Fix description of 'C-z' in User manual
[emacs.git] / src / floatfns.c
blobf514fcbea8cb4d5779d357fc96bc3a6d33c2e814
1 /* Primitive operations on floating point for GNU Emacs Lisp interpreter.
3 Copyright (C) 1988, 1993-1994, 1999, 2001-2016 Free Software Foundation,
4 Inc.
6 Author: Wolfgang Rupprecht (according to ack.texi)
8 This file is part of GNU Emacs.
10 GNU Emacs is free software: you can redistribute it and/or modify
11 it under the terms of the GNU General Public License as published by
12 the Free Software Foundation, either version 3 of the License, or (at
13 your option) any later version.
15 GNU Emacs is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 GNU General Public License for more details.
20 You should have received a copy of the GNU General Public License
21 along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */
24 /* C89 requires only the following math.h functions, and Emacs omits
25 the starred functions since we haven't found a use for them:
26 acos, asin, atan, atan2, ceil, cos, *cosh, exp, fabs, floor, fmod,
27 frexp, ldexp, log, log10 [via (log X 10)], *modf, pow, sin, *sinh,
28 sqrt, tan, *tanh.
30 C99 and C11 require the following math.h functions in addition to
31 the C89 functions. Of these, Emacs currently exports only the
32 starred ones to Lisp, since we haven't found a use for the others:
33 acosh, atanh, cbrt, *copysign, erf, erfc, exp2, expm1, fdim, fma,
34 fmax, fmin, fpclassify, hypot, ilogb, isfinite, isgreater,
35 isgreaterequal, isinf, isless, islessequal, islessgreater, *isnan,
36 isnormal, isunordered, lgamma, log1p, *log2 [via (log X 2)], *logb
37 (approximately), lrint/llrint, lround/llround, nan, nearbyint,
38 nextafter, nexttoward, remainder, remquo, *rint, round, scalbln,
39 scalbn, signbit, tgamma, trunc.
42 #include <config.h>
44 #include "lisp.h"
46 #include <math.h>
48 /* 'isfinite' and 'isnan' cause build failures on Solaris 10 with the
49 bundled GCC in c99 mode. Work around the bugs with simple
50 implementations that are good enough. */
51 #undef isfinite
52 #define isfinite(x) ((x) - (x) == 0)
53 #undef isnan
54 #define isnan(x) ((x) != (x))
56 /* Check that X is a floating point number. */
58 static void
59 CHECK_FLOAT (Lisp_Object x)
61 CHECK_TYPE (FLOATP (x), Qfloatp, x);
64 /* Extract a Lisp number as a `double', or signal an error. */
66 double
67 extract_float (Lisp_Object num)
69 CHECK_NUMBER_OR_FLOAT (num);
71 if (FLOATP (num))
72 return XFLOAT_DATA (num);
73 return (double) XINT (num);
76 /* Trig functions. */
78 DEFUN ("acos", Facos, Sacos, 1, 1, 0,
79 doc: /* Return the inverse cosine of ARG. */)
80 (Lisp_Object arg)
82 double d = extract_float (arg);
83 d = acos (d);
84 return make_float (d);
87 DEFUN ("asin", Fasin, Sasin, 1, 1, 0,
88 doc: /* Return the inverse sine of ARG. */)
89 (Lisp_Object arg)
91 double d = extract_float (arg);
92 d = asin (d);
93 return make_float (d);
96 DEFUN ("atan", Fatan, Satan, 1, 2, 0,
97 doc: /* Return the inverse tangent of the arguments.
98 If only one argument Y is given, return the inverse tangent of Y.
99 If two arguments Y and X are given, return the inverse tangent of Y
100 divided by X, i.e. the angle in radians between the vector (X, Y)
101 and the x-axis. */)
102 (Lisp_Object y, Lisp_Object x)
104 double d = extract_float (y);
106 if (NILP (x))
107 d = atan (d);
108 else
110 double d2 = extract_float (x);
111 d = atan2 (d, d2);
113 return make_float (d);
116 DEFUN ("cos", Fcos, Scos, 1, 1, 0,
117 doc: /* Return the cosine of ARG. */)
118 (Lisp_Object arg)
120 double d = extract_float (arg);
121 d = cos (d);
122 return make_float (d);
125 DEFUN ("sin", Fsin, Ssin, 1, 1, 0,
126 doc: /* Return the sine of ARG. */)
127 (Lisp_Object arg)
129 double d = extract_float (arg);
130 d = sin (d);
131 return make_float (d);
134 DEFUN ("tan", Ftan, Stan, 1, 1, 0,
135 doc: /* Return the tangent of ARG. */)
136 (Lisp_Object arg)
138 double d = extract_float (arg);
139 d = tan (d);
140 return make_float (d);
143 DEFUN ("isnan", Fisnan, Sisnan, 1, 1, 0,
144 doc: /* Return non nil if argument X is a NaN. */)
145 (Lisp_Object x)
147 CHECK_FLOAT (x);
148 return isnan (XFLOAT_DATA (x)) ? Qt : Qnil;
151 #ifdef HAVE_COPYSIGN
152 DEFUN ("copysign", Fcopysign, Scopysign, 2, 2, 0,
153 doc: /* Copy sign of X2 to value of X1, and return the result.
154 Cause an error if X1 or X2 is not a float. */)
155 (Lisp_Object x1, Lisp_Object x2)
157 double f1, f2;
159 CHECK_FLOAT (x1);
160 CHECK_FLOAT (x2);
162 f1 = XFLOAT_DATA (x1);
163 f2 = XFLOAT_DATA (x2);
165 return make_float (copysign (f1, f2));
167 #endif
169 DEFUN ("frexp", Ffrexp, Sfrexp, 1, 1, 0,
170 doc: /* Get significand and exponent of a floating point number.
171 Breaks the floating point number X into its binary significand SGNFCAND
172 \(a floating point value between 0.5 (included) and 1.0 (excluded))
173 and an integral exponent EXP for 2, such that:
175 X = SGNFCAND * 2^EXP
177 The function returns the cons cell (SGNFCAND . EXP).
178 If X is zero, both parts (SGNFCAND and EXP) are zero. */)
179 (Lisp_Object x)
181 double f = XFLOATINT (x);
182 int exponent;
183 double sgnfcand = frexp (f, &exponent);
184 return Fcons (make_float (sgnfcand), make_number (exponent));
187 DEFUN ("ldexp", Fldexp, Sldexp, 2, 2, 0,
188 doc: /* Return SGNFCAND * 2**EXPONENT, as a floating point number.
189 EXPONENT must be an integer. */)
190 (Lisp_Object sgnfcand, Lisp_Object exponent)
192 CHECK_NUMBER (exponent);
193 int e = min (max (INT_MIN, XINT (exponent)), INT_MAX);
194 return make_float (ldexp (XFLOATINT (sgnfcand), e));
197 DEFUN ("exp", Fexp, Sexp, 1, 1, 0,
198 doc: /* Return the exponential base e of ARG. */)
199 (Lisp_Object arg)
201 double d = extract_float (arg);
202 d = exp (d);
203 return make_float (d);
206 DEFUN ("expt", Fexpt, Sexpt, 2, 2, 0,
207 doc: /* Return the exponential ARG1 ** ARG2. */)
208 (Lisp_Object arg1, Lisp_Object arg2)
210 double f1, f2, f3;
212 CHECK_NUMBER_OR_FLOAT (arg1);
213 CHECK_NUMBER_OR_FLOAT (arg2);
214 if (INTEGERP (arg1) /* common lisp spec */
215 && INTEGERP (arg2) /* don't promote, if both are ints, and */
216 && XINT (arg2) >= 0) /* we are sure the result is not fractional */
217 { /* this can be improved by pre-calculating */
218 EMACS_INT y; /* some binary powers of x then accumulating */
219 EMACS_UINT acc, x; /* Unsigned so that overflow is well defined. */
220 Lisp_Object val;
222 x = XINT (arg1);
223 y = XINT (arg2);
224 acc = (y & 1 ? x : 1);
226 while ((y >>= 1) != 0)
228 x *= x;
229 if (y & 1)
230 acc *= x;
232 XSETINT (val, acc);
233 return val;
235 f1 = FLOATP (arg1) ? XFLOAT_DATA (arg1) : XINT (arg1);
236 f2 = FLOATP (arg2) ? XFLOAT_DATA (arg2) : XINT (arg2);
237 f3 = pow (f1, f2);
238 return make_float (f3);
241 DEFUN ("log", Flog, Slog, 1, 2, 0,
242 doc: /* Return the natural logarithm of ARG.
243 If the optional argument BASE is given, return log ARG using that base. */)
244 (Lisp_Object arg, Lisp_Object base)
246 double d = extract_float (arg);
248 if (NILP (base))
249 d = log (d);
250 else
252 double b = extract_float (base);
254 if (b == 10.0)
255 d = log10 (d);
256 #if HAVE_LOG2
257 else if (b == 2.0)
258 d = log2 (d);
259 #endif
260 else
261 d = log (d) / log (b);
263 return make_float (d);
266 DEFUN ("sqrt", Fsqrt, Ssqrt, 1, 1, 0,
267 doc: /* Return the square root of ARG. */)
268 (Lisp_Object arg)
270 double d = extract_float (arg);
271 d = sqrt (d);
272 return make_float (d);
275 DEFUN ("abs", Fabs, Sabs, 1, 1, 0,
276 doc: /* Return the absolute value of ARG. */)
277 (register Lisp_Object arg)
279 CHECK_NUMBER_OR_FLOAT (arg);
281 if (FLOATP (arg))
282 arg = make_float (fabs (XFLOAT_DATA (arg)));
283 else if (XINT (arg) < 0)
284 XSETINT (arg, - XINT (arg));
286 return arg;
289 DEFUN ("float", Ffloat, Sfloat, 1, 1, 0,
290 doc: /* Return the floating point number equal to ARG. */)
291 (register Lisp_Object arg)
293 CHECK_NUMBER_OR_FLOAT (arg);
295 if (INTEGERP (arg))
296 return make_float ((double) XINT (arg));
297 else /* give 'em the same float back */
298 return arg;
301 DEFUN ("logb", Flogb, Slogb, 1, 1, 0,
302 doc: /* Returns largest integer <= the base 2 log of the magnitude of ARG.
303 This is the same as the exponent of a float. */)
304 (Lisp_Object arg)
306 Lisp_Object val;
307 EMACS_INT value;
308 double f = extract_float (arg);
310 if (f == 0.0)
311 value = MOST_NEGATIVE_FIXNUM;
312 else if (isfinite (f))
314 int ivalue;
315 frexp (f, &ivalue);
316 value = ivalue - 1;
318 else
319 value = MOST_POSITIVE_FIXNUM;
321 XSETINT (val, value);
322 return val;
326 /* the rounding functions */
328 static Lisp_Object
329 rounding_driver (Lisp_Object arg, Lisp_Object divisor,
330 double (*double_round) (double),
331 EMACS_INT (*int_round2) (EMACS_INT, EMACS_INT),
332 const char *name)
334 CHECK_NUMBER_OR_FLOAT (arg);
336 if (! NILP (divisor))
338 EMACS_INT i1, i2;
340 CHECK_NUMBER_OR_FLOAT (divisor);
342 if (FLOATP (arg) || FLOATP (divisor))
344 double f1, f2;
346 f1 = FLOATP (arg) ? XFLOAT_DATA (arg) : XINT (arg);
347 f2 = (FLOATP (divisor) ? XFLOAT_DATA (divisor) : XINT (divisor));
348 if (! IEEE_FLOATING_POINT && f2 == 0)
349 xsignal0 (Qarith_error);
351 f1 = (*double_round) (f1 / f2);
352 if (FIXNUM_OVERFLOW_P (f1))
353 xsignal3 (Qrange_error, build_string (name), arg, divisor);
354 arg = make_number (f1);
355 return arg;
358 i1 = XINT (arg);
359 i2 = XINT (divisor);
361 if (i2 == 0)
362 xsignal0 (Qarith_error);
364 XSETINT (arg, (*int_round2) (i1, i2));
365 return arg;
368 if (FLOATP (arg))
370 double d = (*double_round) (XFLOAT_DATA (arg));
371 if (FIXNUM_OVERFLOW_P (d))
372 xsignal2 (Qrange_error, build_string (name), arg);
373 arg = make_number (d);
376 return arg;
379 static EMACS_INT
380 ceiling2 (EMACS_INT i1, EMACS_INT i2)
382 return i1 / i2 + ((i1 % i2 != 0) & ((i1 < 0) == (i2 < 0)));
385 static EMACS_INT
386 floor2 (EMACS_INT i1, EMACS_INT i2)
388 return i1 / i2 - ((i1 % i2 != 0) & ((i1 < 0) != (i2 < 0)));
391 static EMACS_INT
392 truncate2 (EMACS_INT i1, EMACS_INT i2)
394 return i1 / i2;
397 static EMACS_INT
398 round2 (EMACS_INT i1, EMACS_INT i2)
400 /* The C language's division operator gives us one remainder R, but
401 we want the remainder R1 on the other side of 0 if R1 is closer
402 to 0 than R is; because we want to round to even, we also want R1
403 if R and R1 are the same distance from 0 and if C's quotient is
404 odd. */
405 EMACS_INT q = i1 / i2;
406 EMACS_INT r = i1 % i2;
407 EMACS_INT abs_r = eabs (r);
408 EMACS_INT abs_r1 = eabs (i2) - abs_r;
409 return q + (abs_r + (q & 1) <= abs_r1 ? 0 : (i2 ^ r) < 0 ? -1 : 1);
412 /* The code uses emacs_rint, so that it works to undefine HAVE_RINT
413 if `rint' exists but does not work right. */
414 #ifdef HAVE_RINT
415 #define emacs_rint rint
416 #else
417 static double
418 emacs_rint (double d)
420 double d1 = d + 0.5;
421 double r = floor (d1);
422 return r - (r == d1 && fmod (r, 2) != 0);
424 #endif
426 static double
427 double_identity (double d)
429 return d;
432 DEFUN ("ceiling", Fceiling, Sceiling, 1, 2, 0,
433 doc: /* Return the smallest integer no less than ARG.
434 This rounds the value towards +inf.
435 With optional DIVISOR, return the smallest integer no less than ARG/DIVISOR. */)
436 (Lisp_Object arg, Lisp_Object divisor)
438 return rounding_driver (arg, divisor, ceil, ceiling2, "ceiling");
441 DEFUN ("floor", Ffloor, Sfloor, 1, 2, 0,
442 doc: /* Return the largest integer no greater than ARG.
443 This rounds the value towards -inf.
444 With optional DIVISOR, return the largest integer no greater than ARG/DIVISOR. */)
445 (Lisp_Object arg, Lisp_Object divisor)
447 return rounding_driver (arg, divisor, floor, floor2, "floor");
450 DEFUN ("round", Fround, Sround, 1, 2, 0,
451 doc: /* Return the nearest integer to ARG.
452 With optional DIVISOR, return the nearest integer to ARG/DIVISOR.
454 Rounding a value equidistant between two integers may choose the
455 integer closer to zero, or it may prefer an even integer, depending on
456 your machine. For example, (round 2.5) can return 3 on some
457 systems, but 2 on others. */)
458 (Lisp_Object arg, Lisp_Object divisor)
460 return rounding_driver (arg, divisor, emacs_rint, round2, "round");
463 DEFUN ("truncate", Ftruncate, Struncate, 1, 2, 0,
464 doc: /* Truncate a floating point number to an int.
465 Rounds ARG toward zero.
466 With optional DIVISOR, truncate ARG/DIVISOR. */)
467 (Lisp_Object arg, Lisp_Object divisor)
469 return rounding_driver (arg, divisor, double_identity, truncate2,
470 "truncate");
474 Lisp_Object
475 fmod_float (Lisp_Object x, Lisp_Object y)
477 double f1, f2;
479 f1 = FLOATP (x) ? XFLOAT_DATA (x) : XINT (x);
480 f2 = FLOATP (y) ? XFLOAT_DATA (y) : XINT (y);
482 f1 = fmod (f1, f2);
484 /* If the "remainder" comes out with the wrong sign, fix it. */
485 if (f2 < 0 ? f1 > 0 : f1 < 0)
486 f1 += f2;
488 return make_float (f1);
491 DEFUN ("fceiling", Ffceiling, Sfceiling, 1, 1, 0,
492 doc: /* Return the smallest integer no less than ARG, as a float.
493 \(Round toward +inf.) */)
494 (Lisp_Object arg)
496 double d = extract_float (arg);
497 d = ceil (d);
498 return make_float (d);
501 DEFUN ("ffloor", Fffloor, Sffloor, 1, 1, 0,
502 doc: /* Return the largest integer no greater than ARG, as a float.
503 \(Round towards -inf.) */)
504 (Lisp_Object arg)
506 double d = extract_float (arg);
507 d = floor (d);
508 return make_float (d);
511 DEFUN ("fround", Ffround, Sfround, 1, 1, 0,
512 doc: /* Return the nearest integer to ARG, as a float. */)
513 (Lisp_Object arg)
515 double d = extract_float (arg);
516 d = emacs_rint (d);
517 return make_float (d);
520 DEFUN ("ftruncate", Fftruncate, Sftruncate, 1, 1, 0,
521 doc: /* Truncate a floating point number to an integral float value.
522 Rounds the value toward zero. */)
523 (Lisp_Object arg)
525 double d = extract_float (arg);
526 if (d >= 0.0)
527 d = floor (d);
528 else
529 d = ceil (d);
530 return make_float (d);
533 void
534 syms_of_floatfns (void)
536 defsubr (&Sacos);
537 defsubr (&Sasin);
538 defsubr (&Satan);
539 defsubr (&Scos);
540 defsubr (&Ssin);
541 defsubr (&Stan);
542 defsubr (&Sisnan);
543 #ifdef HAVE_COPYSIGN
544 defsubr (&Scopysign);
545 #endif
546 defsubr (&Sfrexp);
547 defsubr (&Sldexp);
548 defsubr (&Sfceiling);
549 defsubr (&Sffloor);
550 defsubr (&Sfround);
551 defsubr (&Sftruncate);
552 defsubr (&Sexp);
553 defsubr (&Sexpt);
554 defsubr (&Slog);
555 defsubr (&Ssqrt);
557 defsubr (&Sabs);
558 defsubr (&Sfloat);
559 defsubr (&Slogb);
560 defsubr (&Sceiling);
561 defsubr (&Sfloor);
562 defsubr (&Sround);
563 defsubr (&Struncate);