Xmaxima: ~/.xmaximrc should probably be ~/.xmaximarc.
[maxima/cygwin.git] / src / gamma.lisp
blob8f5f30ddc08e360ec2f61820cd383facb6a2b70b
1 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2 ;;;
3 ;;; Double Factorial, Incomplete Gamma function, ...
4 ;;;
5 ;;; This file will be extended with further functions related to the
6 ;;; Factorial and Gamma functions.
7 ;;;
8 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
9 ;;;
10 ;;; This file contains the following Maxima User functions:
11 ;;;
12 ;;; double_factorial(z)
13 ;;;
14 ;;; gamma_incomplete(a,z)
15 ;;; gamma_incomplete_generalized(a,z1,z2)
16 ;;; gamma_incomplete_regularized(a,z)
17 ;;;
18 ;;; log_gamma(z)
19 ;;;
20 ;;; erf(z)
21 ;;; erfc(z)
22 ;;; erfi(z)
23 ;;; erf_generalized(z1,z2)
24 ;;;
25 ;;; inverse_erf(z)
26 ;;; inverse_erfc(z)
27 ;;;
28 ;;; fresnel_s(z)
29 ;;; fresnel_c(z)
30 ;;;
31 ;;; beta_incomplete(a,b,z)
32 ;;; beta_incomplete_generalized(a,b,z1,z2)
33 ;;; beta_incomplete_regularized(a,b,z)
34 ;;;
35 ;;; Maxima User variable:
36 ;;;
37 ;;; $factorial_expand - Allows argument simplificaton for expressions like
38 ;;; double_factorial(n-1) and double_factorial(2*k+n)
39 ;;; $beta_expand - Switch on further expansions for the Beta functions
40 ;;;
41 ;;; $erf_representation - When T erfc, erfi, erf_generalized, fresnel_s
42 ;;; and fresnel_c are transformed to erf.
43 ;;; $erf_%iargs - Enable simplification of Erf and Erfi for
44 ;;; imaginary arguments
45 ;;; $hypergeometric_representation
46 ;;; - Enables transformation to a Hypergeometric
47 ;;; representation for fresnel_s and fresnel_c
48 ;;;
49 ;;; Maxima User variable (not definied in this file):
50 ;;;
51 ;;; $factlim - biggest integer for numerically evaluation
52 ;;; of the Double factorial
53 ;;; $gamma_expand - Expansion of the Gamma und Incomplete Gamma
54 ;;; function for some special cases
55 ;;;
56 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
57 ;;; This library is free software; you can redistribute it and/or modify it
58 ;;; under the terms of the GNU General Public License as published by the
59 ;;; Free Software Foundation; either version 2 of the License, or (at
60 ;;; your option) any later version.
61 ;;;
62 ;;; This library is distributed in the hope that it will be useful, but
63 ;;; WITHOUT ANY WARRANTY; without even the implied warranty of
64 ;;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
65 ;;; Library General Public License for more details.
66 ;;;
67 ;;; You should have received a copy of the GNU General Public License along
68 ;;; with this library; if not, write to the Free Software
69 ;;; Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
70 ;;;
71 ;;; Copyright (C) 2008 Dieter Kaiser
72 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
74 (in-package :maxima)
76 (declare-top (special $factlim $gamma_expand))
78 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
80 (defmvar $factorial_expand nil)
81 (defmvar $beta_expand nil)
83 (defmvar $erf_representation nil
84 "When T erfc, erfi and erf_generalized are transformed to erf.")
86 (defmvar $erf_%iargs nil
87 "When T erf and erfi simplifies for an imaginary argument.")
89 (defmvar $hypergeometric_representation nil
90 "When T a transformation to a hypergeometric representation is done.")
92 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
93 ;;; The following functions test if numerical evaluation has to be done.
94 ;;; The functions should help to test for numerical evaluation more consitent
95 ;;; and without complicated conditional tests including more than one or two
96 ;;; arguments.
97 ;;;
98 ;;; The functions take a list of arguments. All arguments have to be a CL or
99 ;;; Maxima number. If all arguments are numbers we have two cases:
100 ;;; 1. $numer is T we return T. The function has to be evaluated numerically.
101 ;;; 2. One of the args is a float or a bigfloat. Evaluate numerically.
102 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
104 ;;; Test for numerically evaluation in float precision
106 (defun float-numerical-eval-p (&rest args)
107 (let ((flag nil))
108 (dolist (ll args)
109 (when (not (float-or-rational-p ll))
110 (return-from float-numerical-eval-p nil))
111 (when (floatp ll) (setq flag t)))
112 (if (or $numer flag) t nil)))
114 ;;; Test for numerically evaluation in complex float precision
116 (defun complex-float-numerical-eval-p (&rest args)
117 "Determine if ARGS consists of numerical values by determining if
118 the real and imaginary parts of each arg are nuemrical (but not
119 bigfloats). A non-NIL result is returned if at least one of args is
120 a floating-point value or if numer is true. If the result is
121 non-NIL, it is a list of the arguments reduced via COMPLEX-NUMBER-P"
122 (let (flag values)
123 (dolist (ll args)
124 (multiple-value-bind (bool rll ill)
125 (complex-number-p ll 'float-or-rational-p)
126 (unless bool
127 (return-from complex-float-numerical-eval-p nil))
128 ;; Always save the result from complex-number-p. But for backward
129 ;; compatibility, only set the flag if any item is a float.
130 (push (add rll (mul ill '$%i)) values)
131 (setf flag (or flag (or (floatp rll) (floatp ill))))))
132 (when (or $numer flag)
133 ;; Return the values in the same order as the args!
134 (nreverse values))))
136 ;;; Test for numerically evaluation in bigfloat precision
138 (defun bigfloat-numerical-eval-p (&rest args)
139 (let ((flag nil))
140 (dolist (ll args)
141 (when (not (bigfloat-or-number-p ll))
142 (return-from bigfloat-numerical-eval-p nil))
143 (when ($bfloatp ll)
144 (setq flag t)))
145 (if (or $numer flag) t nil)))
147 ;;; Test for numerically evaluation in complex bigfloat precision
149 (defun complex-bigfloat-numerical-eval-p (&rest args)
150 "Determine if ARGS consists of numerical values by determining if
151 the real and imaginary parts of each arg are nuemrical (including
152 bigfloats). A non-NIL result is returned if at least one of args is
153 a floating-point value or if numer is true. If the result is
154 non-NIL, it is a list of the arguments reduced via COMPLEX-NUMBER-P."
156 (let (flag values)
157 (dolist (ll args)
158 (multiple-value-bind (bool rll ill)
159 (complex-number-p ll 'bigfloat-or-number-p)
160 (unless bool
161 (return-from complex-bigfloat-numerical-eval-p nil))
162 ;; Always save the result from complex-number-p. But for backward
163 ;; compatibility, only set the flag if any item is a bfloat.
164 (push (add rll (mul ill '$%i)) values)
165 (when (or ($bfloatp rll) ($bfloatp ill))
166 (setf flag t))))
167 (when (or $numer flag)
168 ;; Return the values in the same order as the args!
169 (nreverse values))))
171 ;;; Test for numerical evaluation in any precision, real or complex.
172 (defun numerical-eval-p (&rest args)
173 (or (apply 'float-numerical-eval-p args)
174 (apply 'complex-float-numerical-eval-p args)
175 (apply 'bigfloat-numerical-eval-p args)
176 (apply 'complex-bigfloat-numerical-eval-p args)))
178 ;;; Check for an integer or a float or bigfloat representation. When we
179 ;;; have a float or bigfloat representation return the integer value.
181 (defun integer-representation-p (x)
182 (let ((val nil))
183 (cond ((integerp x) x)
184 ((and (floatp x) (= 0 (nth-value 1 (truncate x))))
185 (nth-value 0 (truncate x)))
186 ((and ($bfloatp x)
187 (eq ($sign (sub (setq val ($truncate x)) x)) '$zero))
188 val)
189 (t nil))))
191 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
193 ;;; The changes to the parser to connect the operator !! to double_factorial(z)
195 ;(def-mheader |$!!| (%double_factorial))
197 ;(def-led (|$!!| 160.) (op left)
198 ; (list '$expr
199 ; (mheader '$!!)
200 ; (convert left '$expr)))
202 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
204 ;;; The implementation of the function Double factorial
206 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
208 (defmfun $double_factorial (z)
209 (simplify (list '(%double_factorial) z)))
211 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
213 ;;; Set properties to give full support to the parser and display
215 (defprop $double_factorial %double_factorial alias)
216 (defprop $double_factorial %double_factorial verb)
218 (defprop %double_factorial $double_factorial reversealias)
219 (defprop %double_factorial $double_factorial noun)
221 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
223 ;;; Double factorial is a simplifying function
225 (defprop %double_factorial simp-double-factorial operators)
227 ;;; Double factorial distributes over bags
229 (defprop %double_factorial (mlist $matrix mequal) distribute_over)
231 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
233 ;;; Double factorial has mirror symmetry
235 (defprop %double_factorial t commutes-with-conjugate)
237 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
239 ;;; Differentiation of Double factorial
241 (defprop %double_factorial
242 ((z)
243 ((mtimes)
244 ((rat) 1 2)
245 ((%double_factorial) z)
246 ((mplus)
247 ((%log) 2)
248 ((mqapply)
249 (($psi array) 0)
250 ((mplus) 1 ((mtimes) ((rat) 1 2) z)))
251 ((mtimes)
252 ((rat) 1 2) $%pi
253 ((%log) ((mtimes) 2 ((mexpt) $%pi -1)))
254 ((%sin) ((mtimes) $%pi z))))))
255 grad)
257 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
259 (defun simp-double-factorial (expr z simpflag)
260 (oneargcheck expr)
261 (setq z (simpcheck (cadr expr) simpflag))
262 (cond
263 ((and (fixnump z) (> z -1) (or (minusp $factlim) (< z $factlim)))
264 ;; Positive Integer less then $factlim or $factlim is -1. Call gfact.
265 (gfact z (floor (/ z 2)) 2))
267 ((and (mnump z)
268 (eq ($sign z) '$neg)
269 (zerop1 (sub (simplify (list '(%truncate) (div z 2))) (div z 2))))
270 ;; Even negative integer or real representation. Not defined.
271 (simp-domain-error
272 (intl:gettext
273 "double_factorial: double_factorial(~:M) is undefined.") z))
275 ((or (integerp z) ; at this point odd negative integer. Evaluate.
276 (complex-float-numerical-eval-p z))
277 (cond
278 ((and (integerp z) (= z -1)) 1) ; Special cases -1 and -3
279 ((and (integerp z) (= z -3)) -1)
281 ;; Odd negative integer, float or complex float.
282 (complexify
283 (double-factorial
284 (complex ($float ($realpart z)) ($float ($imagpart z))))))))
286 ((and (not (ratnump z))
287 (complex-bigfloat-numerical-eval-p z))
288 ;; bigfloat or complex bigfloat.
289 (bfloat-double-factorial
290 (add ($bfloat ($realpart z)) (mul '$%i ($bfloat ($imagpart z))))))
292 ;; double_factorial(inf) -> inf
293 ((eq z '$inf) '$inf)
295 ((and $factorial_expand
296 (mplusp z)
297 (integerp (cadr z)))
298 (let ((k (cadr z))
299 (n (simplify (cons '(mplus) (cddr z)))))
300 (cond
301 ((= k -1)
302 ;; Special case double_factorial(n-1)
303 ;; Not sure if this simplification is useful.
304 (div (simplify (list '(mfactorial) n))
305 (simplify (list '(%double_factorial) n))))
306 ((= k (* 2 (truncate (/ k 2))))
307 ;; Special case double_factorial(2*k+n), k integer
308 (setq k (/ k 2))
309 ($factor ; we get more simple expression when factoring
310 (mul
311 (power 2 k)
312 (simplify (list '($pochhammer) (add (div n 2) 1) k))
313 (simplify (list '(%double_factorial) n)))))
315 (eqtest (list '(%double_factorial) z) expr)))))
318 (eqtest (list '(%double_factorial) z) expr))))
320 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
321 ;;; Double factorial for a complex float argument. The result is a CL complex.
322 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
324 (defun double-factorial (z)
325 (let ((pival (float pi)))
327 (expt
328 (/ 2 pival)
329 (/ (- 1 (cos (* pival z))) 4))
330 (expt 2e0 (/ z 2))
331 (gamma-lanczos (+ 1 (/ z 2))))))
333 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
334 ;;; Double factorial for a bigfloat or complex bigfloat argument
335 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
337 (defun bfloat-double-factorial (z)
338 (let* ((pival ($bfloat '$%pi))
339 (bigfloat1 ($bfloat bigfloatone))
340 (bigfloat2 (add bigfloat1 bigfloat1))
341 (bigfloat4 (add bigfloat2 bigfloat2))
342 ($ratprint nil))
343 (cmul
344 (cpower
345 (cdiv bigfloat2 pival)
346 (cdiv (sub bigfloat1
347 (simplify (list '(%cos) (cmul pival z)))) bigfloat4))
348 (cmul
349 (cpower bigfloat2 (cdiv z bigfloat2))
350 (simplify (list '(%gamma) (add bigfloat1 (cdiv z bigfloat2))))))))
352 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
354 ;;; The implementation of the Incomplete Gamma function
356 ;;; A&S 6.5.3:
358 ;;; inf
359 ;;; /
360 ;;; [ a - 1 - t
361 ;;; gamma_incomplete(a, z) = I t %e dt
362 ;;; ]
363 ;;; /
364 ;;; z
366 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
368 (defvar *debug-gamma* nil)
370 (defmfun $gamma_incomplete (a z)
371 (simplify (list '(%gamma_incomplete) a z)))
373 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
375 ;;; Set properties to give full support to the parser and display
377 (defprop $gamma_incomplete %gamma_incomplete alias)
378 (defprop $gamma_incomplete %gamma_incomplete verb)
380 (defprop %gamma_incomplete $gamma_incomplete reversealias)
381 (defprop %gamma_incomplete $gamma_incomplete noun)
383 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
385 ;;; Incomplete Gamma function is a simplifying function
387 (defprop %gamma_incomplete simp-gamma-incomplete operators)
389 ;;; Incomplete Gamma distributes over bags
391 (defprop %gamma_incomplete (mlist $matrix mequal) distribute_over)
393 ;;; Incomplete Gamma function has not mirror symmetry for z on the negative
394 ;;; real axis. We support a conjugate-function which test this case.
396 (defprop %gamma_incomplete conjugate-gamma-incomplete conjugate-function)
398 (defun conjugate-gamma-incomplete (args)
399 (let ((a (first args)) (z (second args)))
400 (cond ((off-negative-real-axisp z)
401 ;; Definitely not on the negative real axis for z. Mirror symmetry.
402 (simplify
403 (list
404 '(%gamma_incomplete)
405 (simplify (list '($conjugate) a))
406 (simplify (list '($conjugate) z)))))
408 ;; On the negative real axis or no information. Unsimplified.
409 (list
410 '($conjugate simp)
411 (simplify (list '(%gamma_incomplete) a z)))))))
413 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
415 ;;; Derivative of the Incomplete Gamma function
417 (putprop '%gamma_incomplete
418 `((a z)
419 ,(lambda (a z)
420 (cond ((member ($sign a) '($pos $pz))
421 ;; The derivative wrt a in terms of hypergeometric_regularized 2F2
422 ;; function and the Generalized Incomplete Gamma function
423 ;; (functions.wolfram.com), only for a>0.
424 `((mplus)
425 ((mtimes)
426 ((mexpt) ((%gamma) ,a) 2)
427 ((mexpt) ,z ,a)
428 (($hypergeometric_regularized)
429 ((mlist) ,a ,a)
430 ((mlist) ((mplus) 1 ,a) ((mplus) 1 ,a))
431 ((mtimes) -1 ,z)))
432 ((mtimes) -1
433 ((%gamma_incomplete_generalized) ,a 0 ,z)
434 ((%log) ,z))
435 ((mtimes)
436 ((%gamma) ,a)
437 ((mqapply) (($psi array) 0) ,a))))
439 ;; No derivative. Maxima generates a noun form.
440 nil)))
441 ;; The derivative wrt z
442 ((mtimes) -1
443 ((mexpt) $%e ((mtimes) -1 z))
444 ((mexpt) z ((mplus) -1 a))))
445 'grad)
447 ;;; Integral of the Incomplete Gamma function
449 (defprop %gamma_incomplete
450 ((a z)
452 ((mplus)
453 ((mtimes) -1 ((%gamma_incomplete) ((mplus) 1 a) z))
454 ((mtimes) ((%gamma_incomplete) a z) z)))
455 integral)
457 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
459 ;;; We support a simplim%function. The function is looked up in simplimit and
460 ;;; handles specific values of the function.
462 (defprop %gamma_incomplete simplim%gamma_incomplete simplim%function)
464 (defun simplim%gamma_incomplete (expr var val)
465 ;; Look for the limit of the arguments.
466 (let ((a (limit (cadr expr) var val 'think))
467 (z (limit (caddr expr) var val 'think)))
468 (cond
470 ((eq z '$infinity) ;; http://dlmf.nist.gov/8.11#i
471 (cond ((and (zerop1 ($realpart (caddr expr)))
472 (eq ($csign (m+ -1 (cadr expr))) '$neg))
474 (t (throw 'limit t))))
476 ;; Handle an argument 0 at this place.
477 ((or (zerop1 z)
478 (eq z '$zeroa)
479 (eq z '$zerob))
480 (let ((sgn ($sign ($realpart a))))
481 (cond ((zerop1 a) '$inf)
482 ((member sgn '($neg $nz)) '$infinity)
483 ;; Call the simplifier of the function.
484 (t (simplify (list '(%gamma_incomplete) a z))))))
486 ;; All other cases are handled by the simplifier of the function.
487 (simplify (list '(%gamma_incomplete) a z))))))
489 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
491 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
493 ;;; Implementation of the Lower Incomplete gamma function:
495 ;;; A&S 6.5.2
497 ;;; z
498 ;;; /
499 ;;; [ a - 1 - t
500 ;;; gamma_incomplete_lower(a, z) = I t %e dt
501 ;;; ]
502 ;;; /
503 ;;; 0
504 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
505 (defmfun $gamma_incomplete_lower (a z)
506 (simplify (list '(%gamma_incomplete_lower) a z)))
508 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
510 (defprop $gamma_incomplete_lower %gamma_incomplete_lower alias)
511 (defprop $gamma_incomplete_lower %gamma_incomplete_lower verb)
513 (defprop %gamma_incomplete_lower $gamma_incomplete_lower reversealias)
514 (defprop %gamma_incomplete_lower $gamma_incomplete_lower noun)
516 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
518 (defprop %gamma_incomplete_lower simp-gamma-incomplete-lower operators)
520 ;;; distribute over bags (aggregates)
522 (defprop %gamma_incomplete_lower (mlist $matrix mequal) distribute_over)
524 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
526 ;; (defprop %gamma_incomplete_lower ??? grad) WHAT TO PUT HERE ??
529 ;; Handles some special cases for the order a and simplifies it to an
530 ;; equivalent form, possibly involving erf and gamma_incomplete_lower
531 ;; to a lower order.
532 (defun simp-gamma-incomplete-lower (expr ignored simpflag)
533 (declare (ignore ignored))
534 (twoargcheck expr)
535 (let ((a (simpcheck (cadr expr) simpflag))
536 (z (simpcheck (caddr expr) simpflag)))
537 (cond
538 ((or
539 (float-numerical-eval-p a z)
540 (complex-float-numerical-eval-p a z)
541 (bigfloat-numerical-eval-p a z)
542 (complex-bigfloat-numerical-eval-p a z))
543 (take '(%gamma_incomplete_generalized) a 0 z))
544 ((gamma-incomplete-lower-expand a z))
546 (eqtest (list '(%gamma_incomplete_lower) a z) expr)))))
548 ;; Try to express gamma_incomplete_lower(a,z) in simpler terms for
549 ;; special values of the order a. If we can't, return NIL to say so.
550 (defun gamma-incomplete-lower-expand (a z)
551 (cond ((and $gamma_expand (integerp a) (eql a 1))
552 ;; gamma_incomplete_lower(0, x) = 1-exp(x)
553 (sub 1 (m^t '$%e (neg z))))
554 ((and $gamma_expand (integerp a) (plusp a))
555 ;; gamma_incomplete_lower(a,z) can be simplified if a is a
556 ;; positive integer. Since gamma_incomplete_lower(a,z) =
557 ;; gamma(a) - gamma_incomplete(a,z), use gamma_incomplete to
558 ;; get the result.
559 (sub (simpgamma (list '(%gamma) a) 1 nil)
560 (take '(%gamma_incomplete) a z)))
561 ((and $gamma_expand (alike1 a 1//2))
562 ;; A&S 6.5.12:
564 ;; gamma_incomplete_lower(1/2,x^2) = sqrt(%pi)*erf(x)
566 ;; gamma_incomplete_lower(1/2,z) = sqrt(%pi)*erf(sqrt(x))
568 (mul (power '$%pi '((rat simp) 1 2))
569 (take '(%erf) (power z '((rat simp) 1 2)))))
570 ((and $gamma_expand (mplusp a) (integerp (cadr a)))
571 ;; gamma_incomplete_lower(a+n,z), where n is an integer
572 (let ((n (cadr a))
573 (a (simplify (cons '(mplus) (cddr a)))))
574 (cond
575 ((> n 0)
576 ;; gamma_incomplete_lower(a+n,z). See DLMF 8.8.7:
578 ;; gamma_incomplete_lower(a+n,z)
579 ;; = pochhammer(a,n)*gamma_incomplete_lower(a,z)
580 ;; -z^a*exp(-z)*sum(gamma(a+n)gamma(a+k+1)*z^k,k,0,n-1)
581 (sub
582 (mul
583 (simplify (list '($pochhammer) a n))
584 (simplify (list '(%gamma_incomplete_lower) a z)))
585 (mul
586 (power z a)
587 (power '$%e (mul -1 z))
588 (let ((gamma-a+n (simpgamma (list '(%gamma) (add a n)) 1 nil))
589 (index (gensumindex))
590 ($simpsum t))
591 (simpsum1
592 (mul
593 (div gamma-a+n
594 (simpgamma (list '(%gamma) (add a index 1)) 1 nil))
595 (power z index))
596 index 0 (add n -1))))))
597 ((< n 0)
598 (setq n (- n))
599 ;; See DLMF 8.8.8. For simplicity let g(a,z) = gamma_incomplete_lower(a,z).
601 ;; g(a,z) = gamma(a)/gamma(a-n)*g(a-n,z)
602 ;; - z^(a-1)*exp(z)*sum(gamma(a)/gamma(a-k)*z^(-k),k,0,n-1)
604 ;; Rewrite:
605 ;; g(a-n,z) = gamma(a-n)/gamma(a)*g(a,z)
606 ;; + gamma(a-n)/gamma(a)*z^(a-1)*exp(-z)
607 ;; * sum(gamma(a)/gamma(a-k)*z^(-k),k,0,n-1)
608 ;; Or
609 ;; g(a-n,z) = gamma(a-n)/gamma(a)*g(a,z)
610 ;; + z^(a-1)*exp(-z)
611 ;; * sum(gamma(a-n)/gamma(a-k)*z^(-k),k,0,n-1)
612 (let ((gamma-a-n (simpgamma (list '(%gamma) (sub a n)) 1 nil))
613 (index (gensumindex))
614 ($simpsum t))
615 (add
616 (mul
617 (div gamma-a-n
618 (list '(%gamma) a))
619 (simplify (list '(%gamma_incomplete_lower) a z)))
620 (mul
621 (power z (sub a 1))
622 (power '$%e (mul -1 z))
623 (simpsum1
624 (mul
625 (div gamma-a-n
626 (simpgamma (list '(%gamma) (sub a index)) 1 nil))
627 (power z (mul -1 index)))
628 index 0 (add n -1)))))))))
629 ((and $gamma_expand (consp a) (eq 'rat (caar a))
630 (integerp (second a))
631 (integerp (third a)))
632 ;; gamma_incomplete_lower of (numeric) rational order.
633 ;; Expand it out so that the resulting order is between 0 and
634 ;; 1.
635 (multiple-value-bind (n order)
636 (floor (/ (second a) (third a)))
637 ;; a = n + order where 0 <= order < 1.
638 (let ((rat-order (rat (numerator order) (denominator order))))
639 (cond
640 ((zerop n)
641 ;; Nothing to do if the order is already between 0 and 1
642 (list '(%gamma_incomplete_lower simp) a z))
644 ;; Use gamma_incomplete(a+n,z) above. and then substitue
645 ;; a=order. This works for n positive or negative.
646 (let* ((ord (gensym))
647 (g (simplify (list '(%gamma_incomplete_lower) (add ord n) z))))
648 ($substitute rat-order ord g)))))))
650 ;; No expansion so return nil to indicate that
651 nil)))
653 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
655 (defun simp-gamma-incomplete (expr ignored simpflag)
656 (declare (ignore ignored))
657 (twoargcheck expr)
658 (let ((a (simpcheck (cadr expr) simpflag))
659 (z (simpcheck (caddr expr) simpflag))
660 ($simpsum t)
661 (ratorder))
662 (cond
664 ;; Check for specific values
666 ((zerop1 z)
667 ;; gamma_incomplete(v,0) is gamma(v) only if the realpart(v) >
668 ;; 0. If realpart(v) <= 0, gamma_incomplete is undefined. For
669 ;; all other cases, return the noun form.
670 (let ((sgn ($sign ($realpart a))))
671 (cond ((member sgn '($neg $zero))
672 (simp-domain-error
673 (intl:gettext
674 "gamma_incomplete: gamma_incomplete(~:M,~:M) is undefined.")
675 a z))
676 ((member sgn '($pos $pz)) ($gamma a))
677 (t (eqtest (list '(%gamma_incomplete) a z) expr)))))
679 ((eq z '$inf) 0)
680 ((and (eq z '$minf)
681 (eql a 0))
682 '$infinity)
684 ;; Check for numerical evaluation in Float or Bigfloat precision
686 ((float-numerical-eval-p a z)
687 (when *debug-gamma*
688 (format t "~&SIMP-GAMMA-INCOMPLETE in float-numerical-eval-p~%"))
689 ;; a and z are Maxima numbers, at least one has a float value
690 (let ((a ($float a))
691 (z ($float z)))
692 (cond
693 ((or (= a 0.0)
694 (and (= 0 (- a (truncate a)))
695 (< a 0.0)))
696 ;; a is zero or a negative float representing an integer.
697 ;; For these cases the numerical routines of gamma-incomplete
698 ;; do not work. Call the numerical routine for the Exponential
699 ;; Integral E(n,z). The routine is called with a positive integer!.
700 (setq a (truncate a))
701 (complexify (* (expt z a) (expintegral-e (- 1 a) z))))
703 (complexify (gamma-incomplete a z))))))
705 ((complex-float-numerical-eval-p a z)
706 (when *debug-gamma*
707 (format t
708 "~&SIMP-GAMMA-INCOMPLETE in complex-float-numerical-eval-p~%"))
709 ;; a and z are Maxima numbers, at least one is a complex value and
710 ;; we have at least one float part
711 (let ((ca (complex ($float ($realpart a)) ($float ($imagpart a))))
712 (cz (complex ($float ($realpart z)) ($float ($imagpart z)))))
713 (cond
714 ((and (= (imagpart ca) 0.0)
715 (or (= (realpart ca) 0.0)
716 (and (= 0 (- (realpart ca) (truncate (realpart ca))))
717 (< (realpart ca) 0.0))))
718 ;; Call expintegral-e. See comment above.
719 (setq ca (truncate (realpart ca)))
720 (complexify (* (expt cz ca) (expintegral-e (- 1 ca) cz))))
722 (complexify (gamma-incomplete ca cz))))))
724 ((bigfloat-numerical-eval-p a z)
725 (when *debug-gamma*
726 (format t "~&SIMP-GAMMA-INCOMPLETE in bigfloat-numerical-eval-p~%"))
727 (let ((a ($bfloat a))
728 (z ($bfloat z)))
729 (cond
730 ((or (eq ($sign a) '$zero)
731 (and (eq ($sign (sub a ($truncate a))) '$zero)
732 (eq ($sign a) '$neg)))
733 ;; Call bfloat-expintegral-e. See comment above.
734 (setq a ($truncate a))
735 ($rectform (mul (power z a) (bfloat-expintegral-e (- 1 a) z))))
737 (bfloat-gamma-incomplete a z)))))
739 ((complex-bigfloat-numerical-eval-p a z)
740 (when *debug-gamma*
741 (format t
742 "~&SIMP-GAMMA-INCOMPLETE in complex-bigfloat-numerical-eval-p~%"))
743 (let ((ca (add ($bfloat ($realpart a))
744 (mul '$%i ($bfloat ($imagpart a)))))
745 (cz (add ($bfloat ($realpart z))
746 (mul '$%i ($bfloat ($imagpart z))))))
747 (cond
748 ((and (eq ($sign ($imagpart ca)) '$zero)
749 (or (eq ($sign ($realpart ca)) '$zero)
750 (and (eq ($sign (sub ($realpart ca)
751 ($truncate ($realpart ca))))
752 '$zero)
753 (eq ($sign ($realpart ca)) '$neg))))
754 ;; Call bfloat-expintegral-e. See comment above.
755 (when *debug-gamma*
756 (format t
757 "~& bigfloat-numerical-eval-p calls bfloat-expintegral-e~%"))
758 (setq a ($truncate ($realpart a)))
759 ($rectform (mul (power cz a)
760 (bfloat-expintegral-e (- 1 a) cz))))
762 (complex-bfloat-gamma-incomplete ca cz)))))
764 ;; Check for transformations and argument simplification
766 ((and $gamma_expand (integerp a))
767 ;; Integer or a symbol declared to be an integer. Expand in a series.
768 (let ((sgn ($sign a)))
769 (cond
770 ((eq sgn '$zero)
771 (add
772 (mul -1
773 (simplify (list '(%expintegral_ei) (mul -1 z))))
774 (mul
775 '((rat simp) 1 2)
776 (sub
777 (simplify (list '(%log) (mul -1 z)))
778 (simplify (list '(%log) (div -1 z)))))
779 (mul -1 (simplify (list '(%log) z)))))
780 ((member sgn '($pos $pz))
781 (mul
782 (simplify (list '(%gamma) a))
783 (power '$%e (mul -1 z))
784 (let ((index (gensumindex)))
785 (simpsum1
786 (div
787 (power z index)
788 (let (($gamma_expand nil))
789 ;; Simplify gamma, but do not expand to avoid division
790 ;; by zero.
791 (simplify (list '(%gamma) (add index 1)))))
792 index 0 (sub a 1)))))
793 ((member sgn '($neg $nz))
794 (sub
795 (mul
796 (div
797 (power -1 (add (mul -1 a) -1))
798 (simplify (list '(%gamma) (add (mul -1 a) 1))))
799 (add
800 (simplify (list '(%expintegral_ei) (mul -1 z)))
801 (mul
802 '((rat simp) -1 2)
803 (sub
804 (simplify (list '(%log) (mul -1 z)))
805 (simplify (list '(%log) (div -1 z)))))
806 (simplify (list '(%log) z))))
807 (mul
808 (power '$%e (mul -1 z))
809 (let ((index (gensumindex)))
810 (simpsum1
811 (div
812 (power z (add index a -1))
813 (simplify (list '($pochhammer) a index)))
814 index 1 (mul -1 a))))))
815 (t (eqtest (list '(%gamma_incomplete) a z) expr)))))
817 ((and $gamma_expand (setq ratorder (max-numeric-ratio-p a 2)))
818 ;; We have a half integral order and $gamma_expand is not NIL.
819 ;; We expand in a series with the Erfc function
820 (setq ratorder (- ratorder (/ 1 2)))
821 (cond
822 ((equal ratorder 0)
823 (mul
824 (power '$%pi '((rat simp) 1 2))
825 (simplify (list '(%erfc) (power z '((rat simp) 1 2))))))
826 ((> ratorder 0)
827 (sub
828 (mul
829 (simplify (list '(%gamma) a))
830 (simplify (list '(%erfc) (power z '((rat simp) 1 2)))))
831 (mul
832 (power -1 (sub ratorder 1))
833 (power '$%e (mul -1 z))
834 (power z '((rat simp) 1 2))
835 (let ((index (gensumindex)))
836 (simpsum1
837 (mul -1 ; we get more simple results
838 (simplify ; when multiplying with -1
839 (list
840 '($pochhammer)
841 (sub '((rat simp) 1 2) ratorder)
842 (sub ratorder (add index 1))))
843 (power (mul -1 z) index))
844 index 0 (sub ratorder 1))))))
845 ((< ratorder 0)
846 (setq ratorder (- ratorder))
847 (sub
848 (div
849 (mul
850 (power -1 ratorder)
851 (power '$%pi '((rat simp) 1 2))
852 (simplify (list '(%erfc) (power z '((rat simp) 1 2)))))
853 (simplify (list '($pochhammer) '((rat simp) 1 2) ratorder)))
854 (mul
855 (power z (sub '((rat simp) 1 2) ratorder))
856 (power '$%e (mul -1 z))
857 (let ((index (gensumindex)))
858 (simpsum1
859 (div
860 (power z index)
861 (simplify
862 (list
863 '($pochhammer)
864 (sub '((rat simp) 1 2) ratorder)
865 (add index 1))))
866 index 0 (sub ratorder 1))))))))
868 ((and $gamma_expand (mplusp a) (integerp (cadr a)))
869 ;; Handle gamma_incomplete(a+n, z), where n is a numerical integer
870 (let ((n (cadr a))
871 (a (simplify (cons '(mplus) (cddr a)))))
872 (cond
873 ((> n 0)
874 ;; See DLMF 8.8.9: https://dlmf.nist.gov/8.8.E9
876 ;; gamma_incomplete(a+n,z) = pochhammer(a,n)*gamma_incomplete(a,z)
877 ;; + z^a*exp(-z)*sum(gamma(a+n)/gamma(a+k+1)*z^k,k,0,n-1)
878 (add
879 (mul
880 (simplify (list '($pochhammer) a n))
881 (simplify (list '(%gamma_incomplete) a z)))
882 (mul
883 (power '$%e (mul -1 z))
884 (power z a)
885 (let ((gamma-a+n (simpgamma (list '(%gamma) (add a n)) 1 nil))
886 (index (gensumindex)))
887 (simpsum1
888 (mul
889 (div gamma-a+n
890 (simpgamma (list '(%gamma) (add a index 1)) 1 nil))
891 (power z index))
892 index 0 (add n -1))))))
893 ((< n 0)
894 (setq n (- n))
895 ;; See http://functions.wolfram.com/06.06.17.0004.01
897 ;; gamma_incomplate(a,z) =
898 ;; (-1)^n*pochhammer(1-a,n)
899 ;; *[gamma_incomplete(a-n,z)
900 ;; + z^(a-n-1)*exp(-z)*sum(z^k/pochhammer(a-n,k),k,1,n)]
902 ;; Rerarrange this in terms of gamma_incomplete(a-n,z):
904 ;; gamma_incomplete(a-n,z) =
905 ;; (-1)^n*gamma_incomplete(a,z)/pochhammer(1-a,n)
906 ;; -z^(a-n-1)*exp(-z)*sum(z^k/pochhammer(a-n,k),k,1,n)
908 ;; Change the summation index to go from k = 0 to n-1:
910 ;; z^(a-n-1)*sum(z^k/pochhammer(a-n,k),k,1,n)
911 ;; = z^(a-n-1)*sum(z^(k+1)/pochhammer(a-n,k+1),k,0,n-1)
912 ;; = z^(a-n)*sum(z^k/pochhammer(a-n,k+1),k,0,n-1)
914 ;; Thuus:
915 ;; gamma_incomplete(a-n,z) =
916 ;; (-1)^n*gamma_incomplete(a,z)/pochhammer(1-a,n)
917 ;; -z^(a-n)*sum(z^k/pochhammer(a-n,k+1),k,0,n-1)
918 (sub
919 ;; (-1)^n*gamma_incomplete(a,z)/pochhammer(1-a,n)
920 (div
921 (mul
922 (power -1 n)
923 (simplify (list '(%gamma_incomplete) a z)))
924 (simplify (list '($pochhammer) (sub 1 a) n)))
925 (mul
926 (power '$%e (mul -1 z))
927 (power z (sub a n))
928 ;; sum(z^k/pochhammer(a-n,k+1),k,0,n-1)
929 (let ((index (gensumindex)))
930 (simpsum1
931 (div
932 (power z index)
933 (simplify (list '($pochhammer) (sub a n) (add index 1))))
934 index 0 (sub n 1)))))))))
935 ((and $gamma_expand (consp a) (eq 'rat (caar a))
936 (integerp (second a))
937 (integerp (third a)))
938 ;; gamma_incomplete of (numeric) rational order. Expand it out
939 ;; so that the resulting order is between 0 and 1.
940 (multiple-value-bind (n order)
941 (floor (/ (second a) (third a)))
942 ;; a = n + order where 0 <= order < 1.
943 (let ((rat-order (rat (numerator order) (denominator order))))
944 (cond
945 ((zerop n)
946 ;; Nothing to do if the order is already between 0 and 1
947 (eqtest (list '(%gamma_incomplete) a z) expr))
949 ;; Use gamma_incomplete(a+n,z) above. and then substitue
950 ;; a=order. This works for n positive or negative.
951 (let* ((ord (gensym))
952 (g (simplify (list '(%gamma_incomplete) (add ord n) z))))
953 ($substitute rat-order ord g)))))))
955 (t (eqtest (list '(%gamma_incomplete) a z) expr)))))
957 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
958 ;;; Numerical evaluation of the Incomplete Gamma function
960 ;;; gamma-incomplete (a,z) - real and complex double float
961 ;;; bfloat-gamma-incomplete (a z) - bigfloat
962 ;;; complex-bfloat-gamma-incomplete (a z) - complex bigfloat
964 ;;; Expansion in a power series for abs(x) < R, where R is
965 ;;; *gamma-radius* + real(a) if real(a) > 0 or *gamma-radius*
966 ;;; otherwise.
968 ;;; (A&S 6.5.29):
970 ;;; inf
971 ;;; ===
972 ;;; \ gamma(a)
973 ;;; gamma(a,z) = exp(-x)*z^a * > ------------ * z^n
974 ;;; / gamma(a+1+n)
975 ;;; ===
976 ;;; n=0
978 ;;; This expansion does not work for an integer a<=0, because the Gamma function
979 ;;; in the denominator is not defined for a=0 and negative integers. For this
980 ;;; case we use the Exponential Integral E for numerically evaluation. The
981 ;;; Incomplete Gamma function and the Exponential integral are connected by
983 ;;; gamma(a,z) = z^a * expintegral_e(1-a,z)
985 ;;; When the series is not used, two forms of the continued fraction
986 ;;; are used. When z is not near the negative real axis use the
987 ;;; continued fractions (A&S 6.5.31):
989 ;;; 1 1-a 1 2-a 2
990 ;;; gamma(a,z) = exp(-z) z^a *( -- --- --- --- --- ... )
991 ;;; z+ 1+ z+ 1+ z+
993 ;;; The accuracy is controlled by *gamma-incomplete-eps* for double float
994 ;;; precision. For bigfloat precision epsilon is 10^(-$fpprec). The expansions
995 ;;; in a power series or continued fractions stops if *gamma-incomplete-maxit*
996 ;;; is exceeded and an Maxima error is thrown.
998 ;;; The above fraction does not converge on the negative real axis and
999 ;;; converges very slowly near the axis. In this case, use the
1000 ;;; relationship
1002 ;;; gamma(a,z) = gamma(a) - gamma_lower(a,z)
1004 ;;; The continued fraction for gamma_incomplete_lower(a,z) is
1005 ;;; (http://functions.wolfram.com/06.06.10.0009.01):
1007 ;;; gamma_lower(a,z) = exp(-z) * z^a / cf(a,z)
1009 ;;; where
1011 ;;; -a*z z (a+1)*z 2*z (a+2)*z
1012 ;;; cf(a,z) = a + ---- ---- ------- ---- -------
1013 ;;; a+1+ a+2- a+3+ a+4- a+5+
1015 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1017 (defvar *gamma-incomplete-maxit* 10000)
1018 (defvar *gamma-incomplete-eps* (* 2 flonum-epsilon))
1019 (defvar *gamma-incomplete-min* 1.0e-32)
1021 (defvar *gamma-radius* 1.0
1022 "Controls the radius within a series expansion is done.")
1023 (defvar *gamma-imag* 1.0)
1025 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1027 ;;; The numerical evaluation for CL float or complex values a and x
1028 ;;; When the flag regularized is T, the result is divided by gamma(a) and
1029 ;;; Maxima returns the numercial result for gamma_incomplete_regularized
1031 (defun gamma-incomplete (a x &optional (regularized nil))
1032 (setq x (+ x (cond ((complexp x) #C(0.0 0.0)) ((realp x) 0.0))))
1034 (let ((factor
1035 ;; Compute the factor needed to scale the series or continued
1036 ;; fraction. This is x^a*exp(-x) or x^a*exp(-x)/gamma(a)
1037 ;; depending on whether we want a non-regularized or
1038 ;; regularized form. We want to compute the factor carefully
1039 ;; to avoid unnecessary overflow if possible.
1040 (cond (regularized
1041 (or (try-float-computation
1042 #'(lambda ()
1043 ;; gammafloat is more accurate for real
1044 ;; values of a.
1045 (cond ((complexp a)
1046 (/ (* (expt x a) (exp (- x)))
1047 (gamma-lanczos a)))
1049 (/ (* (expt x a) (exp (- x)))
1050 (gammafloat a))))))
1051 ;; Easy way failed. Use logs to compute the
1052 ;; result. This loses some precision, especially
1053 ;; for large values of a and/or x because we use
1054 ;; many bits to hold the exponent part.
1055 (exp (- (* a (log x))
1057 (log-gamma-lanczos (if (complexp a)
1059 (complex a)))))))
1061 (or (try-float-computation
1062 #'(lambda ()
1063 (* (expt x a) (exp (- x)))))
1064 ;; Easy way failed, so use the log form.
1065 (exp (- (* a (log x))
1066 x)))))))
1067 (multiple-value-bind (result lower-incomplete-tail-p)
1068 (%gamma-incomplete a x)
1069 (cond (lower-incomplete-tail-p
1070 ;; %gamma-incomplete compute the lower incomplete gamma
1071 ;; function, so we need to subtract that from gamma(a),
1072 ;; more or less.
1073 (cond (regularized
1074 (- 1 (* result factor)))
1075 ((complexp a)
1076 (- (gamma-lanczos a) (* result factor)))
1078 (- (gammafloat a) (* result factor)))))
1080 ;; Continued fraction used. Just multiply by the factor
1081 ;; to get the final result.
1082 (* factor result))))))
1084 ;; Compute the key part of the gamma incomplete function using either
1085 ;; a series expression or a continued fraction expression. Two values
1086 ;; are returned: the value itself and a boolean, indicating what the
1087 ;; computed value is. If the boolean non-NIL, then the computed value
1088 ;; is the lower incomplete gamma function.
1089 (defun %gamma-incomplete (a x)
1090 (let ((gm-maxit *gamma-incomplete-maxit*)
1091 (gm-eps *gamma-incomplete-eps*)
1092 (gm-min *gamma-incomplete-min*))
1093 (when *debug-gamma*
1094 (format t "~&GAMMA-INCOMPLETE with ~A and ~A~%" a x))
1095 (cond
1096 ;; The series expansion is done for x within a circle of radius
1097 ;; R, where R = *gamma-radius*+(realpart(a)) for realpart(a) > 0
1098 ;; and R = *gamma-radisu* for realpart(a) < 0. Otherwise a
1099 ;; continued fraction is used.
1100 ((and (> (abs x) (+ *gamma-radius*
1101 (if (> (realpart a) 0.0) (realpart a) 0.0))))
1102 (cond ((and (< (realpart x) 0)
1103 (< (abs (imagpart x))
1104 (* *gamma-imag* (abs (realpart x)))))
1105 ;; For x near the negative real axis, use the
1106 ;; relationship gamma_incomplete(a,z) = gamma(a) -
1107 ;; gamma_incomplete_lower(a,z), where
1108 ;; gamma_incomplete_lower(a,z) is the lower poart of the
1109 ;; incomplete gamma function. We can evaluate that
1110 ;; using a continued fraction from
1111 ;; http://functions.wolfram.com/06.06.10.0009.01. (Note
1112 ;; that the alternative fraction,
1113 ;; http://functions.wolfram.com/06.06.10.0007.01,
1114 ;; appears to be less accurate.)
1116 ;; Also note that this appears to be valid for all
1117 ;; values of x (real or complex), but we don't want to
1118 ;; use this everywhere for gamma_incomplete. Consider
1119 ;; what happens for large real x. gamma_incomplete(a,x)
1120 ;; is small, but gamma_incomplete(a,x) = gamma(x) - cf
1121 ;; will have large roundoff errors.
1122 (when *debug-gamma*
1123 (format t "~&GAMMA-INCOMPLETE in continued fractions for lower integral~%"))
1124 (let ((a (bigfloat:to a))
1125 (x (bigfloat:to x))
1126 (bigfloat::*debug-cf-eval* *debug-gamma*)
1127 (bigfloat::*max-cf-iterations* *gamma-incomplete-maxit*))
1128 (values (/ (bigfloat::lentz #'(lambda (n)
1129 (+ n a))
1130 #'(lambda (n)
1131 (if (evenp n)
1132 (* (ash n -1) x)
1133 (- (* (+ a (ash n -1)) x))))))
1134 t)))
1136 ;; Expansion in continued fractions for gamma_incomplete.
1137 (when *debug-gamma*
1138 (format t "~&GAMMA-INCOMPLETE in continued fractions~%"))
1139 (do* ((i 1 (+ i 1))
1140 (an (- a 1.0) (* i (- a i)))
1141 (b (+ 3.0 x (- a)) (+ b 2.0))
1142 (c (/ 1.0 gm-min))
1143 (d (/ 1.0 (- b 2.0)))
1144 (h d)
1145 (del 0.0))
1146 ((> i gm-maxit)
1147 (merror (intl:gettext "gamma_incomplete: continued fractions failed for gamma_incomplete(~:M, ~:M).") a x))
1148 (setq d (+ (* an d) b))
1149 (when (< (abs d) gm-min) (setq d gm-min))
1150 (setq c (+ b (/ an c)))
1151 (when (< (abs c) gm-min) (setq c gm-min))
1152 (setq d (/ 1.0 d))
1153 (setq del (* d c))
1154 (setq h (* h del))
1155 (when (< (abs (- del 1.0)) gm-eps)
1156 ;; Return nil to indicate we used the continued fraction.
1157 (return (values h nil)))))))
1159 ;; Expansion in a series
1160 (when *debug-gamma*
1161 (format t "~&GAMMA-INCOMPLETE in series~%"))
1162 (do* ((i 1 (+ i 1))
1163 (ap a (+ ap 1.0))
1164 (del (/ 1.0 a) (* del (/ x ap)))
1165 (sum del (+ sum del)))
1166 ((> i gm-maxit)
1167 (merror (intl:gettext "gamma_incomplete: series expansion failed for gamma_incomplete(~:M, ~:M).") a x))
1168 (when (< (abs del) (* (abs sum) gm-eps))
1169 (when *debug-gamma* (format t "~&Series converged.~%"))
1170 ;; Return T to indicate we used the series and the series
1171 ;; is for the integral from 0 to x, not x to inf.
1172 (return (values sum t))))))))
1174 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1176 ;;; This function is called for a and x real
1178 (defun bfloat-gamma-incomplete (a x)
1179 (let* ((gm-maxit *gamma-incomplete-maxit*)
1180 (gm-eps (power ($bfloat 10.0) (- $fpprec)))
1181 (gm-min (mul gm-eps gm-eps))
1182 ($ratprint nil))
1183 (cond
1184 ;; The series expansion is done for x within a circle of radius
1185 ;; R, where R = *gamma-radius*+(realpart(a)) for realpart(a) > 0
1186 ;; and R = *gamma-radisu* for realpart(a) < 0. Otherwise a
1187 ;; continued fraction is used.
1188 ((eq ($sign (sub (simplify (list '(mabs) x))
1189 (add *gamma-radius*
1190 (if (eq ($sign a) '$pos) a 0.0))))
1191 '$pos)
1192 (cond
1193 ((and (eq ($sign x) '$pos))
1194 ;; Expansion in continued fractions of the Incomplete Gamma function
1195 (do* ((i 1 (+ i 1))
1196 (an (sub a 1.0) (mul i (sub a i)))
1197 (b (add 3.0 x (mul -1 a)) (add b 2.0))
1198 (c (div 1.0 gm-min))
1199 (d (div 1.0 (sub b 2.0)))
1200 (h d)
1201 (del 0.0))
1202 ((> i gm-maxit)
1203 (merror (intl:gettext "gamma_incomplete: continued fractions failed for gamma_incomplete(~:M, ~:M).") a x))
1204 (when *debug-gamma*
1205 (format t "~&in continued fractions:~%")
1206 (mformat t "~& : i = ~M~%" i)
1207 (mformat t "~& : h = ~M~%" h))
1208 (setq d (add (mul an d) b))
1209 (when (eq ($sign (sub (simplify (list '(mabs) d)) gm-min)) '$neg)
1210 (setq d gm-min))
1211 (setq c (add b (div an c)))
1212 (when (eq ($sign (sub (simplify (list '(mabs) c)) gm-min)) '$neg)
1213 (setq c gm-min))
1214 (setq d (div 1.0 d))
1215 (setq del (mul d c))
1216 (setq h (mul h del))
1217 (when (eq ($sign (sub (simplify (list '(mabs) (sub del 1.0))) gm-eps))
1218 '$neg)
1219 (return
1220 (mul h
1221 (power x a)
1222 (power ($bfloat '$%e) (mul -1 x)))))))
1224 ;; Expand to multiply everything out.
1225 ($expand
1226 ;; Expansion in continued fraction for the lower incomplete gamma.
1227 (sub (simplify (list '(%gamma) a))
1228 ;; NOTE: We want (power x a) instead of bigfloat:expt
1229 ;; because this preserves how maxima computes x^a when
1230 ;; x is negative and a is rational. For, example
1231 ;; (-8)^(1/2) is -2. bigfloat:expt returns the
1232 ;; principal value.
1233 (mul (power x a)
1234 (power ($bfloat '$%e) (mul -1 x))
1235 (let ((a (bigfloat:to a))
1236 (x (bigfloat:to x)))
1237 (to (bigfloat:/
1238 (bigfloat:lentz
1239 #'(lambda (n)
1240 (bigfloat:+ n a))
1241 #'(lambda (n)
1242 (if (evenp n)
1243 (bigfloat:* (ash n -1) x)
1244 (bigfloat:- (bigfloat:* (bigfloat:+ a (ash n -1))
1245 x))))))))))))))
1248 ;; Series expansion of the Incomplete Gamma function
1249 (do* ((i 1 (+ i 1))
1250 (ap a (add ap 1.0))
1251 (del (div 1.0 a) (mul del (div x ap)))
1252 (sum del (add sum del)))
1253 ((> i gm-maxit)
1254 (merror (intl:gettext "gamma_incomplete: series expansion failed for gamma_incomplete(~:M, ~:M).") a x))
1255 (when *debug-gamma*
1256 (format t "~&GAMMA-INCOMPLETE in series:~%")
1257 (mformat t "~& : i = ~M~%" i)
1258 (mformat t "~& : ap = ~M~%" ap)
1259 (mformat t "~& : x/ap = ~M~%" (div x ap))
1260 (mformat t "~& : del = ~M~%" del)
1261 (mformat t "~& : sum = ~M~%" sum))
1262 (when (eq ($sign (sub (simplify (list '(mabs) del))
1263 (mul (simplify (list '(mabs) sum)) gm-eps)))
1264 '$neg)
1265 (when *debug-gamma* (mformat t "~&Series converged to ~M.~%" sum))
1266 (return
1267 (sub (simplify (list '(%gamma) a))
1268 ($rectform
1269 (mul sum
1270 (power x a)
1271 (power ($bfloat '$%e) (mul -1 x))))))))))))
1273 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1275 (defun complex-bfloat-gamma-incomplete (a x)
1276 (let* ((gm-maxit *gamma-incomplete-maxit*)
1277 (gm-eps (power ($bfloat 10.0) (- $fpprec)))
1278 (gm-min (mul gm-eps gm-eps))
1279 ($ratprint nil))
1280 (when *debug-gamma*
1281 (format t "~&COMPLEX-BFLOAT-GAMMA-INCOMPLETE~%")
1282 (format t " : a = ~A~%" a)
1283 (format t " : x = ~A~%" x))
1284 (cond
1285 ;; The series expansion is done for x within a circle of radius
1286 ;; R, where R = *gamma-radius*+(realpart(a)) for realpart(a) > 0
1287 ;; and R = *gamma-radisu* for realpart(a) < 0. Otherwise a
1288 ;; continued fraction is used.
1289 ((and (eq ($sign (sub (simplify (list '(mabs) x))
1290 (add *gamma-radius*
1291 (if (eq ($sign ($realpart a)) '$pos)
1292 ($realpart a)
1293 0.0))))
1294 '$pos))
1295 (cond
1296 ((not (and (eq ($sign ($realpart x)) '$neg)
1297 (eq ($sign (sub (simplify (list '(mabs) ($imagpart x)))
1298 (simplify (list '(mabs) ($realpart x)))))
1299 '$neg)))
1300 ;; Expansion in continued fractions of the Incomplete Gamma function
1301 (when *debug-gamma*
1302 (format t "~&in continued fractions:~%"))
1303 (do* ((i 1 (+ i 1))
1304 (an (sub a 1.0) (mul i (sub a i)))
1305 (b (add 3.0 x (mul -1 a)) (add b 2.0))
1306 (c (cdiv 1.0 gm-min))
1307 (d (cdiv 1.0 (sub b 2.0)))
1308 (h d)
1309 (del 0.0))
1310 ((> i gm-maxit)
1311 (merror (intl:gettext "gamma_incomplete: continued fractions failed for gamma_incomplete(~:M, ~:M).") a x))
1312 (setq d (add (cmul an d) b))
1313 (when (eq ($sign (sub (simplify (list '(mabs) d)) gm-min)) '$neg)
1314 (setq d gm-min))
1315 (setq c (add b (cdiv an c)))
1316 (when (eq ($sign (sub (simplify (list '(mabs) c)) gm-min)) '$neg)
1317 (setq c gm-min))
1318 (setq d (cdiv 1.0 d))
1319 (setq del (cmul d c))
1320 (setq h (cmul h del))
1321 (when (eq ($sign (sub (simplify (list '(mabs) (sub del 1.0)))
1322 gm-eps))
1323 '$neg)
1324 (return
1325 ($bfloat ; force evaluation of expressions with sin or cos
1326 (cmul h
1327 (cmul
1328 (cpower x a)
1329 (cpower ($bfloat '$%e) ($bfloat (mul -1 x))))))))))
1331 ;; Expand to multiply everything out.
1332 ($expand
1333 ;; Expansion in continued fraction for the lower incomplete gamma.
1334 (sub ($rectform (simplify (list '(%gamma) a)))
1335 ;; NOTE: We want (power x a) instead of bigfloat:expt
1336 ;; because this preserves how maxima computes x^a when
1337 ;; x is negative and a is rational. For, example
1338 ;; (-8)^(1/2) is -2. bigfloat:expt returns the
1339 ;; principal value.
1340 (mul ($rectform (power x a))
1341 ($rectform (power ($bfloat '$%e) (mul -1 x)))
1342 (let ((a (bigfloat:to a))
1343 (x (bigfloat:to x)))
1344 (to (bigfloat:/
1345 (bigfloat:lentz
1346 #'(lambda (n)
1347 (bigfloat:+ n a))
1348 #'(lambda (n)
1349 (if (evenp n)
1350 (bigfloat:* (ash n -1) x)
1351 (bigfloat:- (bigfloat:* (bigfloat:+ a (ash n -1))
1352 x))))))))))))))
1354 ;; Series expansion of the Incomplete Gamma function
1355 (when *debug-gamma*
1356 (format t "~&GAMMA-INCOMPLETE in series:~%"))
1357 (do* ((i 1 (+ i 1))
1358 (ap a (add ap 1.0))
1359 (del (cdiv 1.0 a) (cmul del (cdiv x ap)))
1360 (sum del (add sum del)))
1361 ((> i gm-maxit)
1362 (merror (intl:gettext "gamma_incomplete: series expansion failed for gamma_incomplete(~:M, ~:M).") a x))
1363 (when (eq ($sign (sub (simplify (list '(mabs) del))
1364 (mul (simplify (list '(mabs) sum)) gm-eps)))
1365 '$neg)
1366 (when *debug-gamma* (format t "~&Series converged.~%"))
1367 (return
1368 ($bfloat ; force evaluation of expressions with sin or cos
1369 (sub (simplify (list '(%gamma) a))
1370 (cmul sum
1371 (cmul
1372 (cpower x a)
1373 (cpower ($bfloat '$%e) (mul -1 x)))))))))))))
1375 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1377 ;;; Implementation of the Generalized Incomplete Gamma function
1379 ;;; z2
1380 ;;; /
1381 ;;; [ a - 1 - t
1382 ;;; gamma_incomplete_generalized(a, z1, z2) = I t %e dt
1383 ;;; ]
1384 ;;; /
1385 ;;; z1
1387 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1389 (defmfun $gamma_incomplete_generalized (a z1 z2)
1390 (simplify (list '(%gamma_incomplete_generalized) a z1 z2)))
1392 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1394 ;;; Set the properties alias, reversealias, noun and verb
1396 (defprop $gamma_incomplete_generalized %gamma_incomplete_generalized alias)
1397 (defprop $gamma_incomplete_generalized %gamma_incomplete_generalized verb)
1399 (defprop %gamma_incomplete_generalized
1400 $gamma_incomplete_generalized reversealias)
1401 (defprop %gamma_incomplete_generalized
1402 $gamma_incomplete_generalized noun)
1404 ;;; Generalized Incomplete Gamma function has not mirror symmetry for z1 or z2
1405 ;;; on the negative real axis.
1406 ;;; We support a conjugate-function which test this case.
1408 (defprop %gamma_incomplete_generalized
1409 conjugate-gamma-incomplete-generalized conjugate-function)
1411 (defun conjugate-gamma-incomplete-generalized (args)
1412 (let ((a (first args)) (z1 (second args)) (z2 (third args)))
1413 (cond ((and (off-negative-real-axisp z1) (off-negative-real-axisp z2))
1414 ;; z1 and z2 definitely not on the negative real axis.
1415 ;; Mirror symmetry.
1416 (simplify
1417 (list
1418 '(%gamma_incomplete_generalized)
1419 (simplify (list '($conjugate) a))
1420 (simplify (list '($conjugate) z1))
1421 (simplify (list '($conjugate) z2)))))
1423 ;; On the negative real axis or no information. Unsimplified.
1424 (list
1425 '($conjugate simp)
1426 (simplify (list '(%gamma_incomplete_generalized) a z1 z2)))))))
1428 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1430 ;;; Generalized Incomplete Gamma function is a simplifying function
1432 (defprop %gamma_incomplete_generalized
1433 simp-gamma-incomplete-generalized operators)
1435 ;;; Generalized Incomplete Gamma distributes over bags
1437 (defprop %gamma_incomplete_generalized (mlist $matrix mequal) distribute_over)
1439 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1441 ;;; Differentiation of Generalized Incomplete Gamma function
1443 (defprop %gamma_incomplete_generalized
1444 ((a z1 z2)
1445 ;; The derivative wrt a in terms of hypergeometric_regularized 2F2 function
1446 ;; and the Generalized Incomplete Gamma function (functions.wolfram.com)
1447 ((mplus)
1448 ((mtimes)
1449 ((mexpt) ((%gamma) a) 2)
1450 ((mexpt) z1 a)
1451 (($hypergeometric_regularized)
1452 ((mlist) a a)
1453 ((mlist) ((mplus) 1 a) ((mplus) 1 a))
1454 ((mtimes) -1 z1)))
1455 ((mtimes) -1
1456 ((mexpt) ((%gamma) a) 2)
1457 ((mexpt) z2 a)
1458 (($hypergeometric_regularized)
1459 ((mlist) a a)
1460 ((mlist) ((mplus) 1 a) ((mplus) 1 a))
1461 ((mtimes) -1 z2)))
1462 ((mtimes) -1
1463 ((%gamma_incomplete_generalized) a 0 z1)
1464 ((%log) z1))
1465 ((mtimes)
1466 ((%gamma_incomplete_generalized) a 0 z2)
1467 ((%log) z2)))
1468 ;; The derivative wrt z1
1469 ((mtimes) -1
1470 ((mexpt) $%e ((mtimes) -1 z1))
1471 ((mexpt) z1 ((mplus) -1 a)))
1472 ;; The derivative wrt z2
1473 ((mtimes)
1474 ((mexpt) $%e ((mtimes) -1 z2))
1475 ((mexpt) z2 ((mplus) -1 a))))
1476 grad)
1478 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1480 (defun simp-gamma-incomplete-generalized (expr ignored simpflag)
1481 (declare (ignore ignored))
1482 (arg-count-check 3 expr)
1483 (let ((a (simpcheck (cadr expr) simpflag))
1484 (z1 (simpcheck (caddr expr) simpflag))
1485 (z2 (simpcheck (cadddr expr) simpflag))
1486 ($simpsum t))
1488 (cond
1490 ;; Check for specific values
1492 ((zerop1 z2)
1493 (let ((sgn ($sign ($realpart a))))
1494 (cond
1495 ((member sgn '($pos $pz))
1496 (sub
1497 (simplify (list '(%gamma_incomplete) a z1))
1498 (simplify (list '(%gamma) a))))
1500 (eqtest (list '(%gamma_incomplete_generalized) a z1 z2) expr)))))
1502 ((zerop1 z1)
1503 (let ((sgn ($sign ($realpart a))))
1504 (cond
1505 ((member sgn '($pos $pz))
1506 (sub
1507 (simplify (list '(%gamma) a))
1508 (simplify (list '(%gamma_incomplete) a z2))))
1510 (eqtest (list '(%gamma_incomplete_generalized) a z1 z2) expr)))))
1512 ((zerop1 (sub z1 z2)) 0)
1514 ((eq z2 '$inf) (simplify (list '(%gamma_incomplete) a z1)))
1515 ((eq z1 '$inf) (mul -1 (simplify (list '(%gamma_incomplete) a z2))))
1517 ;; Check for numerical evaluation in Float or Bigfloat precision
1518 ;; Use the numerical routines of the Incomplete Gamma function
1520 ((float-numerical-eval-p a z1 z2)
1521 (complexify
1522 (- (gamma-incomplete ($float a) ($float z1))
1523 (gamma-incomplete ($float a) ($float z2)))))
1525 ((complex-float-numerical-eval-p a z1 z2)
1526 (let ((ca (complex ($float ($realpart a)) ($float ($imagpart a))))
1527 (cz1 (complex ($float ($realpart z1)) ($float ($imagpart z1))))
1528 (cz2 (complex ($float ($realpart z2)) ($float ($imagpart z2)))))
1529 (complexify (- (gamma-incomplete ca cz1) (gamma-incomplete ca cz2)))))
1531 ((bigfloat-numerical-eval-p a z1 z2)
1532 (sub (bfloat-gamma-incomplete ($bfloat a) ($bfloat z1))
1533 (bfloat-gamma-incomplete ($bfloat a) ($bfloat z2))))
1535 ((complex-bigfloat-numerical-eval-p a z1 z2)
1536 (let ((ca (add ($bfloat ($realpart a))
1537 (mul '$%i ($bfloat ($imagpart a)))))
1538 (cz1 (add ($bfloat ($realpart z1))
1539 (mul '$%i ($bfloat ($imagpart z1)))))
1540 (cz2 (add ($bfloat ($realpart z2))
1541 (mul '$%i ($bfloat ($imagpart z2))))))
1542 (sub (complex-bfloat-gamma-incomplete ca cz1)
1543 (complex-bfloat-gamma-incomplete ca cz2))))
1545 ;; Check for transformations and argument simplification
1547 ((and $gamma_expand (mplusp a) (integerp (cadr a)))
1548 ;; Expand gamma_incomplete_generalized(a+n,z1,z2) with n an integer
1549 (let ((n (cadr a))
1550 (a (simplify (cons '(mplus) (cddr a)))))
1551 (cond
1552 ((> n 0)
1553 (mul
1554 (simplify (list '($pochhammer) a n))
1555 (add
1556 (simplify (list '(%gamma_incomplete_generalized) a z1 z2))
1557 (mul
1558 (power '$%e (mul -1 z1))
1559 (let ((index (gensumindex)))
1560 (simpsum1
1561 (div
1562 (power z1 (add a index -1))
1563 (simplify (list '($pochhammer) a index)))
1564 index 1 n)))
1565 (mul -1
1566 (power '$%e (mul -1 z2))
1567 (let ((index (gensumindex)))
1568 (simpsum1
1569 (div
1570 (power z2 (add a index -1))
1571 (simplify (list '($pochhammer) a index)))
1572 index 1 n))))))
1574 ((< n 0)
1575 (setq n (- n))
1576 (add
1577 (mul
1578 (div
1579 (power -1 n)
1580 ($factor (simplify (list '($pochhammer) (sub 1 a) n))))
1581 (simplify (list '(%gamma_incomplete_generalized) a z1 z2)))
1582 (mul -1
1583 (power '$%e (mul -1 z2))
1584 (let ((index (gensumindex)))
1585 (simpsum1
1586 (div
1587 (power z1 (add a index (- n) -1))
1588 (simplify (list '($pochhammer) (sub a n) index)))
1589 index 1 n)))
1590 (mul
1591 (power '$%e (mul -1 z2))
1592 (let ((index (gensumindex)))
1593 (simpsum1
1594 (div
1595 (power z2 (add a index (- n) -1))
1596 (simplify (list '($pochhammer) (sub a n) index)))
1597 index 1 n))))))))
1599 (t (eqtest (list '(%gamma_incomplete_generalized) a z1 z2) expr)))))
1601 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1603 ;;; Implementation of the Regularized Incomplete Gamma function
1605 ;;; A&S 6.5.1
1607 ;;; gamma_incomplete(a, z)
1608 ;;; gamma_incomplete_regularized(a, z) = ----------------------
1609 ;;; gamma(a)
1610 ;;;
1611 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1613 (defmfun $gamma_incomplete_regularized (a z)
1614 (simplify (list '(%gamma_incomplete_regularized) a z)))
1616 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1618 (defprop $gamma_incomplete_regularized %gamma_incomplete_regularized alias)
1619 (defprop $gamma_incomplete_regularized %gamma_incomplete_regularized verb)
1621 (defprop %gamma_incomplete_regularized
1622 $gamma_incomplete_regularized reversealias)
1623 (defprop %gamma_incomplete_regularized
1624 $gamma_incomplete_regularized noun)
1626 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1628 ;;; Generalized Incomplete Gamma function has not mirror symmetry for z1 or z2
1629 ;;; on the negative real axis.
1630 ;;; We support a conjugate-function which test this case.
1632 (defprop %gamma_incomplete_regularized
1633 conjugate-gamma-incomplete-regularized conjugate-function)
1635 (defun conjugate-gamma-incomplete-regularized (args)
1636 (let ((a (first args)) (z (second args)))
1637 (cond ((off-negative-real-axisp z)
1638 ;; z definitely not on the negative real axis. Mirror symmetry.
1639 (simplify
1640 (list
1641 '(%gamma_incomplete_regularized)
1642 (simplify (list '($conjugate) a))
1643 (simplify (list '($conjugate) z)))))
1645 ;; On the negative real axis or no information. Unsimplified.
1646 (list
1647 '($conjugate simp)
1648 (simplify (list '(%gamma_incomplete_regularized) a z)))))))
1650 ;;; Regularized Incomplete Gamma function is a simplifying function
1652 (defprop %gamma_incomplete_regularized
1653 simp-gamma-incomplete-regularized operators)
1655 ;;; Regularized Incomplete Gamma distributes over bags
1657 (defprop %gamma_incomplete_regularized (mlist $matrix mequal) distribute_over)
1659 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1661 ;;; Differentiation of Regularized Incomplete Gamma function
1663 (defprop %gamma_incomplete_regularized
1664 ((a z)
1665 ;; The derivative wrt a in terms of hypergeometric_regularized 2F2 function
1666 ;; and the Regularized Generalized Incomplete Gamma function
1667 ;; (functions.wolfram.com)
1668 ((mplus)
1669 ((mtimes)
1670 ((%gamma) a)
1671 ((mexpt) z a)
1672 (($hypergeometric_regularized)
1673 ((mlist) a a)
1674 ((mlist) ((mplus) 1 a) ((mplus) 1 a))
1675 ((mtimes) -1 z)))
1676 ((mtimes)
1677 ((%gamma_incomplete_generalized_regularized) a z 0)
1678 ((mplus)
1679 ((%log) z)
1680 ((mtimes) -1 ((mqapply) (($psi array) 0) a)))))
1681 ;; The derivative wrt z
1682 ((mtimes)
1684 ((mexpt) $%e ((mtimes) -1 z))
1685 ((mexpt) z ((mplus) -1 a))
1686 ((mexpt) ((%gamma) a) -1)))
1687 grad)
1689 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1691 (defun simp-gamma-incomplete-regularized (expr ignored simpflag)
1692 (declare (ignore ignored))
1693 (twoargcheck expr)
1694 (let ((a (simpcheck (cadr expr) simpflag))
1695 (z (simpcheck (caddr expr) simpflag))
1696 ($simpsum t)
1697 (ratorder 0))
1699 (cond
1701 ;; Check for specific values
1703 ((zerop1 z)
1704 (let ((sgn ($sign ($realpart a))))
1705 (cond ((member sgn '($neg $zero))
1706 (simp-domain-error
1707 (intl:gettext
1708 "gamma_incomplete_regularized: gamma_incomplete_regularized(~:M,~:M) is undefined.")
1709 a z))
1710 ((member sgn '($pos $pz)) 1)
1711 (t (eqtest (list '(%gamma_incomplete_regularized) a z) expr)))))
1713 ((zerop1 a) 0)
1714 ((eq z '$inf) 0)
1716 ;; Check for numerical evaluation in Float or Bigfloat precision
1718 ((float-numerical-eval-p a z)
1719 (complexify
1720 ;; gamma_incomplete returns a regularized result
1721 (gamma-incomplete ($float a) ($float z) t)))
1723 ((complex-float-numerical-eval-p a z)
1724 (let ((ca (complex ($float ($realpart a)) ($float ($imagpart a))))
1725 (cz (complex ($float ($realpart z)) ($float ($imagpart z)))))
1726 ;; gamma_incomplete returns a regularized result
1727 (complexify (gamma-incomplete ca cz t))))
1729 ((bigfloat-numerical-eval-p a z)
1730 (div (bfloat-gamma-incomplete ($bfloat a) ($bfloat z))
1731 (simplify (list '(%gamma) ($bfloat a)))))
1733 ((complex-bigfloat-numerical-eval-p a z)
1734 (let ((ca (add ($bfloat ($realpart a))
1735 (mul '$%i ($bfloat ($imagpart a)))))
1736 (cz (add ($bfloat ($realpart z))
1737 (mul '$%i ($bfloat ($imagpart z))))))
1738 ($rectform
1739 (div
1740 (complex-bfloat-gamma-incomplete ca cz)
1741 (simplify (list '(%gamma) ca))))))
1743 ;; Check for transformations and argument simplification
1745 ((and $gamma_expand (integerp a))
1746 ;; An integer. Expand the expression.
1747 (let ((sgn ($sign a)))
1748 (cond
1749 ((member sgn '($pos $pz))
1750 (mul
1751 (power '$%e (mul -1 z))
1752 (let ((index (gensumindex)))
1753 (simpsum1
1754 (div
1755 (power z index)
1756 (let (($gamma_expand nil))
1757 (simplify (list '(%gamma) (add index 1)))))
1758 index 0 (sub a 1)))))
1759 ((member sgn '($neg $nz)) 0)
1760 (t (eqtest (list '(%gamma_incomplete_regularized) a z) expr)))))
1762 ((and $gamma_expand (setq ratorder (max-numeric-ratio-p a 2)))
1763 ;; We have a half integral order and $gamma_expand is not NIL.
1764 ;; We expand in a series with the Erfc function
1765 (setq ratorder (- ratorder (/ 1 2)))
1766 (when *debug-gamma*
1767 (format t "~&SIMP-GAMMA-INCOMPLETE-REGULARIZED in RATORDER~%")
1768 (format t "~& : a = ~A~%" a)
1769 (format t "~& : ratorder = ~A~%" ratorder))
1770 (cond
1771 ((equal ratorder 0)
1772 (simplify (list '(%erfc) (power z '((rat simp) 1 2)))))
1774 ((> ratorder 0)
1775 (add
1776 (simplify (list '(%erfc) (power z '((rat simp) 1 2))))
1777 (mul
1778 (power -1 (sub ratorder 1))
1779 (power '$%e (mul -1 z))
1780 (power z '((rat simp) 1 2))
1781 (div 1 (simplify (list '(%gamma) a)))
1782 (let ((index (gensumindex)))
1783 (simpsum1
1784 (mul
1785 (power (mul -1 z) index)
1786 (simplify (list '($pochhammer)
1787 (sub '((rat simp) 1 2) ratorder)
1788 (sub ratorder (add index 1)))))
1789 index 0 (sub ratorder 1))))))
1791 ((< ratorder 0)
1792 (setq ratorder (- ratorder))
1793 (add
1794 (simplify (list '(%erfc) (power z '((rat simp) 1 2))))
1795 (mul -1
1796 (power '$%e (mul -1 z))
1797 (power z (sub '((rat simp) 1 2) ratorder))
1798 (inv (simplify (list '(%gamma) (sub '((rat simp) 1 2) ratorder))))
1799 (let ((index (gensumindex)))
1800 (simpsum1
1801 (div
1802 (power z index)
1803 (simplify (list '($pochhammer)
1804 (sub '((rat simp) 1 2) ratorder)
1805 (add index 1))))
1806 index 0 (sub ratorder 1))))))))
1808 ((and $gamma_expand (mplusp a) (integerp (cadr a)))
1809 (when *debug-gamma*
1810 (format t "~&SIMP-GAMMA-INCOMPLETE-REGULARIZED in COND (mplusp)~%"))
1811 (let ((n (cadr a))
1812 (a (simplify (cons '(mplus) (cddr a)))))
1813 (cond
1814 ((> n 0)
1815 (add
1816 (simplify (list '(%gamma_incomplete_regularized) a z))
1817 ;; We factor the second summand.
1818 ;; Some factors vanish and the result is more readable.
1819 ($factor
1820 (mul
1821 (power '$%e (mul -1 z))
1822 (power z (add a -1))
1823 (div 1 (simplify (list '(%gamma) a)))
1824 (let ((index (gensumindex)))
1825 (simpsum1
1826 (div
1827 (power z index)
1828 (simplify (list '($pochhammer) a index)))
1829 index 1 n))))))
1830 ((< n 0)
1831 (setq n (- n))
1832 (add
1833 (simplify (list '(%gamma_incomplete_regularized) a z))
1834 ;; We factor the second summand.
1835 ($factor
1836 (mul -1
1837 (power '$%e (mul -1 z))
1838 (power z (sub a (add n 1)))
1839 (div 1 (simplify (list '(%gamma) (add a (- n)))))
1840 (let ((index (gensumindex)))
1841 (simpsum1
1842 (div
1843 (power z index)
1844 (simplify (list '($pochhammer) (add a (- n)) index)))
1845 index 1 n)))))))))
1846 ((and $gamma_expand (consp a) (eq 'rat (caar a))
1847 (integerp (second a))
1848 (integerp (third a)))
1849 ;; gamma_incomplete_regularized of numeric rational order.
1850 ;; Expand it out so that the resulting order is between 0 and
1851 ;; 1. Use gamma_incomplete_regularized(a+n,z) to do the dirty
1852 ;; work.
1853 (multiple-value-bind (n order)
1854 (floor (/ (second a) (third a)))
1855 ;; a = n + order where 0 <= order < 1.
1856 (let ((rat-order (rat (numerator order) (denominator order))))
1857 (cond
1858 ((zerop n)
1859 ;; Nothing to do if the order is already between 0 and 1
1860 (eqtest (list '(%gamma_incomplete_regularized) a z) expr))
1862 ;; Use gamma_incomplete_regularized(a+n,z) above. and
1863 ;; then substitue a=order. This works for n positive or
1864 ;; negative.
1865 (let* ((ord (gensym))
1866 (g (simplify (list '(%gamma_incomplete_regularized) (add ord n) z))))
1867 ($substitute rat-order ord g)))))))
1869 (t (eqtest (list '(%gamma_incomplete_regularized) a z) expr)))))
1871 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1873 ;;; Implementation of the Logarithm of the Gamma function
1875 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1877 (defmfun $log_gamma (z)
1878 (simplify (list '(%log_gamma) z)))
1880 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1882 (defprop $log_gamma %log_gamma alias)
1883 (defprop $log_gamma %log_gamma verb)
1885 (defprop %log_gamma $log_gamma reversealias)
1886 (defprop %log_gamma $log_gamma noun)
1888 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1890 (defprop %log_gamma simp-log-gamma operators)
1892 ;;; Logarithm of the Gamma function distributes over bags
1894 (defprop %log_gamma (mlist $matrix mequal) distribute_over)
1896 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1898 (defprop %log_gamma
1899 ((z)
1900 ((mqapply) (($psi array) 0) z))
1901 grad)
1903 ;; integrate(log_gamma(x),x) = psi[-2](x)
1904 (defun log-gamma-integral (x)
1905 (take '(mqapply) (take '($psi) -2) x))
1907 (putprop '%log_gamma (list (list 'x) #'log-gamma-integral) 'integral)
1909 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1911 (defun simp-log-gamma (expr z simpflag)
1912 (oneargcheck expr)
1913 (setq z (simpcheck (cadr expr) simpflag))
1914 (cond
1916 ;; Check for specific values
1918 ((and (mnump z)
1919 (or (zerop1 z)
1920 (and (eq ($sign z) '$neg)
1921 (zerop1 (sub z ($truncate z))))))
1922 ;; We have zero, a negative integer or a float or bigfloat representation.
1923 (simp-domain-error
1924 (intl:gettext "log_gamma: log_gamma(~:M) is undefined.") z))
1926 ((eq z '$inf) '$inf)
1928 ;; Check for numerical evaluation
1930 ((float-numerical-eval-p z)
1931 (complexify (log-gamma-lanczos (complex ($float z) 0))))
1933 ((complex-float-numerical-eval-p z)
1934 (complexify
1935 (log-gamma-lanczos
1936 (complex ($float ($realpart z)) ($float ($imagpart z))))))
1938 ((bigfloat-numerical-eval-p z)
1939 (bfloat-log-gamma ($bfloat z)))
1941 ((complex-bigfloat-numerical-eval-p z)
1942 (complex-bfloat-log-gamma
1943 (add ($bfloat ($realpart z)) (mul '$%i ($bfloat ($imagpart z))))))
1945 ;; Transform to Logarithm of Factorial for integer values
1946 ;; At this point the integer value is positive and not zero.
1948 ((integerp z)
1949 (simplify (list '(%log) (simplify (list '(mfactorial) (- z 1))))))
1951 (t (eqtest (list '(%log_gamma) z) expr))))
1953 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1954 ;;; The functions log-gamma-lanczos, bfloat-log-gamma and
1955 ;;; complex-bfloat-log-gamma are modified versions of the related functions
1956 ;;; gamma-lanczos, bffac and cbffac. The functions return the Logarithm of
1957 ;;; the Gamma function. If we have to calculate the quotient of Gamma functions,
1958 ;;; e. g. for the Beta function, it is much more appropriate to use the
1959 ;;; logarithmic versions to avoid overflow.
1961 ;;; Be careful log(gamma(z)) is only for realpart(z) positive equal to
1962 ;;; log_gamma(z). For a negative realpart(z) log_gamma differ by multiple of
1963 ;;; %pi from log(gamma(z)). But we always have exp(log_gamma(z))= gamma(z).
1964 ;;; The terms to get the transformation for log_gamma(-z) are taken from
1965 ;;; functions.wolfram.com.
1966 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1968 (defun log-gamma-lanczos (z)
1969 (declare (type (complex flonum) z)
1970 (optimize (safety 3)))
1971 (let ((c (make-array 15 :element-type 'flonum
1972 :initial-contents
1973 '(0.99999999999999709182
1974 57.156235665862923517
1975 -59.597960355475491248
1976 14.136097974741747174
1977 -0.49191381609762019978
1978 .33994649984811888699e-4
1979 .46523628927048575665e-4
1980 -.98374475304879564677e-4
1981 .15808870322491248884e-3
1982 -.21026444172410488319e-3
1983 .21743961811521264320e-3
1984 -.16431810653676389022e-3
1985 .84418223983852743293e-4
1986 -.26190838401581408670e-4
1987 .36899182659531622704e-5))))
1988 (declare (type (simple-array flonum (15)) c))
1989 (if (minusp (realpart z))
1990 (let ((z (- z)))
1994 (- (float pi))
1995 (complex 0 1)
1996 (abs (floor (realpart z)))
1997 (- 1 (abs (signum (imagpart z)))))
1998 (log (float pi))
1999 (- (log (- z)))
2000 (- (log (sin (* (float pi) (- z (floor (realpart z)))))))
2002 (float pi)
2003 (complex 0 1)
2004 (floor (realpart z))
2005 (signum (imagpart z))))
2006 (log-gamma-lanczos z)))
2007 (let* ((z (- z 1))
2008 (zh (+ z 1/2))
2009 (zgh (+ zh 607/128))
2010 (lnzp (* (/ zh 2) (log zgh)))
2011 (ss
2012 (do ((sum 0.0)
2013 (pp (1- (length c)) (1- pp)))
2014 ((< pp 1)
2015 sum)
2016 (incf sum (/ (aref c pp) (+ z pp))))))
2017 (+ (log (sqrt (float (* 2 pi))))
2018 (log (+ ss (aref c 0)))
2019 (+ (- zgh) (* 2 lnzp)))))))
2021 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2023 (defun bfloat-log-gamma (z)
2024 (let (($ratprint nil)
2025 (bigfloat%pi ($bfloat '$%pi)))
2026 (cond
2027 ((eq ($sign z) '$neg)
2028 (let ((z (mul -1 z)))
2029 (sub
2030 (add
2031 (mul -1 bigfloat%pi '$%i
2032 (simplify (list '(mabs) (simplify (list '($floor) ($realpart z)))))
2033 (sub 1
2034 (simplify
2035 (list '(mabs) (simplify (list '(%signum) ($imagpart z)))))))
2036 (simplify (list '(%log) bigfloat%pi))
2037 (mul -1 (simplify (list '(%log) (mul -1 z))))
2038 (mul -1
2039 (simplify (list '(%log)
2040 (simplify (list '(%sin)
2041 (mul
2042 bigfloat%pi
2043 (sub z (simplify (list '($floor) ($realpart z))))))))))
2044 (mul
2045 bigfloat%pi '$%i
2046 (simplify (list '($floor) ($realpart z)))
2047 (simplify (list '(%signum) ($imagpart z)))))
2048 (bfloat-log-gamma z))))
2050 (let* ((k (* 2 (+ 1 ($entier (* 0.41 $fpprec)))))
2051 (m ($bfloat bigfloatone))
2052 (z+k (add z k -1))
2053 (y (power z+k 2))
2054 (x ($bfloat bigfloatzero))
2055 (ii))
2056 (dotimes (i (/ k 2))
2057 (setq ii (* 2 (+ i 1)))
2058 (setq m (mul m (add z ii -2) (add z ii -1)))
2059 (setq x (div
2060 (add x
2061 (div ($bern (+ k (- ii) 2))
2062 (* (+ k (- ii) 1) (+ k (- ii) 2))))
2063 y)))
2064 (add
2065 (div (simplify (list '(%log) (mul 2 bigfloat%pi z+k))) 2)
2066 (mul z+k (add (simplify (list '(%log) z+k)) x -1))
2067 (mul -1 (simplify (list '(%log) m)))))))))
2069 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2071 (defun complex-bfloat-log-gamma (z)
2072 (let (($ratprint nil)
2073 (bigfloat%pi ($bfloat '$%pi)))
2074 (cond
2075 ((eq ($sign ($realpart z)) '$neg)
2076 (let ((z (mul -1 z)))
2077 (sub
2078 (add
2079 (mul -1 bigfloat%pi '$%i
2080 (simplify (list '(mabs) (simplify (list '($floor) ($realpart z)))))
2081 (sub 1
2082 (simplify
2083 (list '(mabs) (simplify (list '(%signum) ($imagpart z)))))))
2084 (simplify (list '(%log) bigfloat%pi))
2085 (mul -1 (simplify (list '(%log) (mul -1 z))))
2086 (mul -1
2087 (simplify (list '(%log)
2088 (simplify (list '(%sin)
2089 (mul
2090 bigfloat%pi
2091 (sub z (simplify (list '($floor) ($realpart z))))))))))
2092 (mul
2093 bigfloat%pi '$%i
2094 (simplify (list '($floor) ($realpart z)))
2095 (simplify (list '(%signum) ($imagpart z)))))
2096 (complex-bfloat-log-gamma z))))
2098 (let* ((k (* 2 (+ 1 ($entier (* 0.41 $fpprec)))))
2099 (m ($bfloat bigfloatone))
2100 (z+k (add z k -1))
2101 (y ($rectform (power z+k 2)))
2102 (x ($bfloat bigfloatzero))
2103 (ii))
2104 (dotimes (i (/ k 2))
2105 (setq ii (* 2 (+ i 1)))
2106 (setq m ($rectform (mul m (add z ii -2) (add z ii -1))))
2107 (setq x ($rectform
2108 (div
2109 (add x
2110 (div ($bern (+ k (- ii) 2))
2111 (* (+ k (- ii) 1) (+ k (- ii) 2))))
2112 y))))
2113 ($rectform
2114 (add
2115 (div (simplify (list '(%log) (mul 2 bigfloat%pi z+k))) 2)
2116 (mul z+k (add (simplify (list '(%log) z+k)) x -1))
2117 (mul -1 (simplify (list '(%log) m))))))))))
2119 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2121 ;;; Implementation of the Error function Erf(z)
2123 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2125 (defmfun $erf (z)
2126 (simplify (list '(%erf) z)))
2128 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2130 (defprop $erf %erf alias)
2131 (defprop $erf %erf verb)
2133 (defprop %erf $erf reversealias)
2134 (defprop %erf $erf noun)
2136 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2138 ;;; erf has mirror symmetry
2140 (defprop %erf t commutes-with-conjugate)
2142 ;;; erf is an odd function
2144 (defprop %erf odd-function-reflect reflection-rule)
2146 ;;; erf is a simplifying function
2148 (defprop %erf simp-erf operators)
2150 ;;; erf distributes over bags
2152 (defprop %erf (mlist $matrix mequal) distribute_over)
2154 ;;; Derivative of the Error function erf
2156 (defprop %erf
2157 ((z)
2158 ((mtimes) 2
2159 ((mexpt) $%pi ((rat) -1 2))
2160 ((mexpt) $%e ((mtimes) -1 ((mexpt) z 2)))))
2161 grad)
2163 ;;; Integral of the Error function erf
2165 (defprop %erf
2166 ((z)
2167 ((mplus)
2168 ((mtimes)
2169 ((mexpt) $%pi ((rat) -1 2))
2170 ((mexpt) $%e ((mtimes) -1 ((mexpt) z 2))))
2171 ((mtimes) z ((%erf) z))))
2172 integral)
2174 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2176 (defun erf-hypergeometric (z)
2177 ; See A&S 7.1.21 or http://functions.wolfram.com/06.25.26.0001.01
2178 (mul 2
2180 (power '$%pi '((rat simp) -1 2))
2181 (list '(%hypergeometric simp)
2182 (list '(mlist simp) '((rat simp) 1 2))
2183 (list '(mlist simp) '((rat simp) 3 2))
2184 (mul -1 (power z 2)))))
2186 (defun simp-erf (expr z simpflag)
2187 (oneargcheck expr)
2188 (setq z (simpcheck (cadr expr) simpflag))
2189 (cond
2191 ;; Check for specific values
2193 ((zerop1 z) z)
2194 ((eq z '$inf) 1)
2195 ((eq z '$minf) -1)
2197 ;; Check for numerical evaluation
2199 ((float-numerical-eval-p z)
2200 (bigfloat::bf-erf ($float z)))
2201 ((complex-float-numerical-eval-p z)
2202 (complexify
2203 (bigfloat::bf-erf (complex ($float ($realpart z)) ($float ($imagpart z))))))
2204 ((bigfloat-numerical-eval-p z)
2205 (to (bigfloat::bf-erf (bigfloat:to ($bfloat z)))))
2206 ((complex-bigfloat-numerical-eval-p z)
2207 (to (bigfloat::bf-erf
2208 (bigfloat:to (add ($bfloat ($realpart z)) (mul '$%i ($bfloat ($imagpart z))))))))
2210 ;; Argument simplification
2212 ((taylorize (mop expr) (second expr)))
2213 ((and $erf_%iargs
2214 (not $erf_representation)
2215 (multiplep z '$%i))
2216 (mul '$%i (simplify (list '(%erfi) (coeff z '$%i 1)))))
2217 ((apply-reflection-simp (mop expr) z $trigsign))
2219 ;; Representation through more general functions
2221 ($hypergeometric_representation
2222 (erf-hypergeometric z))
2224 ;; Transformation to Erfc or Erfi
2226 ((and $erf_representation
2227 (not (eq $erf_representation '$erf)))
2228 (case $erf_representation
2229 (%erfc
2230 (sub 1 (take '(%erfc) z)))
2231 (%erfi
2232 (mul -1 '$%i (take '(%erfi) (mul '$%i z))))
2234 (eqtest (list '(%erf) z) expr))))
2237 (eqtest (list '(%erf) z) expr))))
2239 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2241 (defun erf (z)
2242 ;; We use the slatec routine for float values.
2243 (slatec:derf (float z)))
2245 ;; Compute erf(z) using the relationship
2247 ;; erf(z) = sqrt(z^2)/z*(1 - gamma_incomplete(1/2,z^2)/sqrt(%pi))
2249 ;; When z is real sqrt(z^2)/z is signum(z). For complex z,
2250 ;; sqrt(z^2)/z = 1 if -%pi/2 < arg(z) <= %pi/2 and -1 otherwise.
2252 ;; This relationship has serious round-off issues when z is small
2253 ;; because gamma_incomplete(1/2,z^2)/sqrt(%pi) is near 1.
2255 ;; complex-erf is for (lisp) complex numbers; bfloat-erf is for real
2256 ;; bfloats, and complex-bfloat-erf is for complex bfloats. Care is
2257 ;; taken to return real results for real arguments and imaginary
2258 ;; results for imaginary arguments
2259 (defun complex-erf (z)
2260 (let ((result
2262 (if (or (< (realpart z) 0.0) (and (= (realpart z) 0.0) (< (imagpart z) 0.0)))
2265 (- 1.0
2266 (* (/ (sqrt (float pi))) (gamma-incomplete 0.5 (* z z)))))))
2267 (cond
2268 ((= (imagpart z) 0.0)
2269 ;; Pure real argument, the result is real
2270 (complex (realpart result) 0.0))
2271 ((= (realpart z) 0.0)
2272 ;; Pure imaginary argument, the result is pure imaginary
2273 (complex 0.0 (imagpart result)))
2275 result))))
2277 (defun bfloat-erf (z)
2278 ;; Warning! This has round-off problems when abs(z) is very small.
2279 (let ((1//2 ($bfloat '((rat simp) 1 2))))
2280 ;; The argument is real, the result is real too
2281 ($realpart
2282 (mul
2283 (simplify (list '(%signum) z))
2284 (sub 1
2285 (mul
2286 (div 1 (power ($bfloat '$%pi) 1//2))
2287 (bfloat-gamma-incomplete 1//2 ($bfloat (power z 2)))))))))
2289 (defun complex-bfloat-erf (z)
2290 ;; Warning! This has round-off problems when abs(z) is very small.
2291 (let* (($ratprint nil)
2292 (1//2 ($bfloat '((rat simp) 1 2)))
2293 (result
2294 (cmul
2295 (cdiv (cpower (cpower z 2) 1//2) z)
2296 (sub 1
2297 (cmul
2298 (div 1 (power ($bfloat '$%pi) 1//2))
2299 (complex-bfloat-gamma-incomplete
2300 1//2
2301 ($bfloat (cpower z 2))))))))
2302 (cond
2303 ((zerop1 ($imagpart z))
2304 ;; Pure real argument, the result is real
2305 ($realpart result))
2306 ((zerop1 ($realpart z))
2307 ;; Pure imaginary argument, the result is pure imaginary
2308 (mul '$%i ($imagpart result)))
2310 ;; A general complex result
2311 result))))
2313 (in-package :bigfloat)
2315 ;; Erf(z) for all z. Z must be a CL real or complex number or a
2316 ;; BIGFLOAT or COMPLEX-BIGFLOAT object. The result will be of the
2317 ;; same type as Z.
2318 (defun bf-erf (z)
2319 (cond ((typep z 'cl:real)
2320 ;; Use Slatec derf, which should be faster than the series.
2321 (maxima::erf z))
2322 ((<= (abs z) 0.476936)
2323 ;; Use the series A&S 7.1.5 for small x:
2325 ;; erf(z) = 2*z/sqrt(%pi) * sum((-1)^n*z^(2*n)/n!/(2*n+1), n, 0, inf)
2327 ;; The threshold is approximately erf(x) = 0.5. (Doesn't
2328 ;; have to be super accurate.) This gives max accuracy when
2329 ;; using the identity erf(x) = 1 - erfc(x).
2330 (let ((z^2 (* z z)))
2331 (/ (* 2 z (sum-power-series z^2
2332 #'(lambda (k)
2333 (let ((2k (+ k k)))
2334 (- (/ (- 2k 1)
2336 (+ 2k 1)))))))
2337 (sqrt (%pi z)))))
2339 ;; The general case.
2340 (etypecase z
2341 (cl:real (maxima::erf z))
2342 (cl:complex (maxima::complex-erf z))
2343 (bigfloat
2344 (bigfloat (maxima::$bfloat (maxima::$expand (maxima::bfloat-erf (maxima::to z))))))
2345 (complex-bigfloat
2346 (bigfloat (maxima::$bfloat (maxima::$expand (maxima::complex-bfloat-erf (maxima::to z))))))))))
2348 (defun bf-erfc (z)
2349 ;; Compute erfc(z) via 1 - erf(z) is not very accurate if erf(z) is
2350 ;; near 1. Wolfram says
2352 ;; erfc(z) = 1 - sqrt(z^2)/z * (1 - 1/sqrt(pi)*gamma_incomplete_tail(1/2, z^2))
2354 ;; For real(z) > 0, sqrt(z^2)/z is 1 so
2356 ;; erfc(z) = 1 - (1 - 1/sqrt(pi)*gamma_incomplete_tail(1/2,z^2))
2357 ;; = 1/sqrt(pi)*gamma_incomplete_tail(1/2,z^2)
2359 ;; For real(z) < 0, sqrt(z^2)/z is -1 so
2361 ;; erfc(z) = 1 + (1 - 1/sqrt(pi)*gamma_incomplete_tail(1/2,z^2))
2362 ;; = 2 - 1/sqrt(pi)*gamma_incomplete(1/2,z^2)
2363 (flet ((gamma-inc (z)
2364 (etypecase z
2365 (cl:number
2366 (maxima::gamma-incomplete 0.5 z))
2367 (bigfloat
2368 (bigfloat:to (maxima::$bfloat
2369 (maxima::bfloat-gamma-incomplete (maxima::$bfloat maxima::1//2)
2370 (maxima::to z)))))
2371 (complex-bigfloat
2372 (bigfloat:to (maxima::$bfloat
2373 (maxima::complex-bfloat-gamma-incomplete (maxima::$bfloat maxima::1//2)
2374 (maxima::to z))))))))
2375 (if (>= (realpart z) 0)
2376 (/ (gamma-inc (* z z))
2377 (sqrt (%pi z)))
2378 (- 2
2379 (/ (gamma-inc (* z z))
2380 (sqrt (%pi z)))))))
2382 (in-package :maxima)
2384 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2386 ;;; Implementation of the Generalized Error function Erf(z1,z2)
2388 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2390 (defmfun $erf_generalized (z1 z2)
2391 (simplify (list '(%erf_generalized) z1 z2)))
2393 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2395 (defprop $erf_generalized %erf_generalized alias)
2396 (defprop $erf_generalized %erf_generalized verb)
2398 (defprop %erf_generalized $erf_generalized reversealias)
2399 (defprop %erf_generalized $erf_generalized noun)
2401 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2403 ;;; Generalized Erf has mirror symmetry
2405 (defprop %erf_generalized t commutes-with-conjugate)
2407 ;;; Generalized Erf distributes over bags
2409 (defprop %erf_generalized (mlist $matrix mequal) distribute_over)
2411 ;;; Generalized Erf is antisymmetric Erf(z1,z2) = - Erf(z2,z1)
2413 (eval-when
2414 #+gcl (load eval)
2415 #-gcl (:load-toplevel :execute)
2416 (let (($context '$global) (context '$global))
2417 (meval '(($declare) %erf_generalized $antisymmetric))
2418 ;; Let's remove built-in symbols from list for user-defined properties.
2419 (setq $props (remove '%erf_generalized $props))))
2421 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2423 (defprop %erf_generalized simp-erf-generalized operators)
2425 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2427 (defprop %erf_generalized
2428 ((z1 z2)
2429 ;; derivative wrt z1
2430 ((mtimes) -2
2431 ((mexpt) $%pi ((rat) -1 2))
2432 ((mexpt) $%e ((mtimes) -1 ((mexpt) z1 2))))
2433 ;; derviative wrt z2
2434 ((mtimes) 2
2435 ((mexpt) $%pi ((rat) -1 2))
2436 ((mexpt) $%e ((mtimes) -1 ((mexpt) z2 2)))))
2437 grad)
2439 ;;; ----------------------------------------------------------------------------
2441 (defprop %erf_generalized simplim%erf_generalized simplim%function)
2443 (defun simplim%erf_generalized (expr var val)
2444 ;; Look for the limit of the arguments.
2445 (let ((z1 (limit (cadr expr) var val 'think))
2446 (z2 (limit (caddr expr) var val 'think)))
2447 (cond
2448 ;; Handle infinities at this place.
2449 ((or (eq z2 '$inf)
2450 (alike1 z2 '((mtimes) -1 $minf)))
2451 (sub 1 (take '(%erf) z1)))
2452 ((or (eq z2 '$minf)
2453 (alike1 z2 '((mtimes) -1 $inf)))
2454 (sub (mul -1 (take '(%erf) z1)) 1))
2455 ((or (eq z1 '$inf)
2456 (alike1 z1 '((mtimes) -1 $minf)))
2457 (sub (take '(%erf) z2) 1))
2458 ((or (eq z1 '$minf)
2459 (alike1 z1 '((mtimes) -1 $inf)))
2460 (add (take '(%erf) z2) 1))
2462 ;; All other cases are handled by the simplifier of the function.
2463 (simplify (list '(%erf_generalized) z1 z2))))))
2465 ;;; ----------------------------------------------------------------------------
2467 (defun simp-erf-generalized (expr ignored simpflag)
2468 (declare (ignore ignored))
2469 (twoargcheck expr)
2470 (let ((z1 (simpcheck (cadr expr) simpflag))
2471 (z2 (simpcheck (caddr expr) simpflag)))
2472 (cond
2474 ;; Check for specific values
2476 ((and (zerop1 z1) (zerop1 z2)) 0)
2477 ((zerop1 z1) (take '(%erf) z2))
2478 ((zerop1 z2) (mul -1 (take '(%erf) z1)))
2479 ((or (eq z2 '$inf)
2480 (alike1 z2 '((mtimes) -1 $minf)))
2481 (sub 1 (take '(%erf) z1)))
2482 ((or (eq z2 '$minf)
2483 (alike1 z2 '((mtimes) -1 $inf)))
2484 (sub (mul -1 (take '(%erf) z1)) 1))
2485 ((or (eq z1 '$inf)
2486 (alike1 z1 '((mtimes) -1 $minf)))
2487 (sub (take '(%erf) z2) 1))
2488 ((or (eq z1 '$minf)
2489 (alike1 z1 '((mtimes) -1 $inf)))
2490 (add (take '(%erf) z2) 1))
2492 ;; Check for numerical evaluation. Use erf(z1,z2) = erf(z2)-erf(z1)
2494 ((float-numerical-eval-p z1 z2)
2495 (- (bigfloat::bf-erf ($float z2))
2496 (bigfloat::bf-erf ($float z1))))
2497 ((complex-float-numerical-eval-p z1 z2)
2498 (complexify
2500 (bigfloat::bf-erf
2501 (complex ($float ($realpart z2)) ($float ($imagpart z2))))
2502 (bigfloat::bf-erf
2503 (complex ($float ($realpart z1)) ($float ($imagpart z1)))))))
2504 ((bigfloat-numerical-eval-p z1 z2)
2505 (to (bigfloat:-
2506 (bigfloat::bf-erf (bigfloat:to ($bfloat z2)))
2507 (bigfloat::bf-erf (bigfloat:to ($bfloat z1))))))
2508 ((complex-bigfloat-numerical-eval-p z1 z2)
2509 (to (bigfloat:-
2510 (bigfloat::bf-erf
2511 (bigfloat:to (add ($bfloat ($realpart z2)) (mul '$%i ($bfloat ($imagpart z2))))))
2512 (bigfloat::bf-erf
2513 (bigfloat:to (add ($bfloat ($realpart z1)) (mul '$%i ($bfloat ($imagpart z1)))))))))
2515 ;; Argument simplification
2517 ((and $trigsign (great (mul -1 z1) z1) (great (mul -1 z2) z2))
2518 (mul -1 (simplify (list '(%erf_generalized) (mul -1 z1) (mul -1 z2)))))
2520 ;; Representation through more general functions
2522 ($hypergeometric_representation
2523 (sub (erf-hypergeometric z2) (erf-hypergeometric z1)))
2525 ;; Transformation to Erf
2527 ($erf_representation
2528 (sub (simplify (list '(%erf) z2)) (simplify (list '(%erf) z1))))
2531 (eqtest (list '(%erf_generalized) z1 z2) expr)))))
2533 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2535 ;;; Implementation of the Complementary Error function Erfc(z)
2537 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2539 (defmfun $erfc (z)
2540 (simplify (list '(%erfc) z)))
2542 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2544 (defprop $erfc %erfc alias)
2545 (defprop $erfc %erfc verb)
2547 (defprop %erfc $erfc reversealias)
2548 (defprop %erfc $erfc noun)
2550 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2552 (defprop %erfc t commutes-with-conjugate)
2554 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2556 (defprop %erfc simp-erfc operators)
2558 ;;; Complementary Error function distributes over bags
2560 (defprop %erfc (mlist $matrix mequal) distribute_over)
2562 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2564 (defprop %erfc
2565 ((z)
2566 ((mtimes) -2
2567 ((mexpt) $%pi ((rat) -1 2))
2568 ((mexpt) $%e ((mtimes) -1 ((mexpt) z 2)))))
2569 grad)
2571 ;;; Integral of the Error function erfc
2573 (defprop %erfc
2574 ((z)
2575 ((mplus)
2576 ((mtimes) -1
2577 ((mexpt) $%pi ((rat) -1 2))
2578 ((mexpt) $%e ((mtimes) -1 ((mexpt) z 2))))
2579 ((mtimes) z ((%erfc) z))))
2580 integral)
2582 ;;; ----------------------------------------------------------------------------
2584 (defprop %erfc simplim%erfc simplim%function)
2586 (defun simplim%erfc (expr var val)
2587 ;; Look for the limit of the arguments.
2588 (let ((z (limit (cadr expr) var val 'think)))
2589 (cond
2590 ;; Handle infinities at this place.
2591 ((eq z '$inf) 0)
2592 ((eq z '$minf) 2)
2593 ((eq z '$infinity) ;; parallel to code in simplim%erf-%tanh
2594 (destructuring-let (((rpart . ipart) (trisplit (cadr expr)))
2595 (ans ()) (rlim ()))
2596 (setq rlim (limit rpart var val 'think))
2597 (setq ans
2598 (limit (m* rpart (m^t ipart -1)) var val 'think))
2599 (setq ans ($asksign (m+ `((mabs) ,ans) -1)))
2600 (cond ((or (eq ans '$pos) (eq ans '$zero))
2601 (cond ((eq rlim '$inf) 0)
2602 ((eq rlim '$minf) 2)
2603 (t '$und)))
2604 (t '$und))))
2605 ((eq z '$ind) '$ind)
2607 ;; All other cases are handled by the simplifier of the function.
2608 (simplify (list '(%erfc) z))))))
2610 ;;; ----------------------------------------------------------------------------
2612 (defun simp-erfc (expr z simpflag)
2613 (oneargcheck expr)
2614 (setq z (simpcheck (cadr expr) simpflag))
2615 (cond
2617 ;; Check for specific values
2619 ((zerop1 z) 1)
2620 ((eq z '$inf) 0)
2621 ((eq z '$minf) 2)
2623 ;; Check for numerical evaluation.
2625 ((numerical-eval-p z)
2626 (to (bigfloat::bf-erfc (bigfloat:to z))))
2628 ;; Argument simplification
2630 ((taylorize (mop expr) (second expr)))
2631 ((and $trigsign (great (mul -1 z) z))
2632 (sub 2 (simplify (list '(%erfc) (mul -1 z)))))
2634 ;; Representation through more general functions
2636 ($hypergeometric_representation
2637 (sub 1 (erf-hypergeometric z)))
2639 ;; Transformation to Erf or Erfi
2641 ((and $erf_representation
2642 (not (eq $erf_representation '$erfc)))
2643 (case $erf_representation
2644 (%erf
2645 (sub 1 (take '(%erf) z)))
2646 (%erfi
2647 (add 1 (mul '$%i (take '(%erfi) (mul '$%i z)))))
2649 (eqtest (list '(%erfc) z) expr))))
2652 (eqtest (list '(%erfc) z) expr))))
2654 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2656 ;;; Implementation of the Imaginary Error function Erfi(z)
2658 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2660 (defmfun $erfi (z)
2661 (simplify (list '(%erfi) z)))
2663 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2665 (defprop $erfi %erfi alias)
2666 (defprop $erfi %erfi verb)
2668 (defprop %erfi $erfi reversealias)
2669 (defprop %erfi $erfi noun)
2671 ;;; erfi has mirror symmetry
2673 (defprop %erfi t commutes-with-conjugate)
2675 ;;; erfi is an odd function
2677 (defprop %erfi odd-function-reflect reflection-rule)
2679 ;;; erfi is an simplifying function
2681 (defprop %erfi simp-erfi operators)
2683 ;;; erfi distributes over bags
2685 (defprop %erfi (mlist $matrix mequal) distribute_over)
2687 ;;; Derivative of the Error function erfi
2689 (defprop %erfi
2690 ((z)
2691 ((mtimes) 2
2692 ((mexpt) $%pi ((rat) -1 2))
2693 ((mexpt) $%e ((mexpt) z 2))))
2694 grad)
2696 ;;; Integral of the Error function erfi
2698 (defprop %erfi
2699 ((z)
2700 ((mplus)
2701 ((mtimes) -1
2702 ((mexpt) $%pi ((rat) -1 2))
2703 ((mexpt) $%e ((mexpt) z 2)))
2704 ((mtimes) z ((%erfi) z))))
2705 integral)
2707 ;;; ----------------------------------------------------------------------------
2709 (defprop %erfi simplim%erfi simplim%function)
2711 (defun simplim%erfi (expr var val)
2712 ;; Look for the limit of the arguments.
2713 (let ((z (limit (cadr expr) var val 'think)))
2714 (cond
2715 ;; Handle infinities at this place.
2716 ((eq z '$inf) '$inf)
2717 ((eq z '$minf) '$minf)
2719 ;; All other cases are handled by the simplifier of the function.
2720 (simplify (list '(%erfi) z))))))
2722 ;;; ----------------------------------------------------------------------------
2724 (in-package :bigfloat)
2725 (defun bf-erfi (z)
2726 (flet ((erfi (z)
2727 ;; Use the relationship erfi(z) = -%i*erf(%i*z)
2728 (let* ((iz (complex (- (imagpart z)) (realpart z))) ; %i*z
2729 (result (bf-erf iz)))
2730 (complex (imagpart result) (- (realpart result))))))
2731 ;; Take care to return real results when the argument is real.
2732 (if (realp z)
2733 (if (minusp z)
2734 (- (bf-erfi (- z)))
2735 (realpart (erfi z)))
2736 (erfi z))))
2738 (in-package :maxima)
2740 (defun simp-erfi (expr z simpflag)
2741 (oneargcheck expr)
2742 (setq z (simpcheck (cadr expr) simpflag))
2743 (cond
2745 ;; Check for specific values
2747 ((zerop1 z) z)
2748 ((eq z '$inf) '$inf)
2749 ((eq z '$minf) '$minf)
2751 ;; Check for numerical evaluation. Use erfi(z) = -%i*erf(%i*z).
2753 ((numerical-eval-p z)
2754 (to (bigfloat::bf-erfi (bigfloat:to z))))
2756 ;; Argument simplification
2758 ((taylorize (mop expr) (second expr)))
2759 ((and $erf_%iargs
2760 (multiplep z '$%i))
2761 (mul '$%i (simplify (list '(%erf) (coeff z '$%i 1)))))
2762 ((apply-reflection-simp (mop expr) z $trigsign))
2764 ;; Representation through more general functions
2766 ($hypergeometric_representation
2767 (mul -1 '$%i (erf-hypergeometric (mul '$%i z))))
2769 ;; Transformation to Erf or Erfc
2771 ((and $erf_representation
2772 (not (eq $erf_representation '$erfi)))
2773 (case $erf_representation
2774 (%erf
2775 (mul -1 '$%i (take '(%erf) (mul '$%i z))))
2776 (%erfc
2777 (sub (mul '$%i (take '(%erfc) (mul '$%i z))) '$%i))
2779 (eqtest (list '(%erfi) z) expr))))
2782 (eqtest (list '(%erfi) z) expr))))
2784 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2786 ;;; The implementation of the Inverse Error function
2788 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2790 (defmfun $inverse_erf (z)
2791 (simplify (list '(%inverse_erf) z)))
2793 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2795 ;;; Set properties to give full support to the parser and display
2797 (defprop $inverse_erf %inverse_erf alias)
2798 (defprop $inverse_erf %inverse_erf verb)
2800 (defprop %inverse_erf $inverse_erf reversealias)
2801 (defprop %inverse_erf $inverse_erf noun)
2803 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2805 ;;; The Inverse Error function is a simplifying function
2807 (defprop %inverse_erf simp-inverse-erf operators)
2809 ;;; The Inverse Error function distributes over bags
2811 (defprop %inverse_erf (mlist $matrix mequal) distribute_over)
2813 ;;; inverse_erf is the inverse of the erf function
2815 (defprop %inverse_erf %erf $inverse)
2816 (defprop %erf %inverse_erf $inverse)
2818 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2820 ;;; Differentiation of the Inverse Error function
2822 (defprop %inverse_erf
2823 ((z)
2824 ((mtimes)
2825 ((rat) 1 2)
2826 ((mexpt) $%pi ((rat) 1 2))
2827 ((mexpt) $%e ((mexpt) ((%inverse_erf) z) 2))))
2828 grad)
2830 ;;; Integral of the Inverse Error function
2832 (defprop %inverse_erf
2833 ((z)
2834 ((mtimes) -1
2835 ((mexpt) $%pi ((rat) -1 2))
2836 ((mexpt) $%e ((mtimes) -1 ((mexpt) ((%inverse_erf) z) 2)))))
2837 integral)
2839 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2841 ;;; We support a simplim%function. The function is looked up in simplimit and
2842 ;;; handles specific values of the function.
2844 (defprop %inverse_erf simplim%inverse_erf simplim%function)
2846 (defun simplim%inverse_erf (expr var val)
2847 ;; Look for the limit of the argument.
2848 (let ((z (limit (cadr expr) var val 'think)))
2849 (cond
2850 ;; Handle an argument 1 at this place
2851 ((onep1 z) '$inf)
2852 ((onep1 (mul -1 z)) '$minf)
2854 ;; All other cases are handled by the simplifier of the function.
2855 (simplify (list '(%inverse_erf) z))))))
2857 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2859 (defun simp-inverse-erf (expr z simpflag)
2860 (oneargcheck expr)
2861 (setq z (simpcheck (cadr expr) simpflag))
2862 (cond
2863 ((or (onep1 z)
2864 (onep1 (mul -1 z)))
2865 (simp-domain-error
2866 (intl:gettext "inverse_erf: inverse_erf(~:M) is undefined.") z))
2867 ((zerop1 z) z)
2868 ((numerical-eval-p z)
2869 (to (bigfloat::bf-inverse-erf (bigfloat:to z))))
2870 ((taylorize (mop expr) (cadr expr)))
2872 (eqtest (list '(%inverse_erf) z) expr))))
2874 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2876 ;;; The implementation of the Inverse Complementary Error function
2878 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2880 (defmfun $inverse_erfc (z)
2881 (simplify (list '(%inverse_erfc) z)))
2883 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2885 ;;; Set properties to give full support to the parser and display
2887 (defprop $inverse_erfc %inverse_erfc alias)
2888 (defprop $inverse_erfc %inverse_erfc verb)
2890 (defprop %inverse_erfc $inverse_erfc reversealias)
2891 (defprop %inverse_erfc $inverse_erfc noun)
2893 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2895 ;;; Inverse Complementary Error function is a simplifying function
2897 (defprop %inverse_erfc simp-inverse-erfc operators)
2899 ;;; Inverse Complementary Error function distributes over bags
2901 (defprop %inverse_erfc (mlist $matrix mequal) distribute_over)
2903 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2905 ;;; inverse_erfc is the inverse of the erfc function
2907 (defprop %inverse_erfc %erfc $inverse)
2908 (defprop %erfc %inverse_erfc $inverse)
2911 ;;; Differentiation of the Inverse Complementary Error function
2913 (defprop %inverse_erfc
2914 ((z)
2915 ((mtimes)
2916 ((rat) -1 2)
2917 ((mexpt) $%pi ((rat) 1 2))
2918 ((mexpt) $%e ((mexpt) ((%inverse_erfc) z) 2))))
2919 grad)
2921 ;;; Integral of the Inverse Complementary Error function
2923 (defprop %inverse_erfc
2924 ((z)
2925 ((mtimes)
2926 ((mexpt) $%pi ((rat) -1 2))
2927 ((mexpt) $%e
2928 ((mtimes) -1 ((mexpt) ((%inverse_erfc) z) 2)))))
2929 integral)
2931 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2933 ;;; We support a simplim%function. The function is looked up in simplimit and
2934 ;;; handles specific values of the function.
2936 (defprop %inverse_erfc simplim%inverse_erfc simplim%function)
2938 (defun simplim%inverse_erfc (expr var val)
2939 ;; Look for the limit of the argument.
2940 (let ((z (limit (cadr expr) var val 'think)))
2941 (cond
2942 ;; Handle an argument 0 at this place
2943 ((or (zerop1 z)
2944 (eq z '$zeroa)
2945 (eq z '$zerob))
2946 '$inf)
2947 ((zerop1 (add z -2)) '$minf)
2949 ;; All other cases are handled by the simplifier of the function.
2950 (simplify (list '(%inverse_erfc) z))))))
2952 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2954 (defun simp-inverse-erfc (expr z simpflag)
2955 (oneargcheck expr)
2956 (setq z (simpcheck (cadr expr) simpflag))
2957 (cond
2958 ((or (zerop1 z)
2959 (zerop1 (add z -2)))
2960 (simp-domain-error
2961 (intl:gettext "inverse_erfc: inverse_erfc(~:M) is undefined.") z))
2962 ((onep1 z) 0)
2963 ((numerical-eval-p z)
2964 (to (bigfloat::bf-inverse-erfc (bigfloat:to z))))
2965 ((taylorize (mop expr) (cadr expr)))
2967 (eqtest (list '(%inverse_erfc) z) expr))))
2969 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
2971 ;;; Implementation of the Newton algorithm to calculate numerical values
2972 ;;; of the Inverse Error functions in float or bigfloat precision.
2973 ;;; The algorithm is a modified version of the routine in newton1.mac.
2975 (defvar *debug-newton* nil)
2976 (defvar *newton-maxcount* 1000)
2977 (defvar *newton-epsilon-factor* 50)
2978 (defvar *newton-epsilon-factor-float* 10)
2980 (defun float-newton (expr var x0 eps)
2981 (do ((s (sdiff expr var))
2982 (xn x0)
2983 (sn)
2984 (count 0 (+ count 1)))
2985 ((> count *newton-maxcount*)
2986 (merror
2987 (intl:gettext "float-newton: Newton does not converge for ~:M") expr))
2988 (setq sn ($float (maxima-substitute xn var expr)))
2989 (when (< (abs sn) eps) (return xn))
2990 (when *debug-newton* (format t "~&xn = ~A~%" xn))
2991 (setq xn ($float (sub xn (div sn (maxima-substitute xn var s)))))))
2993 (defun bfloat-newton (expr var x0 eps)
2994 (do ((s (sdiff expr var))
2995 (xn x0)
2996 (sn)
2997 (count 0 (+ count 1)))
2998 ((> count *newton-maxcount*)
2999 (merror
3000 (intl:gettext "bfloat-newton: Newton does not converge for ~:M") expr))
3001 (setq sn ($bfloat (maxima-substitute xn var expr)))
3002 (when (eq ($sign (sub (simplify (list '(mabs) sn)) eps)) '$neg)
3003 (return xn))
3004 (when *debug-newton* (format t "~&xn = ~A~%" xn))
3005 (setq xn ($bfloat (sub xn (div sn (maxima-substitute xn var s)))))))
3008 (in-package :bigfloat)
3010 ;; Compute inverse_erf(z) for z a real or complex number, including
3011 ;; bigfloat objects. The value is computing using a Newton iteration
3012 ;; to solve erf(x) = z.
3013 (defun bf-inverse-erf (z)
3014 (cond ((zerop z)
3016 ((= (abs z) 1)
3017 (maxima::merror
3018 (intl:gettext "bf-inverse-erf: inverse_erf(~M) is undefined")
3020 ((minusp (realpart z))
3021 ;; inverse_erf is odd because erf is.
3022 (- (bf-inverse-erf (- z))))
3024 (labels
3025 ((approx (z)
3026 ;; Find an approximate solution for x = inverse_erf(z).
3027 (let ((result
3028 (cond ((<= (abs z) 1)
3029 ;; For small z, inverse_erf(z) = z*sqrt(%pi)/2
3030 ;; + O(z^3). Thus, x = z*sqrt(%pi)/2 is our
3031 ;; initial starting point.
3032 (* z (sqrt (%pi z)) 1/2))
3034 ;; For |z| > 1 and realpart(z) >= 0, we have
3035 ;; the asymptotic series z = erf(x) = 1 -
3036 ;; exp(-x^2)/x/sqrt(%pi).
3038 ;; Then
3039 ;; x = sqrt(-log(x*sqrt(%pi)*(1-z))
3041 ;; We can use this as a fixed-point iteration
3042 ;; to find x, and we can start the iteration at
3043 ;; x = 1. Just do one more iteration. I (RLT)
3044 ;; think that's close enough to get the Newton
3045 ;; algorithm to converge.
3046 (let* ((sp (sqrt (%pi z)))
3047 (a (sqrt (- (log (* sp (- 1 z)))))))
3048 (setf a (sqrt (- (log (* a sp (- 1 z))))))
3049 (setf a (sqrt (- (log (* a sp (- 1 z)))))))))))
3050 (when maxima::*debug-newton*
3051 (format t "approx = ~S~%" result))
3052 result)))
3053 (let ((two/sqrt-pi (/ 2 (sqrt (%pi z))))
3054 (eps
3055 ;; Try to pick a reasonable epsilon value for the
3056 ;; Newton iteration.
3057 (cond ((<= (abs z) 1)
3058 (typecase z
3059 (cl:real (* maxima::*newton-epsilon-factor-float*
3060 maxima::flonum-epsilon))
3061 (t (* maxima::*newton-epsilon-factor* (epsilon z)))))
3063 (* maxima::*newton-epsilon-factor* (epsilon z))))))
3064 (when maxima::*debug-newton*
3065 (format t "eps = ~S~%" eps))
3066 (flet ((diff (x)
3067 ;; Derivative of erf(x)
3068 (* two/sqrt-pi (exp (- (* x x))))))
3069 (bf-newton #'bf-erf
3070 #'diff
3072 (approx z)
3073 eps)))))))
3075 (defun bf-inverse-erfc (z)
3076 (cond ((zerop z)
3077 (maxima::merror
3078 (intl:gettext "bf-inverse-erfc: inverse_erfc(~M) is undefined")
3080 ((= z 1)
3081 (float 0 z))
3083 (flet
3084 ((approx (z)
3085 ;; Find an approximate solution for x =
3086 ;; inverse_erfc(z). We have inverse_erfc(z) =
3087 ;; inverse_erf(1-z), so that's a good starting point.
3088 ;; We don't need full precision, so a float value is
3089 ;; good enough. But if 1-z is 1, inverse_erf is
3090 ;; undefined, so we need to do something else.
3091 (let ((result
3092 (let ((1-z (float (- 1 z) 0.0)))
3093 (cond ((= 1 1-z)
3094 (if (minusp (realpart z))
3095 (bf-inverse-erf (+ 1 (* 5 maxima::flonum-epsilon)))
3096 (bf-inverse-erf (- 1 (* 5 maxima::flonum-epsilon)))))
3098 (bf-inverse-erf 1-z))))))
3099 (when maxima::*debug-newton*
3100 (format t "approx = ~S~%" result))
3101 result)))
3102 (let ((-two/sqrt-pi (/ -2 (sqrt (%pi z))))
3103 (eps (* maxima::*newton-epsilon-factor* (epsilon z))))
3104 (when maxima::*debug-newton*
3105 (format t "eps = ~S~%" eps))
3106 (flet ((diff (x)
3107 ;; Derivative of erfc(x)
3108 (* -two/sqrt-pi (exp (- (* x x))))))
3109 (bf-newton #'bf-erfc
3110 #'diff
3112 (approx z)
3113 eps)))))))
3115 ;; Newton iteration for solving f(x) = z, given f and the derivative
3116 ;; of f.
3117 (defun bf-newton (f df z start eps)
3118 (do ((x start)
3119 (delta (/ (- (funcall f start) z)
3120 (funcall df start))
3121 (/ (- (funcall f x) z)
3122 (funcall df x)))
3123 (count 0 (1+ count)))
3124 ((or (< (abs delta) (* (abs x) eps))
3125 (> count maxima::*newton-maxcount*))
3126 (if (> count maxima::*newton-maxcount*)
3127 (maxima::merror
3128 (intl:gettext "bf-newton: failed to converge after ~M iterations: delta = ~S, x = ~S eps = ~S")
3129 count delta x eps)
3131 (when maxima::*debug-newton*
3132 (format t "x = ~S, abs(delta) = ~S relerr = ~S~%"
3133 x (abs delta) (/ (abs delta) (abs x))))
3134 (setf x (- x delta))))
3136 (in-package :maxima)
3138 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3140 ;;; Implementation of the Fresnel Integral S(z)
3142 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3144 (defmfun $fresnel_s (z)
3145 (simplify (list '(%fresnel_s) z)))
3147 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3149 ;;; Set properties to give full support to the parser and display
3151 (defprop $fresnel_s %fresnel_s alias)
3152 (defprop $fresnel_s %fresnel_s verb)
3154 (defprop %fresnel_s $fresnel_s reversealias)
3155 (defprop %fresnel_s $fresnel_s noun)
3157 ;;; fresnel_s is a simplifying function
3159 (defprop %fresnel_s simp-fresnel-s operators)
3161 ;;; fresnel_s distributes over bags
3163 (defprop %fresnel_s (mlist $matrix mequal) distribute_over)
3165 ;;; fresnel_s has mirror symmetry
3167 (defprop %fresnel_s t commutes-with-conjugate)
3169 ;;; fresnel_s is an odd function
3171 ;;; Maxima has two mechanism to define a reflection property
3172 ;;; 1. Declare the feature oddfun or evenfun
3173 ;;; 2. Put a reflection rule on the property list
3175 ;;; The second way is used for the trig functions. We use it here too.
3177 ;;; This would be the first way to give the property of an odd function.
3178 ;(eval-when
3179 ; #+gcl (load eval)
3180 ; #-gcl (:load-toplevel :execute)
3181 ; (let (($context '$global) (context '$global))
3182 ; (meval '(($declare) %fresnel_s $oddfun))
3183 ; ;; Let's remove built-in symbols from list for user-defined properties.
3184 ; (setq $props (remove '%fresnel_s $props))))
3186 (defprop %fresnel_s odd-function-reflect reflection-rule)
3188 ;;; Differentiation of the Fresnel Integral S
3190 (defprop %fresnel_s
3191 ((z)
3192 ((%sin) ((mtimes) ((rat) 1 2) $%pi ((mexpt) z 2))))
3193 grad)
3195 ;;; Integration of the Fresnel Integral S
3197 (defprop %fresnel_s
3198 ((z)
3199 ((mplus)
3200 ((mtimes) z ((%fresnel_s) z))
3201 ((mtimes)
3202 ((mexpt) $%pi -1)
3203 ((%cos) ((mtimes) ((rat) 1 2) $%pi ((mexpt) z 2))))))
3204 integral)
3206 ;;; Limits of the Fresnel Integral S
3208 (defprop %fresnel_s simplim%fresnel_s simplim%function)
3209 (defun simplim%fresnel_s (exp var val)
3210 (let ((arg (limit (cadr exp) var val 'think)))
3211 (cond ((eq arg '$inf)
3212 '((rat simp) 1 2))
3213 ((eq arg '$minf)
3214 '((rat simp) -1 2))
3216 `((%fresnel_s) ,arg)))))
3218 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3220 (defvar *fresnel-maxit* 1000)
3221 (defvar *fresnel-eps* (* 2 flonum-epsilon))
3222 (defvar *fresnel-min* 1e-32)
3224 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3225 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3227 (in-package :bigfloat)
3229 (defun bf-fresnel (z)
3230 (let* ((eps (epsilon z))
3231 (maxit maxima::*fresnel-maxit*)
3232 (xmin 1.5)
3233 (bf-pi (%pi z))
3234 ;; For very small x, we have
3235 ;; fresnel_s(x) = %pi/6*z^3
3236 ;; fresnel_c(x) = x
3237 (s (* (/ bf-pi 6) z z z))
3238 (c z))
3239 (when (> (abs z) eps)
3240 (cond
3241 ((< (abs z) xmin)
3242 (when maxima::*debug-gamma*
3243 (format t "~& in FRESNEL series expansion.~%"))
3244 (let ((sums 0) (sumc z))
3245 (do ((sum 0)
3246 (sign 1)
3247 (fact (* (/ bf-pi 2) (* z z)))
3248 (term z)
3249 (odd t (not odd))
3250 (test 0)
3251 (n 3 (+ n 2))
3252 (k 1 (+ k 1)))
3253 ((> k maxit)
3254 (maxima::merror (intl:gettext "fresnel: series expansion failed for (COMPLEX-BFLOAT-FRESNEL ~:M).") z))
3255 (setq term (* term (/ fact k)))
3256 (setq sum (+ sum (/ (* sign term) n)))
3257 (setq test (* (abs sum) eps))
3258 (if odd
3259 (progn
3260 (setq sign (- sign))
3261 (setq sums sum)
3262 (setq sum sumc))
3263 (progn
3264 (setq sumc sum)
3265 (setq sum sums)))
3266 (if (< (abs term) test)
3267 (return)))
3268 (setq s sums)
3269 (setq c sumc)))
3271 (let* ((sqrt-pi (sqrt bf-pi))
3272 (z+ (* (complex 1/2 1/2)
3273 (* sqrt-pi
3274 z)))
3275 (z- (* (complex 1/2 -1/2)
3276 (* sqrt-pi
3277 z)))
3278 (erf+ (bf-erf z+))
3279 (erf- (bf-erf z-)))
3280 (setq s (* (complex 1/4 1/4)
3281 (+ erf+ (* (complex 0 -1) erf-))))
3282 (setq c (* (complex 1/4 -1/4)
3283 (+ erf+ (* (complex 0 1) erf-))))))))
3284 (values s c)))
3286 (defun bf-fresnel-s (z)
3287 (if (and (complexp z) (zerop (realpart z)))
3288 ;; A pure imaginary argument. Use fresnel_s(%i*x)=-%i*fresnel_s(x).
3289 (complex 0
3290 (- (bf-fresnel-s (imagpart z))))
3291 (let ((fs (bf-fresnel z)))
3292 (if (realp z)
3293 (realpart fs)
3294 fs))))
3296 (defun bf-fresnel-c (z)
3297 (if (and (complexp z) (zerop (realpart z)))
3298 ;; A pure imaginary argument. Use fresnel_c(%i*x)=%i*fresnel_c(x).
3299 (complex 0
3300 (bf-fresnel-c (imagpart z)))
3301 (let ((fc (nth-value 1 (bf-fresnel z))))
3302 (if (realp z)
3303 (realpart fc)
3304 fc))))
3306 (in-package :maxima)
3307 (defun simp-fresnel-s (expr z simpflag)
3308 (oneargcheck expr)
3309 (setq z (simpcheck (cadr expr) simpflag))
3310 (cond
3312 ;; Check for specific values
3314 ((zerop1 z) z)
3315 ((eq z '$inf) '((rat simp) 1 2))
3316 ((eq z '$minf) '((rat simp) -1 2))
3318 ;; Check for numerical evaluation
3319 ((numerical-eval-p z)
3320 (to (bigfloat::bf-fresnel-s (bigfloat::to z))))
3322 ;; Check for argument simplification
3324 ((taylorize (mop expr) (second expr)))
3325 ((and $%iargs (multiplep z '$%i))
3326 (mul -1 '$%i (simplify (list '(%fresnel_s) (coeff z '$%i 1)))))
3327 ((apply-reflection-simp (mop expr) z $trigsign))
3329 ;; Representation through equivalent functions
3331 ($erf_representation
3332 (mul
3333 (div (add 1 '$%i) 4)
3334 (add
3335 (simplify
3336 (list
3337 '(%erf)
3338 (mul (div (add 1 '$%i) 2) (power '$%pi '((rat simp) 1 2)) z)))
3339 (mul -1 '$%i
3340 (simplify
3341 (list
3342 '(%erf)
3343 (mul (div (sub 1 '$%i) 2)
3344 (power '$%pi '((rat simp) 1 2)) z)))))))
3346 ($hypergeometric_representation
3347 (mul (div (mul '$%pi (power z 3)) 6)
3348 (take '($hypergeometric)
3349 (list '(mlist) (div 3 4))
3350 (list '(mlist) (div 3 2) (div 7 4))
3351 (mul -1 (div (mul (power '$%pi 2) (power z 4)) 16)))))
3354 (eqtest (list '(%fresnel_s) z) expr))))
3356 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3358 ;;; Implementation of the Fresnel Integral C(z)
3360 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3362 (defmfun $fresnel_c (z)
3363 (simplify (list '(%fresnel_c) z)))
3365 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3367 ;;; Set properties to give full support to the parser and display
3369 (defprop $fresnel_c %fresnel_c alias)
3370 (defprop $fresnel_c %fresnel_c verb)
3372 (defprop %fresnel_c $fresnel_c reversealias)
3373 (defprop %fresnel_c $fresnel_c noun)
3375 ;;; fresnel_c is a simplifying function
3377 (defprop %fresnel_c simp-fresnel-c operators)
3379 ;;; fresnel_c distributes over bags
3381 (defprop %fresnel_c (mlist $matrix mequal) distribute_over)
3383 ;;; fresnel_c has mirror symmetry
3385 (defprop %fresnel_c t commutes-with-conjugate)
3387 ;;; fresnel_c is an odd function
3388 ;;; Maxima has two mechanism to define a reflection property
3389 ;;; 1. Declare the feature oddfun or evenfun
3390 ;;; 2. Put a reflection rule on the property list
3392 ;;; The second way is used for the trig functions. We use it here too.
3394 (defprop %fresnel_c odd-function-reflect reflection-rule)
3396 ;;; Differentiation of the Fresnel Integral C
3398 (defprop %fresnel_c
3399 ((z)
3400 ((%cos) ((mtimes) ((rat) 1 2) $%pi ((mexpt) z 2))))
3401 grad)
3403 ;;; Integration of the Fresnel Integral C
3405 (defprop %fresnel_c
3406 ((z)
3407 ((mplus)
3408 ((mtimes) z ((%fresnel_c) z))
3409 ((mtimes) -1
3410 ((mexpt) $%pi -1)
3411 ((%sin) ((mtimes) ((rat) 1 2) $%pi ((mexpt) z 2))))))
3412 integral)
3414 ;;; Limits of the Fresnel Integral C
3416 (defprop %fresnel_c simplim%fresnel_c simplim%function)
3417 (defun simplim%fresnel_c (exp var val)
3418 (let ((arg (limit (cadr exp) var val 'think)))
3419 (cond ((eq arg '$inf)
3420 '((rat simp) 1 2))
3421 ((eq arg '$minf)
3422 '((rat simp) -1 2))
3424 `((%fresnel_c) ,arg)))))
3426 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3428 (defun simp-fresnel-c (expr z simpflag)
3429 (oneargcheck expr)
3430 (setq z (simpcheck (cadr expr) simpflag))
3431 (cond
3433 ;; Check for specific values
3435 ((zerop1 z) z)
3436 ((eq z '$inf) '((rat simp) 1 2))
3437 ((eq z '$minf) '((rat simp) -1 2))
3439 ;; Check for numerical evaluation
3440 ((numerical-eval-p z)
3441 (to (bigfloat::bf-fresnel-c (bigfloat::to z))))
3444 ;; Check for argument simplification
3446 ((taylorize (mop expr) (second expr)))
3447 ((and $%iargs (multiplep z '$%i))
3448 (mul '$%i (simplify (list '(%fresnel_c) (coeff z '$%i 1)))))
3449 ((apply-reflection-simp (mop expr) z $trigsign))
3451 ;; Representation through equivalent functions
3453 ($erf_representation
3454 (mul
3455 (div (sub 1 '$%i) 4)
3456 (add
3457 (simplify
3458 (list
3459 '(%erf)
3460 (mul (div (add 1 '$%i) 2) (power '$%pi '((rat simp) 1 2)) z)))
3461 (mul '$%i
3462 (simplify
3463 (list
3464 '(%erf)
3465 (mul (div (sub 1 '$%i) 2)
3466 (power '$%pi '((rat simp) 1 2)) z)))))))
3468 ($hypergeometric_representation
3469 (mul z
3470 (take '($hypergeometric)
3471 (list '(mlist) (div 1 4))
3472 (list '(mlist) (div 1 2) (div 5 4))
3473 (mul -1 (div (mul (power '$%pi 2) (power z 4)) 16)))))
3476 (eqtest (list '(%fresnel_c) z) expr))))
3479 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3481 ;;; Implementation of the Beta function
3483 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3485 ;;; The code for the implementation of the beta function is in the files
3486 ;;; csimp2.lisp, simp.lisp and mactex.lisp.
3487 ;;; At this place we only implement the operator property SYMMETRIC.
3489 ;;; Beta is symmetric beta(a,b) = beta(b,a)
3491 (eval-when
3492 #+gcl (load eval)
3493 #-gcl (:load-toplevel :execute)
3494 (let (($context '$global) (context '$global))
3495 (meval '(($declare) $beta $symmetric))
3496 ;; Let's remove built-in symbols from list for user-defined properties.
3497 (setq $props (remove '$beta $props))))
3499 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3501 ;;; Implementation of the Incomplete Beta function
3503 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3505 (defvar *beta-incomplete-maxit* 10000)
3506 (defvar *beta-incomplete-eps* 1.0e-15)
3508 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3510 (defmfun $beta_incomplete (a b z)
3511 (simplify (list '(%beta_incomplete) a b z)))
3513 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3515 (defprop $beta_incomplete %beta_incomplete alias)
3516 (defprop $beta_incomplete %beta_incomplete verb)
3518 (defprop %beta_incomplete $beta_incomplete reversealias)
3519 (defprop %beta_incomplete $beta_incomplete noun)
3521 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3523 (defprop %beta_incomplete simp-beta-incomplete operators)
3525 ;;; beta_incomplete distributes over bags
3527 (defprop %beta_incomplete (mlist $matrix mequal) distribute_over)
3529 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3531 (defprop %beta_incomplete
3532 ((a b z)
3533 ;; Derivative wrt a
3534 ((mplus)
3535 ((%beta_incomplete) a b z)
3536 ((mtimes) -1
3537 ((mexpt) ((%gamma) a) 2)
3538 (($hypergeometric_regularized)
3539 ((mlist) a a ((mplus) 1 ((mtimes) -1 b)))
3540 ((mlist) ((mplus) 1 a) ((mplus) 1 a))
3542 ((mexpt) z a)))
3543 ;; Derivative wrt b
3544 ((mplus)
3545 ((mtimes)
3546 (($beta) a b)
3547 ((mplus)
3548 ((mqapply)
3549 (($psi array 0) b)
3550 ((mtimes) -1 ((mqapply) (($psi array) 0) ((mplus) a b)))))
3551 ((mtimes) -1
3552 ((%beta_incomplete) b a ((mplus) 1 ((mtimes) -1 z)))
3553 ((%log) ((mplus) 1 ((mtimes) -1 z))))
3554 ((mtimes)
3555 ((mexpt) ((%gamma) b) 2)
3556 (($hypergeometric_regularized)
3557 ((mlist) b b ((mplus) 1 ((mtimes) -1 a)))
3558 ((mlist) ((mplus) 1 b) ((mplus) 1 b))
3559 ((mplus) 1 ((mtimes) -1 z)))
3560 ((mexpt) ((mplus) 1 ((mtimes) -1 z)) b))))
3561 ;; The derivative wrt z
3562 ((mtimes)
3563 ((mexpt) ((mplus) 1 ((mtimes) -1 z)) ((mplus) -1 b))
3564 ((mexpt) z ((mplus) -1 a))))
3565 grad)
3567 ;;; Integral of the Incomplete Beta function
3569 (defprop %beta_incomplete
3570 ((a b z)
3571 nil
3573 ((mplus)
3574 ((mtimes) -1 ((%beta_incomplete) ((mplus) 1 a) b z))
3575 ((mtimes) ((%beta_incomplete) a b z) z)))
3576 integral)
3578 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3580 (defun simp-beta-incomplete (expr ignored simpflag)
3581 (declare (ignore ignored))
3582 (arg-count-check 3 expr)
3583 (let ((a (simpcheck (cadr expr) simpflag))
3584 (b (simpcheck (caddr expr) simpflag))
3585 (z (simpcheck (cadddr expr) simpflag))
3586 ($simpsum t))
3587 (when *debug-gamma*
3588 (format t "~&SIMP-BETA-INCOMPLETE:~%")
3589 (format t "~& : a = ~A~%" a)
3590 (format t "~& : b = ~A~%" b)
3591 (format t "~& : z = ~A~%" z))
3592 (cond
3594 ;; Check for specific values
3596 ((zerop1 z)
3597 (let ((sgn ($sign ($realpart a))))
3598 (cond ((member sgn '($neg $zero))
3599 (simp-domain-error
3600 (intl:gettext
3601 "beta_incomplete: beta_incomplete(~:M,~:M,~:M) is undefined.")
3602 a b z))
3603 ((member sgn '($pos $pz))
3606 (eqtest (list '(%beta_incomplete) a b z) expr)))))
3608 ((and (onep1 z) (eq ($sign ($realpart b)) '$pos))
3609 ;; z=1 and realpart(b)>0. Simplify to a Beta function.
3610 ;; If we have to evaluate numerically preserve the type.
3611 (cond
3612 ((complex-float-numerical-eval-p a b z)
3613 (simplify (list '($beta) ($float a) ($float b))))
3614 ((complex-bigfloat-numerical-eval-p a b z)
3615 (simplify (list '($beta) ($bfloat a) ($bfloat b))))
3617 (simplify (list '($beta) a b)))))
3619 ((or (zerop1 a)
3620 (and (integer-representation-p a)
3621 (eq ($sign a) '$neg)
3622 (or (and (mnump b)
3623 (not (integer-representation-p b)))
3624 (eq ($sign (add a b)) '$pos))))
3625 ;; The argument a is zero or a is negative and the argument b is
3626 ;; not in a valid range. Beta incomplete is undefined.
3627 ;; It would be more correct to return Complex infinity.
3628 (simp-domain-error
3629 (intl:gettext
3630 "beta_incomplete: beta_incomplete(~:M,~:M,~:M) is undefined.") a b z))
3632 ;; Check for numerical evaluation in Float or Bigfloat precision
3634 ((complex-float-numerical-eval-p a b z)
3635 (cond
3636 ((not (and (integer-representation-p a) (< a 0.0)))
3637 (let ((*beta-incomplete-eps* (bigfloat:epsilon ($float 1.0))))
3638 (beta-incomplete ($float a) ($float b) ($float z))))
3640 ;; Negative integer a and b is in a valid range. Expand.
3641 ($rectform
3642 (beta-incomplete-expand-negative-integer
3643 (truncate a) ($float b) ($float z))))))
3645 ((complex-bigfloat-numerical-eval-p a b z)
3646 (cond
3647 ((not (and (integer-representation-p a) (eq ($sign a) '$neg)))
3648 (let ((*beta-incomplete-eps*
3649 (bigfloat:epsilon (bigfloat:bigfloat 1.0))))
3650 (beta-incomplete ($bfloat a) ($bfloat b) ($bfloat z))))
3652 ;; Negative integer a and b is in a valid range. Expand.
3653 ($rectform
3654 (beta-incomplete-expand-negative-integer
3655 ($truncate a) ($bfloat b) ($bfloat z))))))
3657 ;; Argument simplifications and transformations
3659 ((and (integerp b)
3660 (plusp b)
3661 (or (not (integerp a))
3662 (plusp a)))
3663 ;; Expand for b a positive integer and a not a negative integer.
3664 (mul
3665 (simplify (list '($beta) a b))
3666 (power z a)
3667 (let ((index (gensumindex)))
3668 (simpsum1
3669 (div
3670 (mul
3671 (simplify (list '($pochhammer) a index))
3672 (power (sub 1 z) index))
3673 (simplify (list '(mfactorial) index)))
3674 index 0 (sub b 1)))))
3676 ((and (integerp a) (plusp a))
3677 ;; Expand for a a positive integer.
3678 (mul
3679 (simplify (list '($beta) a b))
3680 (sub 1
3681 (mul
3682 (power (sub 1 z) b)
3683 (let ((index (gensumindex)))
3684 (simpsum1
3685 (div
3686 (mul
3687 (simplify (list '($pochhammer) b index))
3688 (power z index))
3689 (simplify (list '(mfactorial) index)))
3690 index 0 (sub a 1)))))))
3692 ((and (integerp a) (minusp a) (integerp b) (plusp b) (<= b (- a)))
3693 ;; Expand for a a negative integer and b an integer with b <= -a.
3694 (mul
3695 (power z a)
3696 (let ((index (gensumindex)))
3697 (simpsum1
3698 (div
3699 (mul (simplify (list '($pochhammer) (sub 1 b) index))
3700 (power z index))
3701 (mul (add index a)
3702 (simplify (list '(mfactorial) index))))
3703 index 0 (sub b 1)))))
3705 ((and $beta_expand (mplusp a) (integerp (cadr a)) (plusp (cadr a)))
3706 (let ((n (cadr a))
3707 (a (simplify (cons '(mplus) (cddr a)))))
3708 (sub
3709 (mul
3710 (div
3711 (simplify (list '($pochhammer) a n))
3712 (simplify (list '($pochhammer) (add a b) n)))
3713 ($beta_incomplete a b z))
3714 (mul
3715 (power (add a b n -1) -1)
3716 (let ((index (gensumindex)))
3717 (simpsum1
3718 (mul
3719 (div
3720 (simplify (list '($pochhammer)
3721 (add 1 (mul -1 a) (mul -1 n))
3722 index))
3723 (simplify (list '($pochhammer)
3724 (add 2 (mul -1 a) (mul -1 b) (mul -1 n))
3725 index)))
3726 (mul (power (sub 1 z) b)
3727 (power z (add a n (mul -1 index) -1))))
3728 index 0 (sub n 1)))))))
3730 ((and $beta_expand (mplusp a) (integerp (cadr a)) (minusp (cadr a)))
3731 (let ((n (- (cadr a)))
3732 (a (simplify (cons '(mplus) (cddr a)))))
3733 (sub
3734 (mul
3735 (div
3736 (simplify (list '($pochhammer) (add 1 (mul -1 a) (mul -1 b)) n))
3737 (simplify (list '($pochhammer) (sub 1 a) n)))
3738 ($beta_incomplete a b z))
3739 (mul
3740 (div
3741 (simplify
3742 (list '($pochhammer) (add 2 (mul -1 a) (mul -1 b)) (sub n 1)))
3743 (simplify (list '($pochhammer) (sub 1 a) n)))
3744 (let ((index (gensumindex)))
3745 (simpsum1
3746 (mul
3747 (div
3748 (simplify (list '($pochhammer) (sub 1 a) index))
3749 (simplify (list '($pochhammer)
3750 (add 2 (mul -1 a) (mul -1 b))
3751 index)))
3752 (mul (power (sub 1 z) b)
3753 (power z (add a (mul -1 index) -1))))
3754 index 0 (sub n 1)))))))
3757 (eqtest (list '(%beta_incomplete) a b z) expr)))))
3759 (defun beta-incomplete-expand-negative-integer (a b z)
3760 (mul
3761 (power z a)
3762 (let ((index (gensumindex)) ($simpsum t))
3763 (simpsum1
3764 (div
3765 (mul (simplify (list '($pochhammer) (sub 1 b) index))
3766 (power z index))
3767 (mul (add index a) (simplify (list '(mfactorial) index))))
3768 index 0 (sub b 1)))))
3770 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3772 (defun beta-incomplete (a b z)
3773 (cond
3774 ((eq ($sign (sub (mul ($realpart z) ($realpart (add a b 2)))
3775 ($realpart (add a 1))))
3776 '$pos)
3777 ($rectform
3778 (sub
3779 (simplify (list '($beta) a b))
3780 (to (numeric-beta-incomplete b a (sub 1.0 z))))))
3782 (to (numeric-beta-incomplete a b z)))))
3784 (defun numeric-beta-incomplete (a b z)
3785 (when *debug-gamma*
3786 (format t "~&NUMERIC-BETA-INCOMPLETE enters continued fractions~%"))
3787 (let ((a (bigfloat:to a))
3788 (b (bigfloat:to b))
3789 (z (bigfloat:to z)))
3790 (do* ((beta-maxit *beta-incomplete-maxit*)
3791 (beta-eps *beta-incomplete-eps*)
3792 (beta-min (bigfloat:* beta-eps beta-eps))
3793 (ab (bigfloat:+ a b))
3794 (ap (bigfloat:+ a 1.0))
3795 (am (bigfloat:- a 1.0))
3796 (c 1.0)
3797 (d (bigfloat:- 1.0 (bigfloat:/ (bigfloat:* z ab) ap)))
3798 (d (if (bigfloat:< (bigfloat:abs d) beta-min) beta-min d))
3799 (d (bigfloat:/ 1.0 d))
3800 (h d)
3801 (aa 0.0)
3802 (de 0.0)
3803 (m2 0)
3804 (m 1 (+ m 1)))
3805 ((> m beta-maxit)
3806 (merror (intl:gettext "beta_incomplete: continued fractions failed for beta_incomplete(~:M, ~:M, ~:M).") a b z))
3807 (setq m2 (+ m m))
3808 (setq aa (bigfloat:/ (bigfloat:* m z (bigfloat:- b m))
3809 (bigfloat:* (bigfloat:+ am m2)
3810 (bigfloat:+ a m2))))
3811 (setq d (bigfloat:+ 1.0 (bigfloat:* aa d)))
3812 (when (bigfloat:< (bigfloat:abs d) beta-min) (setq d beta-min))
3813 (setq c (bigfloat:+ 1.0 (bigfloat:/ aa c)))
3814 (when (bigfloat:< (bigfloat:abs c) beta-min) (setq c beta-min))
3815 (setq d (bigfloat:/ 1.0 d))
3816 (setq h (bigfloat:* h d c))
3817 (setq aa (bigfloat:/ (bigfloat:* -1
3818 (bigfloat:+ a m)
3819 (bigfloat:+ ab m) z)
3820 (bigfloat:* (bigfloat:+ a m2)
3821 (bigfloat:+ ap m2))))
3822 (setq d (bigfloat:+ 1.0 (bigfloat:* aa d)))
3823 (when (bigfloat:< (bigfloat:abs d) beta-min) (setq d beta-min))
3824 (setq c (bigfloat:+ 1.0 (bigfloat:/ aa c)))
3825 (when (bigfloat:< (bigfloat:abs c) beta-min) (setq c beta-min))
3826 (setq d (bigfloat:/ 1.0 d))
3827 (setq de (bigfloat:* d c))
3828 (setq h (bigfloat:* h de))
3829 (when (bigfloat:< (bigfloat:abs (bigfloat:- de 1.0)) beta-eps)
3830 (when *debug-gamma*
3831 (format t "~&Continued fractions finished.~%")
3832 (format t "~& z = ~A~%" z)
3833 (format t "~& a = ~A~%" a)
3834 (format t "~& b = ~A~%" b)
3835 (format t "~& h = ~A~%" h))
3836 (return
3837 (let ((result
3838 (bigfloat:/
3839 (bigfloat:* h
3840 (bigfloat:expt z a)
3841 (bigfloat:expt (bigfloat:- 1.0 z) b)) a)))
3842 (when *debug-gamma*
3843 (format t "~& result = ~A~%" result))
3844 result))))))
3846 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3848 ;;; Implementation of the Generalized Incomplete Beta function
3850 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3852 (defmfun $beta_incomplete_generalized (a b z1 z2)
3853 (simplify (list '(%beta_incomplete_generalized) a b z1 z2)))
3855 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3857 (defprop $beta_incomplete_generalized %beta_incomplete_generalized alias)
3858 (defprop $beta_incomplete_generalized %beta_incomplete_generalized verb)
3860 (defprop %beta_incomplete_generalized $beta_incomplete_generalized reversealias)
3861 (defprop %beta_incomplete_generalized $beta_incomplete_generalized noun)
3863 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3865 (defprop %beta_incomplete_generalized
3866 simp-beta-incomplete-generalized operators)
3868 ;;; beta_incomplete_generalized distributes over bags
3870 (defprop %beta_incomplete_generalized (mlist $matrix mequal) distribute_over)
3872 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3874 ;;; Generalized Incomplete Gamma function has not mirror symmetry for z1 or z2
3875 ;;; but not on the negative real axis and for z1 or z2 real and > 1.
3876 ;;; We support a conjugate-function which test these cases.
3878 (defprop %beta_incomplete_generalized
3879 conjugate-beta-incomplete-generalized conjugate-function)
3881 (defun conjugate-beta-incomplete-generalized (args)
3882 (let ((a (first args))
3883 (b (second args))
3884 (z1 (third args))
3885 (z2 (fourth args)))
3886 (cond ((and (off-negative-real-axisp z1)
3887 (not (and (zerop1 ($imagpart z1))
3888 (eq ($sign (sub ($realpart z1) 1)) '$pos)))
3889 (off-negative-real-axisp z2)
3890 (not (and (zerop1 ($imagpart z2))
3891 (eq ($sign (sub ($realpart z2) 1)) '$pos))))
3892 (simplify
3893 (list
3894 '(%beta_incomplete_generalized)
3895 (simplify (list '($conjugate) a))
3896 (simplify (list '($conjugate) b))
3897 (simplify (list '($conjugate) z1))
3898 (simplify (list '($conjugate) z2)))))
3900 (list
3901 '($conjugate simp)
3902 (simplify (list '(%beta_incomplete_generalized)
3903 a b z1 z2)))))))
3905 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3907 (defprop %beta_incomplete_generalized
3908 ((a b z1 z2)
3909 ;; Derivative wrt a
3910 ((mplus)
3911 ((mtimes) -1
3912 ((%beta_incomplete) a b z1)
3913 ((%log) z1))
3914 ((mtimes)
3915 ((mexpt) ((%gamma) a) 2)
3916 ((mplus)
3917 ((mtimes)
3918 (($hypergeometric_regularized)
3919 ((mlist) a a ((mplus) 1 ((mtimes) -1 b)))
3920 ((mlist) ((mplus) 1 a) ((mplus) 1 a))
3922 ((mexpt) z1 1))
3923 ((mtimes) -1
3924 (($hypergeometric_regularized)
3925 ((mlist) a a ((mplus) 1 ((mtimes) -1 b)))
3926 ((mlist) ((mplus) 1 a) ((mplus) 1 a))
3928 ((mexpt) z2 a))))
3929 ((mtimes) ((%beta_incomplete) a b z2) ((%log) z2)))
3930 ;; Derivative wrt b
3931 ((mplus)
3932 ((mtimes)
3933 ((%beta_incomplete) b a ((mplus) 1 ((mtimes) -1 z1)))
3934 ((%log) ((mplus) 1 ((mtimes) -1 z1))))
3935 ((mtimes) -1
3936 ((%beta_incomplete) b a ((mplus) 1 ((mtimes) -1 z2)))
3937 ((%log) ((mplus) 1 ((mtimes) -1 z2))))
3938 ((mtimes) -1
3939 ((mexpt) ((%gamma) b) 2)
3940 ((mplus)
3941 ((mtimes)
3942 (($hypergeometric_regularized)
3943 ((mlist) b b ((mplus) 1 ((mtimes) -1 a)))
3944 ((mlist) ((mplus) 1 b) ((mplus) 1 b))
3945 ((mplus) 1 ((mtimes) -1 z1)))
3946 ((mexpt) ((mplus) 1 ((mtimes) -1 z1)) b))
3947 ((mtimes) -1
3948 (($hypergeometric_regularized)
3949 ((mlist) b b ((mplus) 1 ((mtimes) -1 a)))
3950 ((mlist) ((mplus) 1 b) ((mplus) 1 b))
3951 ((mplus) 1 ((mtimes) -1 z2)))
3952 ((mexpt) ((mplus) 1 ((mtimes) -1 z2)) b)))))
3953 ;; The derivative wrt z1
3954 ((mtimes) -1
3955 ((mexpt)
3956 ((mplus) 1 ((mtimes) -1 z1))
3957 ((mplus) -1 b))
3958 ((mexpt) z1 ((mplus) -1 a)))
3959 ;; The derivative wrt z2
3960 ((mtimes)
3961 ((mexpt)
3962 ((mplus) 1 ((mtimes) -1 z2))
3963 ((mplus) -1 b))
3964 ((mexpt) z2 ((mplus) -1 a))))
3965 grad)
3967 ;;; Integral of the Incomplete Beta function
3969 (defprop %beta_incomplete_generalized
3970 ((a b z1 z2)
3971 nil
3973 ;; Integral wrt z1
3974 ((mplus)
3975 ((%beta_incomplete) ((mplus) 1 a) b z1)
3976 ((mtimes) ((%beta_incomplete_generalized) a b z1 z2) z1))
3977 ;; Integral wrt z2
3978 ((mplus)
3979 ((mtimes) -1
3980 ((%beta_incomplete) ((mplus) 1 a) b z2))
3981 ((mtimes) ((%beta_incomplete_generalized) a b z1 z2) z2)))
3982 integral)
3984 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
3986 (defun simp-beta-incomplete-generalized (expr ignored simpflag)
3987 (declare (ignore ignored))
3988 (arg-count-check 4 expr)
3989 (let ((a (simpcheck (second expr) simpflag))
3990 (b (simpcheck (third expr) simpflag))
3991 (z1 (simpcheck (fourth expr) simpflag))
3992 (z2 (simpcheck (fifth expr) simpflag))
3993 ($simpsum t))
3994 (cond
3996 ;; Check for specific values
3998 ((zerop1 z2)
3999 (let ((sgn ($sign ($realpart a))))
4000 (cond ((eq sgn '$neg)
4001 (simp-domain-error
4002 (intl:gettext
4003 "beta_incomplete_generalized: beta_incomplete_generalized(~:M,~:M,~:M,~:M) is undefined.")
4004 a b z1 z2))
4005 ((member sgn '($pos $pz))
4006 (mul -1 ($beta_incomplete a b z1)))
4008 (eqtest
4009 (list '(%beta_incomplete_generalized) a b z1 z2) expr)))))
4011 ((zerop1 z1)
4012 (let ((sgn ($sign ($realpart a))))
4013 (cond ((eq sgn '$neg)
4014 (simp-domain-error
4015 (intl:gettext
4016 "beta_incomplete_generalized: beta_incomplete_generalized(~:M,~:M,~:M,~:M) is undefined.")
4017 a b z1 z2))
4018 ((member sgn '($pos $pz))
4019 (mul -1 ($beta_incomplete a b z2)))
4021 (eqtest
4022 (list '(%beta_incomplete_generalized) a b z1 z2) expr)))))
4024 ((and (onep1 z2) (or (not (mnump a)) (not (mnump b)) (not (mnump z1))))
4025 (let ((sgn ($sign ($realpart b))))
4026 (cond ((member sgn '($pos $pz))
4027 (sub (simplify (list '($beta) a b))
4028 ($beta_incomplete a b z1)))
4030 (eqtest
4031 (list '(%beta_incomplete_generalized) a b z1 z2) expr)))))
4033 ((and (onep1 z1) (or (not (mnump a)) (not (mnump b)) (not (mnump z2))))
4034 (let ((sgn ($sign ($realpart b))))
4035 (cond ((member sgn '($pos $pz))
4036 (sub ($beta_incomplete a b z2)
4037 (simplify (list '($beta) a b))))
4039 (eqtest
4040 (list '(%beta_incomplete_generalized) a b z1 z2) expr)))))
4042 ;; Check for numerical evaluation in Float or Bigfloat precision
4044 ((complex-float-numerical-eval-p a b z1 z2)
4045 (let ((*beta-incomplete-eps* (bigfloat:epsilon ($float 1.0))))
4046 (sub (beta-incomplete ($float a) ($float b) ($float z2))
4047 (beta-incomplete ($float a) ($float b) ($float z1)))))
4049 ((complex-bigfloat-numerical-eval-p a b z1 z2)
4050 (let ((*beta-incomplete-eps*
4051 (bigfloat:epsilon (bigfloat:bigfloat 1.0))))
4052 (sub (beta-incomplete ($bfloat a) ($bfloat b) ($bfloat z2))
4053 (beta-incomplete ($bfloat a) ($bfloat b) ($bfloat z1)))))
4055 ;; Check for argument simplifications and transformations
4057 ((and (integerp a) (plusp a))
4058 (mul
4059 (simplify (list '($beta) a b))
4060 (sub
4061 (mul
4062 (power (sub 1 z1) b)
4063 (let ((index (gensumindex)))
4064 (simpsum1
4065 (div
4066 (mul
4067 (simplify (list '($pochhammer) b index))
4068 (power z1 index))
4069 (simplify (list '(mfactorial) index)))
4070 index 0 (sub a 1))))
4071 (mul
4072 (power (sub 1 z2) b)
4073 (let ((index (gensumindex)))
4074 (simpsum1
4075 (div
4076 (mul
4077 (simplify (list '($pochhammer) b index))
4078 (power z2 index))
4079 (simplify (list '(mfactorial) index)))
4080 index 0 (sub a 1)))))))
4082 ((and (integerp b) (plusp b))
4083 (mul
4084 (simplify (list '($beta) a b))
4085 (sub
4086 (mul
4087 (power z2 a)
4088 (let ((index (gensumindex)))
4089 (simpsum1
4090 (div
4091 (mul
4092 (simplify (list '($pochhammer) a index))
4093 (power (sub 1 z2) index))
4094 (simplify (list '(mfactorial) index)))
4095 index 0 (sub b 1))))
4096 (mul
4097 (power z1 a)
4098 (let ((index (gensumindex)))
4099 (simpsum1
4100 (div
4101 (mul
4102 (simplify (list '($pochhammer) a index))
4103 (power (sub 1 z1) index))
4104 (simplify (list '(mfactorial) index)))
4105 index 0 (sub b 1)))))))
4107 ((and $beta_expand (mplusp a) (integerp (cadr a)) (plusp (cadr a)))
4108 (let ((n (cadr a))
4109 (a (simplify (cons '(mplus) (cddr a)))))
4110 (add
4111 (mul
4112 (div
4113 (simplify (list '($pochhammer) a n))
4114 (simplify (list '($pochhammer) (add a b) n)))
4115 ($beta_incomplete_generalized a b z1 z2))
4116 (mul
4117 (power (add a b n -1) -1)
4118 (let ((index (gensumindex)))
4119 (simpsum1
4120 (mul
4121 (div
4122 (simplify (list '($pochhammer)
4123 (add 1 (mul -1 a) (mul -1 n))
4124 index))
4125 (simplify (list '($pochhammer)
4126 (add 2 (mul -1 a) (mul -1 b) (mul -1 n))
4127 index)))
4128 (sub
4129 (mul (power (sub 1 z1) b)
4130 (power z1 (add a n (mul -1 index) -1)))
4131 (mul (power (sub 1 z2) b)
4132 (power z2 (add a n (mul -1 index) -1)))))
4133 index 0 (sub n 1)))))))
4135 ((and $beta_expand (mplusp a) (integerp (cadr a)) (minusp (cadr a)))
4136 (let ((n (- (cadr a)))
4137 (a (simplify (cons '(mplus) (cddr a)))))
4138 (sub
4139 (mul
4140 (div
4141 (simplify (list '($pochhammer) (add 1 (mul -1 a) (mul -1 b)) n))
4142 (simplify (list '($pochhammer) (sub 1 a) n)))
4143 ($beta_incomplete_generalized a b z1 z2))
4144 (mul
4145 (div
4146 (simplify
4147 (list '($pochhammer) (add 2 (mul -1 a) (mul -1 b)) (sub n 1)))
4148 (simplify (list '($pochhammer) (sub 1 a) n)))
4149 (let ((index (gensumindex)))
4150 (simpsum1
4151 (mul
4152 (div
4153 (simplify (list '($pochhammer) (sub 1 a) index))
4154 (simplify (list '($pochhammer)
4155 (add 2 (mul -1 a) (mul -1 b))
4156 index)))
4157 (sub
4158 (mul (power (sub 1 z2) b)
4159 (power z2 (add a (mul -1 index) -1)))
4160 (mul (power (sub 1 z1) b)
4161 (power z1 (add a (mul -1 index) -1)))))
4162 index 0 (sub n 1)))))))
4165 (eqtest (list '(%beta_incomplete_generalized) a b z1 z2) expr)))))
4167 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
4169 ;;; Implementation of the Regularized Incomplete Beta function
4171 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
4173 (defmfun $beta_incomplete_regularized (a b z)
4174 (simplify (list '(%beta_incomplete_regularized) a b z)))
4176 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
4178 (defprop $beta_incomplete_regularized %beta_incomplete_regularized alias)
4179 (defprop $beta_incomplete_regularized %beta_incomplete_regularized verb)
4181 (defprop %beta_incomplete_regularized $beta_incomplete_regularized reversealias)
4182 (defprop %beta_incomplete_regularized $beta_incomplete_regularized noun)
4184 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
4186 (defprop %beta_incomplete_regularized
4187 simp-beta-incomplete-regularized operators)
4189 ;;; beta_incomplete_regularized distributes over bags
4191 (defprop %beta_incomplete_regularized (mlist $matrix mequal) distribute_over)
4193 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
4195 (defprop %beta_incomplete_regularized
4196 ((a b z)
4197 ;; Derivative wrt a
4198 ((mplus)
4199 ((mtimes) -1
4200 ((%gamma) a)
4201 (($hypergeometric_regularized)
4202 ((mlist) a a ((mplus) 1 ((mtimes) -1 b)))
4203 ((mlist) ((mplus) 1 a) ((mplus) 2 a)) z)
4204 ((mexpt) ((%gamma) b) -1)
4205 ((%gamma) ((mplus) a b))
4206 ((mexpt) z a))
4207 ((mtimes)
4208 ((%beta_incomplete_regularized) a b z)
4209 ((mplus)
4210 ((mtimes) -1 ((mqapply) (($psi array) 0) a))
4211 ((mqapply) (($psi array) 0) ((mplus) a b))
4212 ((%log) z))))
4213 ;; Derivative wrt b
4214 ((mplus)
4215 ((mtimes)
4216 ((%beta_incomplete_regularized) b a ((mplus) 1 ((mtimes) -1 z)))
4217 ((mplus)
4218 ((mqapply) (($psi array) 0) b)
4219 ((mtimes) -1 ((mqapply) (($psi array) 0) ((mplus) a b)))
4220 ((mtimes) -1 ((%log) ((mplus) 1 ((mtimes) -1 z))))))
4221 ((mtimes)
4222 ((mexpt) ((%gamma) a) -1)
4223 ((%gamma) b)
4224 ((%gamma) ((mplus) a b))
4225 (($hypergeometric_regularized)
4226 ((mlist) b b ((mplus) 1 ((mtimes) -1 a)))
4227 ((mlist) ((mplus) 1 b) ((mplus) 1 b))
4228 ((mplus) 1 ((mtimes) -1 z)))
4229 ((mexpt) ((mplus) 1 ((mtimes) -1 z)) b)))
4230 ;; The derivative wrt z
4231 ((mtimes)
4232 ((mexpt) (($beta) a b) -1)
4233 ((mexpt) ((mplus) 1 ((mtimes) -1 z)) ((mplus) -1 b))
4234 ((mexpt) z ((mplus) -1 a))))
4235 grad)
4237 ;;; Integral of the Generalized Incomplete Beta function
4239 (defprop %beta_incomplete_regularized
4240 ((a b z)
4241 nil
4243 ;; Integral wrt z
4244 ((mplus)
4245 ((mtimes) -1 a
4246 ((%beta_incomplete_regularized) ((mplus) 1 a) b z)
4247 ((mexpt) ((mplus) a b) -1))
4248 ((mtimes) ((%beta_incomplete_regularized) a b z) z)))
4249 integral)
4251 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
4253 (defun simp-beta-incomplete-regularized (expr ignored simpflag)
4254 (declare (ignore ignored))
4255 (arg-count-check 3 expr)
4256 (let ((a (simpcheck (second expr) simpflag))
4257 (b (simpcheck (third expr) simpflag))
4258 (z (simpcheck (fourth expr) simpflag))
4259 ($simpsum t))
4260 (cond
4262 ;; Check for specific values
4264 ((zerop1 z)
4265 (let ((sgn ($sign ($realpart a))))
4266 (cond ((eq sgn '$neg)
4267 (simp-domain-error
4268 (intl:gettext
4269 "beta_incomplete_regularized: beta_incomplete_regularized(~:M,~:M,~:M) is undefined.")
4270 a b z))
4271 ((member sgn '($pos $pz))
4274 (eqtest
4275 (list '(%beta_incomplete_regularized) a b z) expr)))))
4277 ((and (onep1 z)
4278 (or (not (mnump a))
4279 (not (mnump b))
4280 (not (mnump z))))
4281 (let ((sgn ($sign ($realpart b))))
4282 (cond ((member sgn '($pos $pz))
4285 (eqtest
4286 (list '(%beta_incomplete_regularized) a b z) expr)))))
4288 ((and (integer-representation-p b)
4289 (if ($bfloatp b) (minusp (cadr b)) (minusp b)) )
4290 ;; Problem: for b a negative integer the Regularized Incomplete
4291 ;; Beta function is defined to be zero. BUT: When we calculate
4292 ;; e.g. beta_incomplete(1.0,-2.0,1/2)/beta(1.0,-2.0) we get the
4293 ;; result -3.0, because beta_incomplete and beta are defined for
4294 ;; for this case. How do we get a consistent behaviour?
4297 ((and (integer-representation-p a)
4298 (if ($bfloatp a) (minusp (cadr a)) (minusp a)) )
4299 (cond
4300 ;; TODO: The following line does not work for bigfloats.
4301 ((and (integer-representation-p b) (<= b (- a)))
4302 ;; Does $beta_incomplete or simpbeta underflow in this case?
4303 (div ($beta_incomplete a b z)
4304 (simplify (list '($beta) a b))))
4306 1)))
4308 ;; Check for numerical evaluation in Float or Bigfloat precision
4310 ((complex-float-numerical-eval-p a b z)
4311 (let ((*beta-incomplete-eps* (bigfloat:epsilon ($float 1.0)))
4312 beta ibeta )
4313 (setq a ($float a) b ($float b))
4314 (if (or (< ($abs (setq beta (simplify (list '($beta) a b)))) 1e-307)
4315 (< ($abs (setq ibeta (beta-incomplete a b ($float z)))) 1e-307) )
4316 ;; In case of underflow (see bug #2999) or precision loss use bigfloats
4317 ;; and emporarily give some extra precision but avoid fpprec dependency.
4318 ;; Is this workaround correct for complex values?
4319 (let ((fpprec 70))
4320 ($float ($beta_incomplete_regularized ($bfloat a) ($bfloat b) z)) )
4321 ($rectform (div ibeta beta)) )))
4323 ((complex-bigfloat-numerical-eval-p a b z)
4324 (let ((*beta-incomplete-eps*
4325 (bigfloat:epsilon (bigfloat:bigfloat 1.0))))
4326 (setq a ($bfloat a) b ($bfloat b))
4327 ($rectform
4328 (div (beta-incomplete a b ($bfloat z))
4329 (simplify (list '($beta) a b))))))
4331 ;; Check for argument simplifications and transformations
4333 ((and (integerp b) (plusp b))
4334 (div ($beta_incomplete a b z)
4335 (simplify (list '($beta) a b))))
4337 ((and (integerp a) (plusp a))
4338 (div ($beta_incomplete a b z)
4339 (simplify (list '($beta) a b))))
4341 ((and $beta_expand (mplusp a) (integerp (cadr a)) (plusp (cadr a)))
4342 (let ((n (cadr a))
4343 (a (simplify (cons '(mplus) (cddr a)))))
4344 (sub
4345 ($beta_incomplete_regularized a b z)
4346 (mul
4347 (power (add a b n -1) -1)
4348 (power (simplify (list '($beta) (add a n) b)) -1)
4349 (let ((index (gensumindex)))
4350 (simpsum1
4351 (mul
4352 (div
4353 (simplify (list '($pochhammer)
4354 (add 1 (mul -1 a) (mul -1 n))
4355 index))
4356 (simplify (list '($pochhammer)
4357 (add 2 (mul -1 a) (mul -1 b) (mul -1 n))
4358 index)))
4359 (power (sub 1 z) b)
4360 (power z (add a n (mul -1 index) -1)))
4361 index 0 (sub n 1)))))))
4363 ((and $beta_expand (mplusp a) (integerp (cadr a)) (minusp (cadr a)))
4364 (let ((n (- (cadr a)))
4365 (a (simplify (cons '(mplus) (cddr a)))))
4366 (sub
4367 ($beta_incomplete_regularized a b z)
4368 (mul
4369 (power (add a b -1) -1)
4370 (power (simplify (list '($beta) a b)) -1)
4371 (let ((index (gensumindex)))
4372 (simpsum1
4373 (mul
4374 (div
4375 (simplify (list '($pochhammer) (sub 1 a) index))
4376 (simplify (list '($pochhammer)
4377 (add 2 (mul -1 a) (mul -1 b))
4378 index)))
4379 (power (sub 1 z) b)
4380 (power z (add a (mul -1 index) -1)))
4381 index 0 (sub n 1)))))))
4384 (eqtest (list '(%beta_incomplete_regularized) a b z) expr)))))
4386 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;