *ef-red*)
3235 (*f-prim-p-1 x *ef-red* *ef-ord* *ef-fs-ord*) )
3236 (t nil) ))
3238 ;; *f-prim-p-1
3240 (defun *f-prim-p-1 (x red ord fs-ord)
3241 (dolist (pe fs-ord t)
3242 (when (equal '(0 1) (gf-pow x (truncate ord (car pe)) red)) (return)) ))
3244 ;; *f-prim-p-2 uses precomputations and exponentiation by composition
3246 (defun *f-prim-p-2 (x q red fs fs-base-q x^q-powers)
3247 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3248 (unless (or (= 2 *gf-char*) (= -1 (gf-jacobi x red q))) ;; red is assumed to be irreducible
3249 (return-from *f-prim-p-2) )
3250 (let ((exponent (car red))
3251 (x+c? (and (= (car x) 1) (= (cadr x) 1)))
3252 y prod -c z )
3253 (do ((i 0 (1+ i)) (j 0 0) (lf (array-dimension fs 0)))
3254 ((= i lf) t)
3255 (declare (fixnum i j lf))
3256 (cond
3257 ((and x+c? (cadr (svref fs i))) ;; linear and pi|ord and pi|p-1
3258 (setq -c (if (= 2 (length x)) 0 (gf-cminus-b (car (last x))))
3259 z (list 0 (gf-at red -c)) )
3260 (when (oddp exponent) (setq z (gf-minus z))) ;; (-1)^n * red(-c)
3261 (setq z (gf-pow z (caddr (svref fs i)) red)) ;; ((-1)^n * red(-c))^ai
3262 (when (or (null z) (equal z '(0 1)))
3263 (return nil) ))
3265 (setq prod (list 0 1))
3266 (dolist (aij (svref fs-base-q i))
3267 (setq y (gf-compose (svref x^q-powers j) x red) ;; f(x^q^j)
3268 y (gf-pow y aij red) ;; f(x^q^j)^aij
3269 prod (gf-times prod y red)
3270 j (1+ j) ))
3271 (when (or (null prod) (equal prod '(0 1))) ;; prod(f(x^q^j)^aij, j,0,m)
3272 (return nil) )) ))))
3275 ;; generalized Jacobi-symbol (Bach-Shallit, Theorem 6.7.1)
3277 (defmfun $gf_jacobi (a b)
3278 (gf-char? "gf_jacobi")
3279 (let* ((*ef-arith?*)
3280 (x (gf-p2x a))
3281 (y (gf-p2x b)) )
3282 (if (= 2 *gf-char*)
3283 (if (null (gf-rem x y)) 0 1)
3284 (gf-jacobi x y *gf-char*) )))
3286 (defmfun $ef_jacobi (a b)
3287 (ef-gf-field? "ef_jacobi")
3288 (let* ((*ef-arith?* t)
3289 (x (gf-p2x a))
3290 (y (gf-p2x b)) )
3291 (if (= 2 (car (cfactorw *gf-card*)))
3292 (if (null (gf-rem x y)) 0 1)
3293 (gf-jacobi x y *gf-card*) )))
3295 (defun gf-jacobi (u v q)
3296 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3297 (if (null (setq u (gf-rem u v))) 0
3298 (let* ((c (cadr u))
3299 (s (if (evenp (car v)) 1 (gf-cjacobi c))) )
3300 (cond
3301 ((= 0 (car u)) s)
3303 (setq u (gf-xctimes u (gf-cinv c)))
3304 (when (every #'oddp (list (ash (1- q) -1) (car u) (car v)))
3305 (setq s (neg s)) )
3306 (* s (gf-jacobi v u q)) )))))
3308 (defun gf-cjacobi (c)
3309 (if *ef-arith?*
3310 (let ((*ef-arith?*)) (gf-jacobi (gf-n2x c) *gf-red* *gf-char*))
3311 ($jacobi c *gf-char*) ))
3314 ;; modular composition (uses Horner and square and multiply)
3315 ;; y(x) mod red
3317 (defmfun $gf_compose (a b)
3318 (gf-red? "gf_compose")
3319 (let ((*ef-arith?*))
3320 (gf-x2p (gf-compose (gf-p2x a) (gf-p2x b) *gf-red*)) ))
3322 (defmfun $ef_compose (a b)
3323 (ef-red? "ef_compose")
3324 (let ((*ef-arith?* t))
3325 (gf-x2p (gf-compose (gf-p2x a) (gf-p2x b) *ef-red*)) ))
3327 (defun gf-compose (x y red)
3328 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3329 (cond
3330 ((or (null x) (null y)) nil)
3331 ((= 0 (car y)) y)
3332 ((= 0 (car x))
3333 (let ((n (gf-at y (cadr x))))
3334 (if (= 0 n) nil (list 0 n)) ))
3336 (do (res) (())
3337 (setq res (gf-nplus res (list 0 (cadr y))))
3338 (when (null (cddr y))
3339 (return (gf-times res (gf-pow x (car y) red) red)) )
3340 (setq res (gf-times res (gf-pow x (- (car y) (caddr y)) red) red)
3341 y (cddr y) ) ))))
3343 ;; a(n) with poly a and integer n
3345 (defun gf-at-errchk (n fun)
3346 (unless (integerp n)
3347 (gf-merror (intl:gettext "`~m': Second argument must be an integer.") fun) ))
3349 (defmfun $gf_at (a n) ;; integer n
3350 (gf-char? "gf_at")
3351 (gf-at-errchk n "gf_at")
3352 (let ((*ef-arith?*))
3353 (gf-at (gf-p2x a) n) ))
3355 (defmfun $ef_at (a n) ;; poly a, integer n
3356 (ef-gf-field? "ef_at")
3357 (gf-at-errchk n "ef_at")
3358 (let ((*ef-arith?* t))
3359 (gf-at (gf-p2x a) n) ))
3361 (defun gf-at (x n) ;; Horner and square and multiply
3362 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3363 (if (or (null x) (integerp x)) x
3364 (maybe-char-is-fixnum-let ((n n))
3365 (do ((i 0) i1) (())
3366 (setq i (gf-cplus-b i (cadr x)))
3367 (when (null (cddr x))
3368 (setq i1 (gf-cpow n (the fixnum (car x))))
3369 (return (gf-ctimes i i1)) )
3370 (setq i1 (gf-cpow n (- (the fixnum (car x)) (the fixnum (caddr x))))
3371 i (gf-ctimes i i1)
3372 x (cddr x) )))))
3374 ;; find a primitive element:
3376 (defmfun $gf_primitive ()
3377 (gf-data? "gf_primitive")
3378 (let ((*ef-arith?*))
3379 (cond
3380 ((null *gf-prim*) nil)
3381 ((equal *gf-prim* '$unknown)
3382 (setq *gf-prim* (gf-prim))
3383 (unless (null *gf-prim*) (gf-x2p *gf-prim*)) )
3384 (t (gf-x2p *gf-prim*)) )))
3386 (defmfun $ef_primitive ()
3387 (ef-data? "ef_primitive")
3388 (let ((*ef-arith?* t))
3389 (cond
3390 ((null *ef-prim*) nil)
3391 ((equal *ef-prim* '$unknown)
3392 (cond
3393 ((= 1 *ef-exp*)
3394 (setq *ef-prim* (let ((*ef-arith?*)) (gf-x2n *gf-prim*))) )
3396 (setq *ef-prim* (ef-prim))
3397 (unless (null *ef-prim*) (gf-x2p *ef-prim*)) )))
3398 (t (gf-x2p *ef-prim*)) )))
3401 (defun gf-prim ()
3402 (*f-prim *gf-char* *gf-exp* #'gf-prim-p) )
3404 (defun ef-prim ()
3405 (*f-prim *gf-card* *ef-exp* #'ef-prim-p) )
3407 (defun *f-prim (inc e prim-p-fn)
3408 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3409 (setq inc (min $gf_coeff_limit inc))
3410 (do ((n inc (1+ n))
3411 (n-lim (expt inc e))
3413 ((>= n n-lim)
3414 (when (= $gf_coeff_limit inc) '$unknown) )
3415 (setq x (let ((*gf-char* inc)) (gf-n2x n)))
3416 (cond
3417 ((= 2 (cadr x))
3418 (setq n (1- (* (ash n -1) inc))) ) ;; go to next monic poly
3419 ((funcall prim-p-fn x)
3420 (return x) ) )))
3423 ;; precomputation for *f-prim-p:
3425 (defun gf-precomp ()
3426 (*f-precomp (1- *gf-char*) *gf-ord* *gf-fs-ord*) )
3428 (defun ef-precomp ()
3429 (*f-precomp (1- *gf-card*) *ef-ord* *ef-fs-ord*) )
3431 (defun *f-precomp (q-1 ord fs-ord)
3432 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3433 (let (fs-q-1
3434 fs-list
3435 ($intfaclim) )
3436 (setq fs-q-1
3437 (sort (get-factor-list q-1) #'< :key #'car) ) ;; .. [pi, ei] ..
3438 (dolist (fj fs-q-1)
3439 (setq fs-ord (remove-if #'(lambda (sj) (= (car fj) (car sj))) fs-ord :count 1)) )
3440 (setq fs-q-1
3441 (mapcar #'(lambda (pe) (list (car pe) t (truncate q-1 (car pe)))) fs-q-1) );; .. [pi, true, (p-1)/pi] ..
3442 (setq fs-ord
3443 (mapcar #'(lambda (pe) (list (car pe) nil)) fs-ord) ) ;; .. [pi, false] ..
3444 (setq fs-list
3445 (merge 'list fs-q-1 fs-ord #'(lambda (a b) (< (car a) (car b)))) )
3446 (cond
3447 (*ef-arith?*
3448 (setq *ef-fsx* (apply #'vector fs-list))
3449 (setq *ef-fsx-base-q*
3450 (apply #'vector
3451 (mapcar #'(lambda (pe) (nreverse (gf-n2l (truncate ord (car pe))))) ;; qi = ord/pi = sum(aij*p^j, j,0,m)
3452 fs-list) ))
3453 (setq *ef-x^q-powers* (gf-x^p-powers *gf-card* *ef-exp* *ef-red*)) ) ;; x^p^j
3455 (setq *gf-fsx* (apply #'vector fs-list))
3456 (setq *gf-fsx-base-p*
3457 (apply #'vector
3458 (mapcar #'(lambda (pe) (nreverse (gf-n2l (truncate ord (car pe))))) ;; qi = ord/pi = sum(aij*p^j, j,0,m)
3459 fs-list) ))
3460 (setq *gf-x^p-powers* (gf-x^p-powers *gf-char* *gf-exp* *gf-red*)) )))) ;; x^p^j
3462 ;; returns an array of polynomials x^p^j, j = 0, 1, .. , (n-1), where n = *gf-exp*
3464 (defun gf-x^p-powers (q n red)
3465 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3466 (declare (integer q) (fixnum n))
3467 (let ((a (make-array n :element-type 'list :initial-element nil)) )
3468 (setf (svref a 0) (list 1 1)) ;; x
3469 (do ((j 1 (1+ j)))
3470 ((= j n) a)
3471 (declare (fixnum j))
3472 (setf (svref a j) (gf-pow (svref a (1- j)) q red)) )))
3474 ;; -----------------------------------------------------------------------------
3477 ;; Primitive polynomials -------------------------------------------------------
3480 ;; test if a is a primitive polynomial over Fq
3482 (defmfun $gf_primitive_poly_p (a &optional p)
3483 (cond
3484 (p (unless (and (integerp p) (primep p))
3485 (gf-merror (intl:gettext "`gf_primitive_poly_p': ~m is not a prime number.") p) ))
3486 (t (gf-char? "gf_primitive_poly_p")
3487 (setq p *gf-char*) ))
3488 (let* ((*ef-arith?*)
3489 (*gf-char* p)
3490 (y (gf-p2x a)) ;; gf-p2x depends on *gf-char*
3491 (n (car y)) )
3492 (gf-primpoly-p y p n) ))
3494 (defmfun $ef_primitive_poly_p (y)
3495 (ef-gf-field? "ef_primitive_poly_p")
3496 (let ((*ef-arith?* t))
3497 (setq y (gf-p2x y))
3498 (gf-primpoly-p y *gf-card* (car y)) ))
3501 ;; based on
3502 ;; TOM HANSEN AND GARY L. MULLEN
3503 ;; PRIMITIVE POLYNOMIALS OVER FINITE FIELDS
3504 ;; (gf-primpoly-p performs a full irreducibility check
3505 ;; and therefore doesn't check whether x^((q^n-1)/(q-1)) = (-1)^n * y(0) mod y)
3507 (defun gf-primpoly-p (y q n)
3508 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3509 (declare (fixnum n))
3510 (unless (= 1 (cadr y)) ;; monic poly assumed
3511 (return-from gf-primpoly-p) )
3512 (prog* ((fs-q (cfactorw q))
3513 (*gf-char* (car fs-q))
3514 (*gf-exp* (if *ef-arith?* (cadr fs-q) n))
3515 (q-1 (1- q)) fs-q-1
3516 (const (last y 2))
3517 ($intfaclim) )
3518 ;; the constant part ...
3519 (unless (= 0 (car const)) (return nil))
3520 (setq const (cadr const))
3521 (when (oddp n) (setq const (gf-cminus-b const))) ;; (-1)^n*const
3522 ;; ... must be primitive in Fq:
3523 (unless (if (and *ef-arith?* (> *gf-exp* 1))
3524 (let ((*ef-arith?*)) (gf-prim-p (gf-n2x const)))
3525 (progn
3526 (setq fs-q-1 (sort (mapcar #'car (get-factor-list q-1)) #'<))
3527 (zn-primroot-p const q q-1 fs-q-1) ))
3528 (return nil) )
3529 ;; the linear case:
3530 (when (= n 1) (return t))
3531 ;; y must be irreducible:
3532 (unless (gf-irr-p y q n) (return nil))
3533 ;; check for all prime factors fi of r = (q^n-1)/(q-1) which do not divide q-1
3534 ;; that x^(r/fi) mod y is not an integer:
3535 (let (x^q-powers r fs-r fs-r-base-q)
3536 ;; pre-computation:
3537 (setq x^q-powers (gf-x^p-powers q n y)
3538 r (truncate (1- (expt q n)) q-1)
3539 fs-r (sort (mapcar #'car (get-factor-list r)) #'<) )
3540 (unless fs-q-1
3541 (setq fs-q-1 (sort (mapcar #'car (get-factor-list q-1)) #'<)) )
3542 (dolist (fj fs-q-1)
3543 (setq fs-r (delete-if #'(lambda (sj) (= fj sj)) fs-r :count 1)) )
3544 (setq fs-r-base-q
3545 (let ((*gf-char* q))
3546 (apply #'vector
3547 (mapcar #'(lambda (f) (nreverse (gf-n2l (truncate r f)))) fs-r ) )))
3548 ;; check:
3549 (return (gf-primpoly-p-exit y fs-r-base-q x^q-powers)) )))
3551 ;; uses exponentiation by pre-computation
3552 (defun gf-primpoly-p-exit (y fs-r-base-q x^q-powers)
3553 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3554 (do ((i 0 (1+ i)) (j 0 0) (dim (array-dimension fs-r-base-q 0)) z zz)
3555 ((= i dim) t)
3556 (declare (fixnum i j dim))
3557 (setq zz (list 0 1))
3558 (dolist (aij (svref fs-r-base-q i)) ;; fi = sum(aij*q^j, j,0,n-1)
3559 (setq z (gf-pow (svref x^q-powers j) aij y)
3560 zz (gf-times zz z y)
3561 j (1+ j) ))
3562 (when (= 0 (car zz)) (return nil)) ))
3565 ;; find a primitive polynomial
3567 (defmfun $gf_primitive_poly (p n)
3568 (unless (and (integerp p) (primep p) (integerp n))
3569 (gf-merror (intl:gettext "`gf_primitive_poly' needs a prime number and an integer.")) )
3570 (gf-set-rat-header)
3571 (let ((*ef-arith?*) (*gf-char* p)) ;; gf-x2p needs *gf-char*
3572 (gf-x2p (gf-primpoly p n)) ))
3574 (defmfun $ef_primitive_poly (n)
3575 (ef-gf-field? "ef_primitive_poly")
3576 (let ((*ef-arith?* t))
3577 (gf-x2p (gf-primpoly *gf-card* n)) ))
3580 (defun gf-primpoly (q n)
3581 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3582 (declare (fixnum n))
3583 (let* ((fs-q (cfactorw q))
3584 (*gf-char* (car fs-q))
3585 (*gf-exp* (if *ef-arith?* (cadr fs-q) n))
3586 (q-1 (1- q))
3587 ($intfaclim)
3588 (fs-q-1 (sort (mapcar #'car (get-factor-list q-1)) #'<))
3589 r fs-r fs-r-base-q )
3590 ;; the linear case:
3591 (when (= 1 n)
3592 (let ((prt (if (= q 2) 1 (zn-primroot q q-1 fs-q-1))))
3593 (return-from gf-primpoly
3594 (list 1 1 0 (gf-cminus-b prt)) )))
3595 ;; pre-computation part 1:
3596 (setq r (truncate (1- (expt q n)) q-1)
3597 fs-r (sort (mapcar #'car (get-factor-list r)) #'<) )
3598 (dolist (fj fs-q-1)
3599 (setq fs-r (delete-if #'(lambda (sj) (= fj sj)) fs-r :count 1)) )
3600 (setq fs-r-base-q
3601 (let ((*gf-char* q))
3602 (apply #'vector
3603 (mapcar #'(lambda (f) (nreverse (gf-n2l (truncate r f)))) fs-r ) )))
3604 ;; search:
3605 (let* ((inc (min $gf_coeff_limit q))
3606 (i-lim (expt inc n))
3608 (do ((i (1+ inc) (1+ i)))
3609 ((>= i i-lim)
3610 (gf-merror (intl:gettext "No primitive polynomial found.~%~\
3611 `gf_coeff_limit' might be too small.~%" )) )
3612 (setq x (let ((*gf-char* inc)) (gf-n2x i))
3613 x (cons n (cons 1 x)) )
3614 (when (gf-primpoly-p2 x *gf-char* *gf-exp* q n fs-q-1 fs-r-base-q)
3615 (return x) )))))
3617 (defun gf-primpoly-p2 (y p e q n fs-q-1 fs-r-base-q)
3618 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3619 (declare (fixnum e n))
3620 (when (= 1 (cadr y)) ;; monic poly assumed
3621 (prog* ((*gf-char* p) (*gf-exp* e) (q-1 (1- q))
3622 (const (last y 2)) )
3623 ;; the constant part ...
3624 (unless (= 0 (car const)) (return nil))
3625 (setq const (cadr const))
3626 (when (oddp n) (setq const (gf-cminus-b const))) ;; (-1)^n*const
3627 ;; ... must be primitive in Fq:
3628 (unless (if (and *ef-arith?* (> *gf-exp* 1))
3629 (let ((*ef-arith?*)) (gf-prim-p (gf-n2x const)))
3630 (zn-primroot-p const q q-1 fs-q-1) )
3631 (return nil) )
3632 ;; y must be irreducible:
3633 (unless (gf-irr-p y q n) (return nil))
3634 ;; y dependend pre-computation and final check:
3635 (return (gf-primpoly-p-exit y fs-r-base-q (gf-x^p-powers q n y))) )))
3637 ;; -----------------------------------------------------------------------------
3640 ;; random elements -------------------------------------------------------------
3643 ;; Produces a random element within the given environment
3645 (defmfun $gf_random (&optional p n)
3646 (let ((*ef-arith?* t))
3647 (cond
3648 (n (let ((*gf-char* p))
3649 (unless *gf-red?* (gf-set-rat-header))
3650 (gf-x2p (gf-random p n)) ))
3651 (t (gf-data? "gf_random")
3652 (gf-x2p (gf-random *gf-char* *gf-exp*)) ))))
3654 (defmfun $ef_random (&optional q n)
3655 (let ((*ef-arith?* t))
3656 (cond
3657 (n (let ((*gf-char* q)) (gf-x2p (gf-random q n))))
3658 (t (ef-data? "ef_random")
3659 (gf-x2p (gf-random *gf-card* *ef-exp*)) ))))
3661 (defun gf-random (q n)
3662 (do ((e 0 (1+ e)) c x)
3663 ((= e n) x)
3664 (setq c (random q))
3665 (when (/= 0 c)
3666 (setq x (cons e (cons c x))) )))
3668 ;; -----------------------------------------------------------------------------
3671 ;; factoring -------------------------------------------------------------------
3674 (defmfun $gf_factor (a &optional p) ;; set p to switch to another modulus
3675 (cond
3676 (p (unless (and (integerp p) (primep p))
3677 (gf-merror (intl:gettext "`gf_factor': Second argument must be a prime number.")) )
3678 (gf-set-rat-header) )
3679 (t (gf-char? "gf_factor")
3680 (setq p *gf-char*) ))
3681 (let* ((*gf-char* p)
3682 (modulus p) (a ($rat a))
3683 (*ef-arith?*) )
3684 (when (> (length (caddar a)) 1)
3685 (gf-merror (intl:gettext "`gf_factor': Polynomial must be univariate.")) )
3686 (setq a (cadr a))
3687 (cond
3688 ((integerp a) (mod a *gf-char*))
3689 (t (setq a
3690 (if $gf_cantor_zassenhaus
3691 (gf-factor (gf-mod (cdr a)) p)
3692 (gf-ns2pmod-factors (pfactor a)) ))
3693 (setq a (gf-disrep-factors a))
3694 (and (consp a) (consp (car a)) (equal (caar a) 'mtimes)
3695 (setq a (simplifya (cons '(mtimes) (cdr a)) nil)) )
3696 a ))))
3698 ;; adjust results from rat3d/pfactor to a positive modulus if $gf_balanced = false
3699 (defun gf-ns2pmod-factors (fs) ;; modifies fs
3700 (if $gf_balanced fs
3701 (maybe-char-is-fixnum-let ((m *gf-char*))
3702 (do ((r fs (cddr r)))
3703 ((null r) fs)
3704 (if (integerp (car r))
3705 (when (< (the integer (car r)) 0)
3706 (incf (car r) m) ) ;; only in the case *ef-arith?* = false
3707 (rplaca r (gf-ns2pmod-factor (cdar r) m)) )))))
3709 (defun gf-ns2pmod-factor (fac m)
3710 (do ((r (cdr fac) (cddr r))) (())
3711 (when (< (the integer (car r)) 0)
3712 (incf (car r) m) )
3713 (when (null (cdr r)) (return fac)) ))
3715 (defun gf-disrep-factors (fs)
3716 (cond
3717 ((integerp fs) (gf-cp2smod fs))
3719 (setq fs (nreverse fs))
3720 (do ((e 0) fac p)
3721 ((null fs) (cons '(mtimes simp factored) p))
3722 (declare (fixnum e))
3723 (setq e (the fixnum (car fs))
3724 fac (cadr fs)
3725 fs (cddr fs)
3726 p (cond
3727 ((integerp fac) (cons (gf-cp2smod fac) p))
3728 ((= 1 e) (cons (gf-disrep (gf-np2smod fac)) p))
3729 (t (cons `((mexpt simp) ,(gf-disrep (gf-np2smod fac)) ,e) p)) ))))))
3731 (defmfun $ef_factor (a)
3732 (ef-gf-field? "ef_factor")
3733 (let ((*ef-arith?* t))
3734 (setq a (let ((modulus)) ($rat a)))
3735 (when (> (length (caddar a)) 1)
3736 (gf-merror (intl:gettext "`ef_factor': Polynomial must be univariate.")) )
3737 (setq a (cadr a))
3738 (cond
3739 ((integerp a) (ef-cmod a))
3740 (t (setq a
3741 (gf-disrep-factors
3742 (gf-factor (gf-mod (cdr a)) *gf-card*) ))
3743 (and (consp a) (consp (car a)) (equal (caar a) 'mtimes)
3744 (setq a (simplifya (cons '(mtimes) (cdr a)) nil)) )
3745 a ))))
3747 (defun gf-factor (x q) ;; non-integer x
3748 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3749 (let ((lc (cadr x)) z)
3750 (unless (= 1 lc)
3751 (setq x (gf-xctimes x (gf-cinv lc))) ) ;; monicize x
3752 (if (gf-irr-p x q (car x))
3753 (setq z (list x 1))
3754 (let ((sqfr (gf-square-free x)) e y)
3755 (dolist (v sqfr)
3756 (setq e (car v)
3757 y (cadr v)
3758 y (gf-distinct-degree-factors y q) )
3759 (dolist (w y)
3760 (setq z (nconc (gf-equal-degree-factors w q e) z)) ))))
3761 (if (= 1 lc) z (cons lc (cons 1 z))) ))
3763 (defun gf-diff (x)
3764 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3765 (if (null x) nil
3766 (maybe-char-is-fixnum-let ((m *gf-char*))
3767 (do ((rx x (cddr rx)) res c)
3768 ((or (null rx) (= 0 (car rx))) (nreverse res))
3769 (setq c (gf-ctimes (mod (the fixnum (car rx)) m) (cadr rx)))
3770 (when (/= 0 c)
3771 (push (1- (car rx)) res)
3772 (push c res) )))) )
3774 ;; c -> c^p^(n-1)
3775 (defun ef-pth-croot (c)
3776 (let ((p *gf-char*) (*ef-arith?* t))
3777 (dotimes (i (1- *gf-exp*) c)
3778 (setq c (gf-cpow c p)) )))
3780 (defun gf-pth-root (x)
3781 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3782 (maybe-char-is-fixnum-let ((p *gf-char*))
3783 (if (null x) nil
3784 (do ((rx x (cddr rx)) res c)
3785 ((null rx) (nreverse res))
3786 (push (truncate (the fixnum (car rx)) p) res)
3787 (setq c (cadr rx))
3788 (when *ef-arith?* ;; p # q
3789 (setq c (ef-pth-croot c)) )
3790 (push c res) ))))
3792 (defun gf-gcd-cofactors (x dx)
3793 (let ((g (gf-gcd x dx)))
3794 (values g (gf-divide x g) (gf-divide dx g)) ))
3796 (defun gf-square-free (x) ;; monic x
3797 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3798 (let (f fs (r (gf-diff x)) g)
3799 (cond
3800 ((equal '(0 1) (setq g (gf-gcd x r))) `((1 ,x)))
3802 (when r ;; # 0
3803 (setq r (gf-divide x g)
3804 x g ) ;; d.h. x # 1
3805 (do ((m 1 (1+ m)))
3806 ((equal '(0 1) r))
3807 (declare (fixnum m))
3808 (multiple-value-setq (r f x) (gf-gcd-cofactors r x))
3809 (unless (equal '(0 1) f)
3810 (push (list m f) fs) )))
3811 (unless (equal '(0 1) x)
3812 (setq fs
3813 (append (mapcar #'(lambda (v) (rplaca v (* (car v) *gf-char*)))
3814 (gf-square-free (gf-pth-root x)) )
3815 fs )))
3816 (nreverse fs) ))))
3818 (defun gf-distinct-degree-factors (x q)
3819 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3820 (let ((w '(1 1)) f fs (*gf-char* (car (cfactorw q))) )
3821 (do ((n 1 (1+ n)))
3822 ((equal '(0 1) x) fs)
3823 (declare (fixnum n))
3824 (when (> (ash n 1) (car x))
3825 (setq fs (cons (list x (car x)) fs))
3826 (return) )
3827 (setq w (gf-nred w x)
3828 w (gf-pow w q x)
3829 f (gf-gcd (gf-plus w (gf-nminus (list 1 1))) x) )
3830 (unless (equal '(0 1) f)
3831 (setq fs (cons (list f n) fs)
3832 x (gf-divide x f) )))
3833 (nreverse fs) ))
3835 (defun gf-nonconst-random (q q^n)
3836 (do (r) (())
3837 (setq r (random q^n))
3838 (when (>= r q) (return (let ((*gf-char* q)) (gf-n2x r)))) ))
3840 ;; computes Tm(x) = x^2^(m-1) + x^2^(m-2) + .. + x^4 + x^2 + x in F2[x]
3842 (defun gf-trace-poly-f2 (x m red) ;; m > 0
3843 (let ((tm (gf-nred x red)))
3844 (do ((i 1 (1+ i)))
3845 ((= i m) tm)
3846 (declare (fixnum i))
3847 (setq x (gf-sq x red)
3848 tm (gf-plus tm x) ))))
3850 ;; Cantor and Zassenhaus' algorithm
3852 (defun gf-equal-degree-factors (x-and-d q mult)
3853 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3854 (let* ((x (car x-and-d)) (d (cadr x-and-d))
3855 (n (car x)) )
3856 (declare (fixnum d n))
3857 (cond
3858 ((= n d) (list x mult))
3860 (let* ((p^k (cfactorw q)) (p (car p^k)) (k (cadr p^k)) (*gf-char* p)
3861 (f '(0 1)) (q^n (expt q n)) m e r r^e )
3862 (if (= 2 p)
3863 (setq m (* k d)) ;; q = 2^k
3864 (setq e (ash (1- (expt q d)) -1)) )
3866 (do () ((and (not (equal '(0 1) f)) (not (equal x f))))
3867 (setq r (gf-nonconst-random q q^n)
3868 f (gf-gcd x r) )
3869 (when (equal '(0 1) f)
3870 (setq r^e
3871 (if (= 2 p) (gf-trace-poly-f2 r m x) ;; q = 2^k
3872 (gf-pow r e x) )) ;; q is odd prime power
3873 (setq f (gf-gcd x (gf-nplus r^e (gf-nminus (list 0 1))))) ))
3875 (append (gf-equal-degree-factors (list (gf-divide x f) d) q mult)
3876 (gf-equal-degree-factors (list f d) q mult) ))))))
3878 ;; -----------------------------------------------------------------------------
3881 ;; gcd, gcdex and test of invertibility ----------------------------------------
3884 (defmfun $ef_gcd (a b)
3885 (ef-gf-field? "ef_gcd")
3886 (let ((*ef-arith?* t))
3887 (gf-x2p (gf-gcd (gf-p2x a) (gf-p2x b))) ))
3889 (defmfun $gf_gcd (a b &optional p)
3890 (let ((*ef-arith?*))
3891 (cond
3892 (p (unless (and (integerp p) (primep p))
3893 (gf-merror (intl:gettext "`gf_gcd': ~m is not a prime number.") p) )
3894 (gf-set-rat-header)
3895 (let* ((*gf-char* p)
3896 (modulus p)
3897 (vars (caddar ($rat a))) )
3898 (when (> (length vars) 1)
3899 (gf-merror (intl:gettext "`gf_gcd': Polynomials must be univariate.")) )
3900 (gf-x2p (gf-gcd (gf-p2x a) (gf-p2x b))) ))
3901 (t (gf-char? "gf_gcd")
3902 (gf-x2p (gf-gcd (gf-p2x a) (gf-p2x b))) ))))
3905 (defmfun $gf_gcdex (a b)
3906 (gf-red? "gf_gcdex")
3907 (let ((*ef-arith?*))
3908 (cons '(mlist simp)
3909 (mapcar #'gf-x2p (gf-gcdex (gf-p2x a) (gf-p2x b) *gf-red*)) )))
3911 (defmfun $ef_gcdex (a b)
3912 (ef-red? "ef_gcdex")
3913 (let ((*ef-arith?* t))
3914 (cons '(mlist simp)
3915 (mapcar #'gf-x2p (gf-gcdex (gf-p2x a) (gf-p2x b) *gf-red*)) )))
3918 (defmfun $gf_unit_p (a)
3919 (gf-red? "gf_unit_p")
3920 (let ((*ef-arith?*))
3921 (gf-unit-p (gf-p2x a) *gf-red*) ))
3923 (defmfun $ef_unit_p (a)
3924 (ef-red? "ef_unit_p")
3925 (let ((*ef-arith?* t))
3926 (gf-unit-p (gf-p2x a) *ef-red*) ))
3928 (defun gf-unit-p (x red)
3929 (= 0 (car (gf-gcd x red))) )
3931 ;; -----------------------------------------------------------------------------
3934 ;; order -----------------------------------------------------------------------
3937 ;; group/element order
3939 (defmfun $gf_order (&optional a)
3940 (gf-data? "gf_order")
3941 (cond
3942 (a (let ((*ef-arith?*))
3943 (setq a (gf-p2x a))
3944 (when (and a (or *gf-irred?* (gf-unit-p a *gf-red*)))
3945 (gf-ord a *gf-ord* *gf-fs-ord* *gf-red*) )))
3946 (t *gf-ord*) ))
3948 (defmfun $ef_order (&optional a)
3949 (ef-data? "ef_order")
3950 (cond
3951 (a (let ((*ef-arith?* t))
3952 (setq a (gf-p2x a))
3953 (when (and a (or *ef-irred?* (gf-unit-p a *ef-red*)))
3954 (gf-ord a *ef-ord* *ef-fs-ord* *ef-red*) )))
3955 (t *ef-ord*) ))
3957 ;; find the lowest value k for which a^k = 1
3959 (defun gf-ord (x ord fs-ord red) ;; assume x # 0
3960 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3961 (let (p (e 0) z)
3962 (declare (fixnum e))
3963 (dolist (pe fs-ord ord)
3964 (setq p (car pe)
3965 e (the fixnum (cadr pe))
3966 ord (truncate ord (expt p e))
3967 z (gf-pow$ x ord red) ) ;; use exponentiation by precomputation
3968 (do ()
3969 ((equal z '(0 1)))
3970 (setq ord (* ord p))
3971 (when (= e 1) (return))
3972 (decf e)
3973 (setq z (gf-pow$ z p red)) ))))
3975 (defun gf-ord-by-table (x)
3976 (let ((index (svref $gf_logs (gf-x2n x))))
3977 (truncate *gf-ord* (gcd *gf-ord* index)) ))
3980 ;; Fq^n = F[x]/(f) is no field <=> f splits into factors
3982 ;; f = f1^e1 * ... * fk^ek where fi are irreducible of degree ni.
3984 ;; We compute the order of the group (F[x]/(fi^ei))* by
3986 ;; ((q^ni)^ei - (q^ni)^(ei-1)) = ((q^ni) - 1) * (q^ni)^(ei-1)
3988 ;; and ord((Fq^n)*) with help of the Chinese Remainder Theorem.
3990 (defun gf-group-order (q red)
3991 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
3992 (let (ne-list q^n (e 0) (ord 1))
3993 (declare (fixnum e))
3994 (do ((x (gf-factor red q))) ;; red is assumed to be a monic poly
3995 ((null x))
3996 (push (list (caar x) (cadr x)) ne-list) ;; (list ni ei), f = prod(fi^ei) with fi of degree ni
3997 (setq x (cddr x)) )
3998 (dolist (a ne-list)
3999 (setq q^n (expt q (the fixnum (car a)))
4000 e (the fixnum (cadr a))
4001 ord (* ord (1- q^n) (expt q^n (the fixnum (1- e)))) ))
4002 ord ))
4004 ;; -----------------------------------------------------------------------------
4007 ;; degree, minimal polynomial, trace and norm ----------------------------------
4011 ;; degree: Find the lowest value d for which x^(q^d) = x
4013 (defun gf-degree-errchk (a n fun)
4014 (when (and (not (null a)) (>= (car a) n))
4015 (gf-merror (intl:gettext "`~m': Leading exponent must be smaller than ~m.") fun n) ))
4017 (defmfun $gf_degree (a)
4018 (gf-field? "gf_degree")
4019 (let ((*ef-arith?*))
4020 (setq a (gf-p2x a))
4021 (gf-degree-errchk a *gf-exp* "gf_degree")
4022 (*f-deg a *gf-exp* *gf-red* *gf-x^p-powers*) ))
4024 (defmfun $ef_degree (a)
4025 (ef-field? "ef_degree")
4026 (let ((*ef-arith?* t))
4027 (setq a (gf-p2x a))
4028 (gf-degree-errchk a *ef-exp* "ef_degree")
4029 (*f-deg a *ef-exp* *ef-red* *ef-x^q-powers*) ))
4031 (defun *f-deg (x n red x^q-powers)
4032 (do ((d 1 (1+ d)))
4033 ((= d n) d)
4034 (declare (fixnum d))
4035 (when (equal x (gf-compose (svref x^q-powers d) x red)) ;; f(x)^q = f(x^q)
4036 (return d) ) ))
4039 ;; produce the minimal polynomial
4041 (defmfun $gf_minimal_poly (a)
4042 (gf-field? "gf_minimal_poly")
4043 (let ((*ef-arith?*))
4044 (setq a (gf-p2x a))
4045 (gf-degree-errchk a *gf-exp* "gf_minimal_poly")
4046 (gf-minpoly a *gf-red* *gf-x^p-powers*) ))
4048 (defmfun $ef_minimal_poly (a)
4049 (ef-field? "ef_minimal_poly")
4050 (let ((*ef-arith?* t))
4051 (setq a (gf-p2x a))
4052 (gf-degree-errchk a *ef-exp* "ef_minimal_poly")
4053 (gf-minpoly a *ef-red* *ef-x^q-powers*) ))
4055 ;; 2 (d-1)
4056 ;; q q q
4057 ;; f(z) = (z - x) (z - x ) (z - x ) ... (z - x ) , where d = degree(x)
4059 (defun gf-minpoly (x red x^q-powers)
4060 (if (null x) '$z
4061 (let ((n (car red))
4062 (powers (list (gf-minus x)))
4063 (prod (list 0 (list 0 1)))
4064 xqi zx cx ) (declare (fixnum n))
4065 (do ((i 1 (1+ i)))
4066 ((= i n)) (declare (fixnum i))
4067 (setq xqi (gf-compose (svref x^q-powers i) x red))
4068 (when (equal x xqi) (return))
4069 (push (gf-nminus xqi) powers) )
4070 (dolist (pow powers)
4071 (setq zx (gf-zx prod)
4072 cx (gf-ncx pow prod red)
4073 prod (gf-nzx+cx zx cx)) )
4074 ($substitute '$z '$x (gf-x2p (gf-nxx2x prod))) )))
4076 (defun gf-zx (x) ;; (gf-zx '(3 (5 1 3 1) 2 (4 1))) -> (4 (5 1 3 1) 3 (4 1))
4077 ;; 3 5 3 2 4 4 5 3 3 4
4078 ;; z (x + x ) + z x -> z (x + x ) + z x
4079 (do* ((res (list (1+ (car x)) (cadr x)))
4080 (r (cdr res) (cddr r))
4081 (rx (cddr x) (cddr rx)) )
4082 ((null rx) res)
4083 (rplacd r (list (1+ (car rx)) (cadr rx))) ))
4085 (defun gf-ncx (c x red) ;; modifies x
4086 ;; (gf-ncx '(1 1) '(3 (4 1 3 1) 2 (2 1)) '(6 1))
4087 ;; -> (3 (5 1 4 1) 2 (3 1))
4088 (if (null c) c
4089 (do ((r (cdr x) (cddr r)))
4090 ((null r) x)
4091 (rplaca r (gf-times c (car r) red)) )))
4093 (defun gf-nzx+cx (zx cx) ;; modifies zx
4094 (do ((r (cdr zx))
4095 (s (cdr cx) (cddr s)) r+s )
4096 ((null (cdr r)) (nconc zx (list 0 (car s))))
4097 (setq r+s (gf-plus (caddr r) (car s)))
4098 (if r+s
4099 (rplaca (setq r (cddr r)) r+s)
4100 (rplacd r (cdddr r)) )))
4102 (defun gf-nxx2x (xx) ;; modifies xx
4103 ;; (gf-nxx2x '(4 (0 3) 2 (0 1))) -> (4 3 2 1)
4104 (do ((r (cdr xx) (cddr r)))
4105 ((null r) xx)
4106 (rplaca r (cadar r)) ))
4109 ;; 2 (n-1)
4110 ;; q q q
4111 ;; trace(a) = a + a + a + .. + a
4113 (defmfun $gf_trace (a)
4114 (gf-field? "gf_trace")
4115 (let ((*ef-arith?*))
4116 (gf-trace (gf-p2x a) *gf-red* *gf-x^p-powers*) ))
4118 (defmfun $ef_trace (a)
4119 (ef-field? "ef_trace")
4120 (let ((*ef-arith?* t))
4121 (gf-trace (gf-p2x a) *ef-red* *ef-x^q-powers*) ))
4123 (defun gf-trace (x red x^q-powers)
4124 (let ((n (car red))
4125 (su x) )
4126 (do ((i 1 (1+ i)))
4127 ((= i n) (gf-x2p su)) (declare (fixnum i))
4128 (setq su (gf-plus su (gf-compose (svref x^q-powers i) x red))) )))
4131 ;; 2 (n-1) n 2 (n-1)
4132 ;; 1 + q + q + .. + q (q - 1)/(q - 1) q q q
4133 ;; norm(a) = a = a = a * a * a * .. * a
4135 (defmfun $gf_norm (a)
4136 (gf-field? "gf_norm")
4137 (let ((*ef-arith?*))
4138 (gf-norm (gf-p2x a) *gf-red* *gf-x^p-powers*) ))
4140 (defmfun $ef_norm (a)
4141 (ef-field? "ef_norm")
4142 (let ((*ef-arith?* t))
4143 (gf-norm (gf-p2x a) *ef-red* *ef-x^q-powers*) ))
4145 (defun gf-norm (x red x^q-powers)
4146 (let ((n (car red))
4147 (prod x) )
4148 (do ((i 1 (1+ i)))
4149 ((= i n) (gf-x2p prod)) (declare (fixnum i))
4150 (setq prod (gf-times prod (gf-compose (svref x^q-powers i) x red) red)) )))
4152 ;; -----------------------------------------------------------------------------
4155 ;; normal elements and normal basis --------------------------------------------
4158 ;; Tests if an element is normal
4160 (defmfun $gf_normal_p (a)
4161 (gf-field? "gf_normal_p")
4162 (let ((*ef-arith?*)) (gf-normal-p (gf-p2x a))) )
4164 (defun gf-normal-p (x)
4165 (unless (null x)
4166 (let ((modulus *gf-char*)
4167 (mat (gf-maybe-normal-basis x)) )
4168 (equal ($rank mat) *gf-exp*) )))
4170 (defmfun $ef_normal_p (a)
4171 (ef-field? "ef_normal_p")
4172 (let ((*ef-arith?* t)) (ef-normal-p (gf-p2x a))) )
4174 (defun ef-normal-p (x)
4175 (unless (null x)
4176 (let ((mat (ef-maybe-normal-basis x)) )
4177 (/= 0 ($ef_determinant mat)) )))
4180 ;; Finds a normal element e in the field; that is,
4181 ;; an element for which the list [e, e^q, e^(q^2), ... , e^(q^(n-1))] is a basis
4183 (defmfun $gf_normal ()
4184 (gf-field? "gf_normal")
4185 (let ((*ef-arith?*))
4186 (gf-x2p (gf-normal *gf-char* *gf-exp* #'gf-normal-p)) ))
4188 (defmfun $ef_normal ()
4189 (ef-field? "ef_normal")
4190 (let ((*ef-arith?* t))
4191 (gf-x2p (gf-normal *gf-card* *ef-exp* #'ef-normal-p)) ))
4193 (defun gf-normal (q n normal-p-fn)
4194 (let* ((inc (min $gf_coeff_limit q))
4195 (i-lim (expt inc n))
4197 (do ((i inc (1+ i)))
4198 ((>= i i-lim)
4199 (gf-merror (intl:gettext "No normal element found.~%~\
4200 `gf_coeff_limit' might be too small.~%" )) )
4201 (setq x (let ((*gf-char* inc)) (gf-n2x i)))
4202 (when (funcall normal-p-fn x ) (return-from gf-normal x)) )))
4205 ;; Finds a normal element in the field by producing random elements and checking
4206 ;; if each one is normal
4208 (defmfun $gf_random_normal ()
4209 (gf-field? "gf_random_normal")
4210 (let ((*ef-arith?*)) (gf-x2p (gf-random-normal))) )
4212 (defun gf-random-normal ()
4213 (do ((x (gf-random *gf-char* *gf-exp*) (gf-random *gf-char* *gf-exp*)))
4214 ((gf-normal-p x) x) ))
4216 (defmfun $ef_random_normal ()
4217 (ef-field? "ef_random_normal")
4218 (let ((*ef-arith?* t)) (gf-x2p (ef-random-normal))) )
4220 (defun ef-random-normal ()
4221 (do ((x (gf-random *gf-card* *ef-exp*) (gf-random *gf-card* *ef-exp*)))
4222 ((ef-normal-p x) x) ))
4224 ;; Produces a normal basis as a matrix;
4225 ;; the columns are the coefficients of the powers e^(q^i) of the normal element
4227 (defmfun $gf_normal_basis (a)
4228 (gf-field? "gf_normal_basis")
4229 (let* ((*ef-arith?*)
4230 (x (gf-p2x a))
4231 (modulus *gf-char*)
4232 (mat (gf-maybe-normal-basis x)) )
4233 (unless (equal ($rank mat) *gf-exp*)
4234 (gf-merror (intl:gettext "Argument to `gf_normal_basis' must be a normal element.")) )
4235 mat ))
4237 (defmfun $ef_normal_basis (a)
4238 (ef-field? "ef_normal_basis")
4239 (let* ((*ef-arith?* t)
4240 (mat (ef-maybe-normal-basis (gf-p2x a))) )
4241 (unless (/= 0 ($ef_determinant mat))
4242 (merror (intl:gettext "Argument to `ef_normal_basis' must be a normal element." )) )
4243 mat ))
4245 (defun gf-maybe-normal-basis (x)
4246 (*f-maybe-normal-basis x *gf-x^p-powers* *gf-exp* *gf-red*) )
4248 (defun ef-maybe-normal-basis (x)
4249 (*f-maybe-normal-basis x *ef-x^q-powers* *ef-exp* *ef-red*) )
4251 (defun *f-maybe-normal-basis (x x^q-powers e red)
4252 #+ (or ccl ecl gcl sbcl) (declare (optimize (speed 3) (safety 0)))
4253 (declare (fixnum e))
4254 (let ((e-1 (- e 1))) (declare (fixnum e-1))
4255 ($transpose
4256 ($genmatrix
4257 #'(lambda (i j) (declare (fixnum i j))
4258 (svref (gf-x2array
4259 (gf-compose (svref x^q-powers (1- i)) x red) e-1 ) (1- j)))
4260 e e ))))
4262 ;; coefficients as an array of designated length
4264 (defun gf-x2array (x len)
4265 #+ (or ccl ecl gcl sbcl) (declare (optimize (speed 3) (safety 0)))
4266 (declare (fixnum len))
4267 (let ((cs (make-array (1+ len) :initial-element 0)))
4268 (do ((k len)) ((null x) cs) (declare (fixnum k))
4269 (cond
4270 ((> k (the fixnum (car x)))
4271 (decf k) )
4272 ((= k (the fixnum (car x)))
4273 (setf (svref cs (- len k)) (cadr x))
4274 (setq x (cddr x))
4275 (decf k) )
4277 (setq x (cddr x)) ) ))))
4279 ;; Produces the normal representation of an element as a list of coefficients
4280 ;; Needs the inverted normal basis as the second argument
4281 ;; (the inversion might be time consuming)
4283 (defmfun $gf_normal_basis_rep (a m-inv)
4284 (gf-field? "gf_normal_basis_rep")
4285 (let ((*ef-arith?*))
4286 (gf-normal-basis-rep (gf-p2x a) m-inv *gf-exp* '$gf_matmult) ))
4288 (defmfun $ef_normal_basis_rep (a m-inv)
4289 (ef-field? "ef_normal_basis_rep")
4290 (let ((*ef-arith?* t))
4291 (gf-normal-basis-rep (gf-p2x a) m-inv *ef-exp* '$ef_matmult) ))
4293 (defun gf-normal-basis-rep (x m-inv e matmult-fn)
4294 (let* ((cs (cons '(mlist simp) (gf-x2l x e)))
4295 (nbrep (mfuncall matmult-fn m-inv cs)) )
4296 (cons '(mlist simp) (mapcar #'cadr (cdr nbrep))) ))
4298 ;; -----------------------------------------------------------------------------
4301 ;; functions for matrices ------------------------------------------------------
4304 (defmfun $gf_matneg (m)
4305 (gf-char? "gf_matneg")
4306 (mfuncall '$matrixmap '$gf_neg m) )
4308 (defmfun $ef_matneg (m)
4309 (ef-gf-field? "ef_matneg")
4310 (mfuncall '$matrixmap '$ef_neg m) )
4313 ;; matrix addition (convenience: mat, list or poly possible as argument)
4315 (defmfun $gf_matadd (&rest args)
4316 (let ((*ef-arith?*))
4317 (reduce #'(lambda (m1 m2) (gf-matadd m1 m2 '$gf_add)) args) ))
4319 (defmfun $ef_matadd (&rest args)
4320 (gf-data? "ef_matadd")
4321 (let ((*ef-arith?* t))
4322 (reduce #'(lambda (m1 m2) (gf-matadd m1 m2 '$ef_add)) args) ))
4324 (defun gf-matadd (m1 m2 add-fn)
4325 (when ($listp m1) (setq m1 ($transpose m1)))
4326 (when ($listp m2) (setq m2 ($transpose m2)))
4327 (cond
4328 ((and ($matrixp m1) ($matrixp m2))
4329 (gf-matadd2 m1 m2 add-fn) )
4330 (($matrixp m1)
4331 (gf-matadd1 m1 m2 add-fn) ) ;; assumed without checking: m2 is poly
4332 (($matrixp m2)
4333 (gf-matadd1 m2 m1 add-fn) )
4335 (mfuncall add-fn m1 m2) ) ))
4337 (defun gf-matadd1 (m poly add-fn)
4338 (do ((r (cdr m) (cdr r)) new)
4339 ((null r) (cons '($matrix simp) (nreverse new)))
4340 (push (cons '(mlist simp)
4341 (mapcar #'(lambda (p) (mfuncall add-fn p poly)) (cdar r)) )
4342 new )))
4344 (defun gf-matadd2-error ()
4345 (gf-merror
4346 (intl:gettext "Arguments to `~m' must have same formal structure.")
4347 (if *ef-arith?* "ef_matadd" "gf_matadd") ))
4349 (defun gf-matadd2 (m1 m2 add-fn)
4350 (setq m1 (cdr m1) m2 (cdr m2))
4351 (unless (= (length (car m1)) (length (car m2)))
4352 (gf-matadd2-error) )
4353 (do ((r1 m1 (cdr r1)) (r2 m2 (cdr r2)) new)
4354 ((or (null r1) (null r2))
4355 (unless (and (null r1) (null r2))
4356 (gf-matadd2-error) )
4357 (cons '($matrix simp) (nreverse new)) )
4358 (push (cons '(mlist simp) (mapcar add-fn (cdar r1) (cdar r2))) new) ))
4361 ;; matrix multiplication (convenience: mat, list or poly possible as argument)
4363 (defmfun $gf_matmult (&rest args)
4364 (gf-red? "gf_matmult")
4365 (let ((*ef-arith?*))
4366 (apply #'*f-matmult `($gf_mult ,@args)) ))
4368 (defmfun $ef_matmult (&rest args)
4369 (ef-red? "ef_matmult")
4370 (let ((*ef-arith?* t))
4371 (apply #'*f-matmult `($ef_mult ,@args)) ))
4373 (defun *f-matmult (mult-fn &rest args)
4374 ($rreduce
4375 #'(lambda (m1 m2) (gf-matmult m1 m2 mult-fn))
4376 (cons '(mlist simp) args) ))
4378 (defun gf-matmult (m1 m2 mult-fn)
4379 (when ($listp m1) (setq m1 (list '($matrix simp) m1)))
4380 (when ($listp m2) (setq m2 ($transpose m2)))
4381 (cond
4382 ((and ($matrixp m1) ($matrixp m2))
4383 (gf-matmult2 m1 m2) )
4384 (($matrixp m1)
4385 (gf-matmult1 m1 m2 mult-fn) ) ;; assumed without checking: m2 is poly
4386 (($matrixp m2)
4387 (gf-matmult1 m2 m1 mult-fn) )
4389 (mfuncall mult-fn m1 m2) ) ))
4391 (defun gf-matmult1 (m poly mult-fn)
4392 (do ((r (cdr m) (cdr r)) new)
4393 ((null r) (cons '($matrix simp) (nreverse new)))
4394 (push (cons '(mlist simp)
4395 (mapcar #'(lambda (p) (mfuncall mult-fn p poly)) (cdar r)) )
4396 new )))
4398 (defun gf-matmult2 (m1 m2)
4399 (setq m1 (cdr m1) m2 (cdr ($transpose m2)))
4400 (unless (= (length (car m1)) (length (car m2)))
4401 (gf-merror
4402 (intl:gettext "`~m': attempt to multiply non conformable matrices.")
4403 (if *ef-arith?* "ef_matmult" "gf_matmult") ))
4404 (let ((red (if *ef-arith?* *ef-red* *gf-red*)) )
4405 (do ((r1 m1 (cdr r1)) new-mat)
4406 ((null r1)
4407 (if (and (not (eq nil $scalarmatrixp))
4408 (= 1 (length new-mat)) (= 1 (length (cdar new-mat))) )
4409 (cadar new-mat)
4410 (cons '($matrix simp) (nreverse new-mat)) ))
4411 (do ((r2 m2 (cdr r2)) new-row)
4412 ((null r2)
4413 (push (cons '(mlist simp) (nreverse new-row)) new-mat) )
4414 (push (gf-x2p
4415 (reduce #'gf-nplus
4416 (mapcar #'(lambda (a b) (gf-times (gf-p2x a) (gf-p2x b) red))
4417 (cdar r1) (cdar r2) )))
4418 new-row ) ))))
4420 ;; -----------------------------------------------------------------------------
4423 ;; invert and determinant by lu ------------------------------------------------
4426 (defun zn-p-errchk (p fun pos)
4427 (unless (and p (integerp p) (primep p))
4428 (gf-merror (intl:gettext "`~m': ~m argument must be prime number.") fun pos) ))
4430 ;; invert by lu
4432 ;; returns nil when LU-decomposition fails
4433 (defun try-lu-and-call (fun m ring)
4434 (let ((old-out *standard-output*)
4435 (redirect (make-string-output-stream))
4436 (res nil) )
4437 (unwind-protect ;; protect against lu failure
4438 (setq *standard-output* redirect ;; discard error messages from lu-factor
4439 res (mfuncall fun m ring) )
4440 (progn
4441 (setq *standard-output* old-out)
4442 (close redirect) )
4443 (return-from try-lu-and-call res) )))
4445 (defmfun $zn_invert_by_lu (m p)
4446 (zn-p-errchk p "zn_invert_by_lu" "Second")
4447 (let ((*ef-arith?*) (*gf-char* p))
4448 (try-lu-and-call '$invert_by_lu m 'gf-coeff-ring) ))
4450 (defmfun $gf_matinv (m)
4451 (format t "`gf_matinv' is deprecated. ~%~\
4452 The user is asked to use `gf_invert_by_lu' instead.~%" )
4453 ($gf_invert_by_lu m) )
4455 (defmfun $gf_invert_by_lu (m)
4456 (gf-field? "gf_invert_by_lu")
4457 (let ((*ef-arith?*)) (try-lu-and-call '$invert_by_lu m 'gf-ring)) )
4459 (defmfun $ef_invert_by_lu (m)
4460 (ef-field? "ef_invert_by_lu")
4461 (let ((*ef-arith?* t)) (try-lu-and-call '$invert_by_lu m 'ef-ring)) )
4463 ;; determinant
4465 (defmfun $zn_determinant (m p)
4466 (zn-p-errchk p "zn_determinant" "Second")
4467 (let* ((*ef-arith?*)
4468 (*gf-char* p)
4469 (det (try-lu-and-call '$determinant_by_lu m 'gf-coeff-ring)) )
4470 (if det det
4471 (mod (mfuncall '$determinant m) p) )))
4473 (defmfun $gf_determinant (m)
4474 (gf-field? "gf_determinant")
4475 (let* ((*ef-arith?*)
4476 (det (try-lu-and-call '$determinant_by_lu m 'gf-ring)) )
4477 (if det det
4478 (let (($matrix_element_mult '$gf_mult) ($matrix_element_add '$gf_add))
4479 (mfuncall '$determinant m) ))))
4481 (defmfun $ef_determinant (m)
4482 (ef-field? "ef_determinant")
4483 (let* ((*ef-arith?* t)
4484 (det (try-lu-and-call '$determinant_by_lu m 'ef-ring)) )
4485 (if det det
4486 (let (($matrix_element_mult '$ef_mult) ($matrix_element_add '$ef_add))
4487 (mfuncall '$determinant m) ))))
4489 ;; -----------------------------------------------------------------------------
4492 ;; discrete logarithm ----------------------------------------------------------
4495 ;; solve g^x = a in Fq^n, where g a generator of (Fq^n)*
4497 (defmfun $gf_index (a)
4498 (gf-data? "gf_index")
4499 (gf-log-errchk1 *gf-prim* "gf_index")
4500 (let ((*ef-arith?*))
4501 (if (= 1 *gf-exp*)
4502 ($zn_log a (gf-x2n *gf-prim*) *gf-char*)
4503 (gf-dlog (gf-p2x a)) )))
4505 (defmfun $ef_index (a)
4506 (ef-data? "ef_index")
4507 (gf-log-errchk1 *ef-prim* "ef_index")
4508 (let ((*ef-arith?* t))
4509 (setq a (gf-p2x a))
4510 (if (= 1 *ef-exp*)
4511 (let ((*ef-arith?*)) (gf-dlog (gf-n2x (cadr a))))
4512 (ef-dlog a) )))
4514 (defmfun $gf_log (a &optional b)
4515 (gf-data? "gf_log")
4516 (gf-log-errchk1 *gf-prim* "gf_log")
4517 (let ((*ef-arith?*))
4518 (cond
4519 ((= 1 *gf-exp*)
4520 ($zn_log a (if b b (gf-x2n *gf-prim*)) *gf-char*) ) ;; $zn_log checks if b is primitive
4522 (setq a (gf-p2x a))
4523 (and b (setq b (gf-p2x b)) (gf-log-errchk2 b #'gf-prim-p "gf_log"))
4524 (if b
4525 (gf-dlogb a b)
4526 (gf-dlog a) )))))
4528 (defun gf-log-errchk1 (prim fun)
4529 (when (null prim)
4530 (gf-merror (intl:gettext "`~m': there is no primitive element.") fun) )
4531 (when (equal prim '$unknown)
4532 (gf-merror (intl:gettext "`~m': a primitive element is not known.") fun) ))
4534 (defun gf-log-errchk2 (x prim-p-fn fun)
4535 (unless (funcall prim-p-fn x)
4536 (gf-merror (intl:gettext
4537 "Second argument to `~m' must be a primitive element." ) fun )))
4539 (defmfun $ef_log (a &optional b)
4540 (ef-data? "ef_log")
4541 (gf-log-errchk1 *ef-prim* "ef_log")
4542 (let ((*ef-arith?* t))
4543 (setq a (gf-p2x a))
4544 (and b (setq b (gf-p2x b)) (gf-log-errchk2 b #'ef-prim-p "ef_log")) )
4545 (cond
4546 ((= 1 *ef-exp*)
4547 (let ((*ef-arith?*))
4548 (setq a (gf-n2x (cadr a)))
4549 (if b
4550 (gf-dlogb a (gf-n2x (cadr b)))
4551 (gf-dlog a) )))
4553 (let ((*ef-arith?* t))
4554 (if b
4555 (ef-dlogb a b)
4556 (ef-dlog a) )))))
4558 (defun gf-dlogb (a b)
4559 (*f-dlogb a b #'gf-dlog *gf-ord*) )
4561 (defun ef-dlogb (a b)
4562 (*f-dlogb a b #'ef-dlog *ef-ord*) )
4564 (defun *f-dlogb (a b dlog-fn ord)
4565 (let* ((a-ind (funcall dlog-fn a)) (b-ind (funcall dlog-fn b))
4566 (d (gcd (gcd a-ind b-ind) ord))
4567 (m (truncate ord d)) )
4568 (zn-quo (truncate a-ind d) (truncate b-ind d) m) ))
4570 ;; Pohlig and Hellman reduction
4572 (defun gf-dlog (a)
4573 (*f-dlog a *gf-prim* *gf-red* *gf-ord* *gf-fs-ord*) )
4575 (defun ef-dlog (a)
4576 (*f-dlog a *ef-prim* *ef-red* *ef-ord* *ef-fs-ord*) )
4578 (defun *f-dlog (a g red ord fs-ord)
4579 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
4580 (cond
4581 ((or (null a) (null g)) nil)
4582 ((>= (car a) (car red)) nil)
4583 ((equal '(0 1) a) 0)
4584 ((equal g a) 1)
4585 ((not (gf-unit-p a red)) nil)
4587 (let (p (e 0) ord/p om xp xk dlogs mods (g-inv (gf-inv g red)))
4588 (declare (fixnum e))
4589 (dolist (f fs-ord)
4590 (setq p (car f) e (cadr f)
4591 ord/p (truncate ord p)
4592 om (gf-pow g ord/p red)
4593 xp 0 )
4594 (do ((b a) (k 0) (pk 1) (acc g-inv) (e1 (1- e)))
4595 (()) (declare (fixnum k))
4596 (setq xk (gf-dlog-rho-brent (gf-pow b ord/p red) om p red))
4597 (incf xp (* xk pk))
4598 (incf k)
4599 (when (= k e) (return))
4600 (setq ord/p (truncate ord/p p)
4601 pk (* pk p) )
4602 (when (/= xk 0) (setq b (gf-times b (gf-pow acc xk red) red)))
4603 (when (/= k e1) (setq acc (gf-pow acc p red))) )
4604 (push (expt p e) mods)
4605 (push xp dlogs) )
4606 (car (chinese dlogs mods)) ))))
4608 ;; iteration for Pollard rho: b = g^y * a^z in each step
4610 (defun gf-dlog-f (b y z a g p red)
4611 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
4612 (let ((m (mod (cadr b) 3))) (declare (fixnum m))
4613 (cond
4614 ((= 0 m)
4615 (values (gf-sq b red) (mod (ash y 1) p) (mod (ash z 1) p)) )
4616 ((= 1 m) ;; avoid stationary case b=1 => b^2=1
4617 (values (gf-times g b red) (mod (+ y 1) p) z ) )
4619 (values (gf-times a b red) y (mod (+ z 1) p) ) ) )))
4621 ;; brute-force:
4623 (defun gf-dlog-naive (a g red)
4624 (do ((i 0 (1+ i))
4625 (gi '(0 1) (gf-times gi g red)) )
4626 ((equal gi a) i) ))
4628 ;; baby-steps-giant-steps:
4630 (defun gf-dlog-baby-giant (a g p red) ;; g is generator of order prime p
4631 (let* ((m (1+ (isqrt p)))
4632 (s (floor (* 1.3 m)))
4633 (gi (gf-inv g red))
4634 g^m babies )
4635 (setf babies
4636 (make-hash-table :size s :test #'equal :rehash-threshold 0.9) )
4637 (do ((r 0) (b a))
4638 (())
4639 (when (equal '(0 1) b)
4640 (clrhash babies)
4641 (return-from gf-dlog-baby-giant r) )
4642 (setf (gethash b babies) r)
4643 (incf r)
4644 (when (= r m) (return))
4645 (setq b (gf-times gi b red)) )
4646 (setq g^m (gf-pow g m red))
4647 (do ((rr 0 (1+ rr))
4648 (bb (list 0 1) (gf-times g^m bb red))
4649 r ) (())
4650 (when (setq r (gethash bb babies))
4651 (clrhash babies)
4652 (return (+ r (* rr m))) )) ))
4654 ;; Pollard rho for dlog computation (Brents variant of collision detection)
4656 (defun gf-dlog-rho-brent (a g p red) ;; g is generator of order p
4657 #+ (or ccl ecl gcl) (declare (optimize (speed 3) (safety 0)))
4658 (cond
4659 ((equal '(0 1) a) 0)
4660 ((equal g a) 1)
4661 ((equal a (gf-sq g red)) 2)
4662 ((equal '(0 1) (gf-times a g red)) (1- p))
4663 ((< p 32) (gf-dlog-naive a g red))
4664 ((< p 1024) (gf-dlog-baby-giant a g p red))
4666 (prog ((b (list 0 1)) (y 0) (z 0) ;; b = g^y * a^z
4667 (bb (list 0 1)) (yy 0) (zz 0) ;; bb = g^yy * a^zz
4668 dy dz )
4670 (do ((i 0)(j 1)) (())
4671 (declare (fixnum i j))
4672 (multiple-value-setq (b y z) (gf-dlog-f b y z a g p red))
4673 (when (equal b bb) (return)) ;; g^y * a^z = g^yy * a^zz
4674 (incf i)
4675 (when (= i j)
4676 (setq j (1+ (ash j 1)))
4677 (setq bb b yy y zz z) ))
4678 (setq dy (mod (- yy y) p) dz (mod (- z zz) p)) ;; g^dy = a^dz = g^(x*dz)
4679 (when (= 1 (gcd dz p))
4680 (return (zn-quo dy dz p)) ) ;; x = dy/dz mod p (since g is generator of order p)
4681 (setq y 0
4683 b (list 0 1)
4684 yy (1+ (random (1- p)))
4685 zz (1+ (random (1- p)))
4686 bb (gf-times (gf-pow g yy red) (gf-pow a zz red) red) )
4687 (go rho) ))))
4689 ;; -----------------------------------------------------------------------------
4692 ;; nth root in Galois Fields ---------------------------------------------------
4695 (defmfun $ef_nth_root (a r)
4696 (ef-field? "ef_nth_root")
4697 (unless (and (integerp r) (> r 0))
4698 (gf-merror (intl:gettext "Second argument to `ef_nth_root' must be a positive integer. Found ~m.") a r) )
4699 (let* ((*ef-arith?* t)
4700 (rts (gf-nrt (gf-p2x a) r *ef-red* *ef-ord*)) )
4701 (gf-nrt-exit rts) ))
4703 (defmfun $gf_nth_root (a r)
4704 (gf-field? "gf_nth_root")
4705 (unless (and (integerp r) (> r 0))
4706 (gf-merror (intl:gettext "Second argument to `gf_nth_root' must be a positive integer. Found ~m.") a r) )
4707 (let* ((*ef-arith?*)
4708 (rts (gf-nrt (gf-p2x a) r *gf-red* *gf-ord*)) )
4709 (gf-nrt-exit rts) ))
4711 (defun gf-nrt-exit (rts)
4712 (when rts
4713 (setq rts (mapcar #'gf-n2x (sort (mapcar #'gf-x2n rts) #'<)))
4714 (cons '(mlist simp) (mapcar #'gf-x2p rts)) ))
4716 ;; e.g. r=i*i*j*k, then x^(1/r) = (((x^(1/i))^(1/i))^(1/j))^(1/k)
4717 (defun gf-nrt (x r red ord)
4718 (setq x (gf-nred x red))
4719 (let (rts)
4720 (cond
4721 ((null x) nil)
4722 ((or (= 1 r) (primep r)) (setq rts (gf-amm x r red ord)))
4724 (let* (($intfaclim) (rs (get-factor-list r)))
4725 (setq rs (sort rs #'< :key #'car))
4726 (setq rs
4727 (apply #'nconc
4728 (mapcar
4729 #'(lambda (pe) (apply #'(lambda (p e) (make-list e :initial-element p)) pe))
4730 rs )))
4731 (setq rts (gf-amm x (car rs) red ord))
4732 (dolist (r (cdr rs))
4733 (setq rts (apply #'nconc (mapcar #'(lambda (y) (gf-amm y r red ord)) rts))) ))))
4734 rts ))
4736 ;; inspired by Bach,Shallit 7.3.2
4737 (defun gf-amm (x r red ord) ;; r prime, red irreducible
4738 (if (= 1 r)
4739 (list x)
4740 (let (k n e u s m re xr xu om g br bu ab alpha beta rt)
4741 (when (= 2 r)
4742 (cond
4743 ((and (= 0 (setq m (mod ord 2)))
4744 (/= 1 (gf-jacobi x red (if *ef-arith?* *gf-card* *gf-char*))) )
4745 (return-from gf-amm nil) )
4746 ((= 1 m) ;; q = 2^n : unique solution
4747 (return-from gf-amm
4748 `(,(gf-pow x (ash (+ (ash ord 1) 2) -2) red)) ))
4749 ((= 2 (mod ord 4))
4750 (setq rt (gf-pow x (ash (+ ord 2) -2) red))
4751 (return-from gf-amm `(,rt ,(gf-minus rt))) )
4752 ((= 4 (mod ord 8))
4753 (let* ((twox (gf-plus x x))
4754 (y (gf-pow twox (ash (- ord 4) -3) red))
4755 (i (gf-times twox (gf-times y y red) red))
4756 (rt (gf-times x (gf-times y (gf-nplus i `(0 ,(gf-cminus-b 1))) red) red)) )
4757 (return-from gf-amm `(,rt ,(gf-minus rt))) ))))
4758 (multiple-value-setq (s m) (truncate ord r))
4759 (when (and (= 0 m) (not (equal '(0 1) (gf-pow x s red))))
4760 (return-from gf-amm nil))
4761 ;; r = 3, first 2 cases:
4762 (when (= 3 r)
4763 (cond
4764 ((= 1 (setq m (mod ord 3))) ;; unique solution
4765 (return-from gf-amm
4766 `(,(gf-pow x (truncate (1+ (ash ord 1)) 3) red)) ))
4767 ((= 2 m) ;; unique solution
4768 (return-from gf-amm
4769 `(,(gf-pow x (truncate (1+ ord) 3) red)) ))))
4770 ;; compute u,e with ord = u*r^e and r,u coprime:
4771 (setq u ord e 0)
4772 (do ((u1 u)) (())
4773 (multiple-value-setq (u1 m) (truncate u1 r))
4774 (when (/= 0 m) (return))
4775 (setq u u1 e (1+ e)) )
4776 (cond
4777 ((= 0 e)
4778 (setq rt (gf-pow x (inv-mod r u) red)) ;; unique solution, see Bach,Shallit 7.3.1
4779 (list rt) )
4781 (setq n (gf-n2x 2))
4782 (do () ((not (equal '(0 1) (gf-pow n s red)))) ;; n is no r-th power
4783 (setq n (gf-n2x (1+ (gf-x2n n)))) )
4784 (setq g (gf-pow n u red)
4785 re (expt r e)
4786 om (gf-pow g (truncate re r) red) ) ;; r-th root of unity
4787 (cond
4788 ((or (/= 3 r) (= 0 (setq m (mod ord 9))))
4789 (setq xr (gf-pow x u red)
4790 xu (gf-pow x re red)
4791 k (*f-dlog xr g red re `((,r ,e))) ;; g^k = xr
4792 br (gf-pow g (truncate k r) red) ;; br^r = xr
4793 bu (gf-pow xu (inv-mod r u) red) ;; bu^r = xu
4794 ab (cdr (zn-gcdex1 u re))
4795 alpha (car ab)
4796 beta (cadr ab) )
4797 (if (< alpha 0) (incf alpha ord) (incf beta ord))
4798 (setq rt (gf-times (gf-pow br alpha red) (gf-pow bu beta red) red)) )
4799 ;; r = 3, remaining cases:
4800 ((= 3 m)
4801 (setq rt (gf-pow x (truncate (+ (ash ord 1) 3) 9) red)) )
4802 ((= 6 m)
4803 (setq rt (gf-pow x (truncate (+ ord 3) 9) red)) ))
4804 (do ((i 1 (1+ i)) (j (list 0 1)) (res (list rt)))
4805 ((= i r) res)
4806 (setq j (gf-times j om red))
4807 (push (gf-times rt j red) res) ))))))
4809 ;; -----------------------------------------------------------------------------
4812 ;; tables of small fields ------------------------------------------------------
4815 (defmfun $gf_add_table ()
4816 (gf-data? "gf_add_table")
4817 (let ((*ef-arith?*)) (gf-add-table *gf-card*)) )
4819 (defmfun $ef_add_table ()
4820 (ef-data? "ef_add_table")
4821 (let ((*ef-arith?* t)) (gf-add-table *ef-card*)) )
4823 (defun gf-add-table (card)
4824 ($genmatrix
4825 #'(lambda (i j)
4826 (gf-x2n (gf-plus (gf-n2x (1- i)) (gf-n2x (1- j)))) )
4827 card
4828 card ))
4830 (defmfun $gf_mult_table (&optional all?)
4831 (gf-data? "gf_mult_table")
4832 (let ((*ef-arith?*))
4833 (gf-mult-table *gf-red* *gf-irred?* *gf-card* all?) ))
4835 (defmfun $ef_mult_table (&optional all?)
4836 (ef-data? "ef_mult_table")
4837 (let ((*ef-arith?* t))
4838 (gf-mult-table *ef-red* *ef-irred?* *ef-card* all?) ))
4840 (defun gf-mult-table (red irred? card all?)
4841 (let (units res)
4842 (cond
4843 ((or irred? ;; field
4844 (equal all? '$all) )
4845 ($genmatrix
4846 #'(lambda (i j)
4847 (gf-x2n (gf-times (gf-n2x i) (gf-n2x j) red)))
4848 (1- card)
4849 (1- card) ))
4850 (t ;; units only
4851 (do ((i 1 (1+ i)) x )
4852 ((= i card) )
4853 (setq x (gf-n2x i))
4854 (when (gf-unit-p x red) (push x units)) )
4855 (dolist (x units (cons '($matrix simp) (nreverse res)))
4856 (push
4857 (cons '(mlist simp)
4858 (mapcar #'(lambda (y) (gf-x2n (gf-times x y red))) units) )
4859 res ) )) )))
4861 (defmfun $gf_power_table (&rest args)
4862 (gf-data? "gf_power_table")
4863 (let ((*ef-arith?*) all? cols)
4864 (multiple-value-setq (all? cols)
4865 (apply #'gf-power-table-args (cons "gf_power_table" args)) )
4866 (unless cols
4867 (setq cols *gf-ord*)
4868 (when all? (incf cols)) )
4869 (gf-power-table *gf-red* *gf-irred?* *gf-card* cols all? ) ))
4871 (defmfun $ef_power_table (&rest args)
4872 (ef-data? "ef_power_table")
4873 (let ((*ef-arith?* t) all? cols)
4874 (multiple-value-setq (all? cols)
4875 (apply #'gf-power-table-args (cons "ef_power_table" args)) )
4876 (unless cols
4877 (setq cols *ef-ord*)
4878 (when all? (incf cols)) )
4879 (gf-power-table *ef-red* *ef-irred?* *ef-card* cols all? ) ))
4881 (defun gf-power-table-args (&rest args)
4882 (let ((str (car args)) all? cols (x (cadr args)) (y (caddr args)))
4883 (when x
4884 (cond
4885 ((integerp x) (setq cols x))
4886 ((equal x '$all) (setq all? t))
4887 (t (gf-merror (intl:gettext
4888 "First argument to `~m' must be `all' or a small fixnum." ) str ))))
4889 (when y
4890 (cond
4891 ((and (integerp x) (equal y '$all)) (setq all? t))
4892 ((and (equal x '$all) (integerp y)) (setq cols y))
4893 (t (format t "Only the first argument to `~a' has been used.~%" str) )))
4894 (values all? cols) ))
4896 (defun gf-power-table (red irred? card cols all?)
4897 (cond
4898 ((or irred? all?)
4899 ($genmatrix
4900 #'(lambda (i j) (gf-x2n (gf-pow (gf-n2x i) j red)))
4901 (1- card)
4902 cols ))
4903 (t ;; units only
4904 (do ((i 1 (1+ i)) x res)
4905 ((= i card) (cons '($matrix simp) (nreverse res)))
4906 (setq x (gf-n2x i))
4907 (when (gf-unit-p x red)
4908 (push
4909 (cons '(mlist simp)
4910 (mapcar #'(lambda (j) (gf-x2n (gf-pow x j red)))
4911 (cdr (mfuncall '$makelist '$j '$j 1 cols)) ))
4912 res ) )) )))
4914 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;