8758c02ab8
New module for gamma-zeta functions, with new and improved functions and algorithms:
Fredrik Johansson
2008-09-05 12:56:43 +00:00
3916b248e2
faster computation of real erf, an accurate implementation of erfc, and raw normal probability functions npdf, ncdf
Fredrik Johansson
2008-08-30 15:56:30 +00:00
ff0f93f6cb
* compute ellipk via agm, which is way faster and works in the entire complex plane * allow single-argument form agm(x) = agm(1,x) * implemented the isinf utility function
Fredrik Johansson
2008-08-30 09:51:04 +00:00
fd4bb9f4e2
plotting implemented (yay!)
Fredrik Johansson
2008-08-30 00:58:35 +00:00
e758135397
fix nthroot for n = 1, 0, -1
Fredrik Johansson
2008-08-29 21:52:06 +00:00
8d57f05a26
fix pickle in python mode
Fredrik Johansson
2008-08-29 18:01:27 +00:00
6adb621ffc
Major overhaul of numerical integration:
Fredrik Johansson
2008-08-29 17:58:06 +00:00
8b60d994ca
use hex() instead of long() when pickling, because pickle insists on using quadratic str() for representing longs
Fredrik Johansson
2008-08-29 17:15:39 +00:00
491132b5fe
use a much better iteration for Euler's constant
Fredrik Johansson
2008-08-26 14:37:49 +00:00
2aa95d0861
use fixed-point arithmetic in pslq, giving a ~7x speedup
Fredrik Johansson
2008-08-26 13:55:09 +00:00
7ec289bc6d
fix inf/2**n and nan/2**n returning zero
Fredrik Johansson
2008-08-23 21:53:18 +00:00
263724643a
fix docs to say version 0.9 instead of 0.8
Fredrik Johansson
2008-08-23 20:42:05 +00:00
b409862597
docs updated for 0.9
Fredrik Johansson
2008-08-23 14:37:56 +00:00
b84ab792b7
use getrandbits instead of randrange in frand, which is much more efficient
Fredrik Johansson
2008-08-23 14:16:16 +00:00
5be8d6138f
update CHANGES and README
Fredrik Johansson
2008-08-23 13:09:16 +00:00
d667941a99
change polyval and related functions to use the same coefficient order as scipy. small fixes and some more tests added
Fredrik Johansson
2008-08-23 13:03:15 +00:00
4ddd3c9c27
fix various limit points of exponential integrals
Fredrik Johansson
2008-08-23 09:55:56 +00:00
4c9feca7aa
return NotImplemented in more __op__ methods, fixing issue 40. also support a special method _mpmath_ for conversion. the code is quite messy and might have gotten slightly less efficient / there could be some corner case that doesn't work
Fredrik Johansson
2008-08-22 23:00:59 +00:00
b07195b99c
implement sign(); fix and write tests for arg()
Fredrik Johansson
2008-08-22 20:35:10 +00:00
5b13218ee1
added several more special hypergeometric functions
Fredrik Johansson
2008-08-22 19:46:20 +00:00
8dc1bb1b93
take care to use MP_BASE for round-to-even bitmasks. this avoids massive random slowdown at huge precision (e.g. fmul at 1M bits would sometimes take 20x longer than it should)
Fredrik Johansson
2008-08-22 11:03:35 +00:00
8a1ae9c820
workaround gmpy bug causing crash in mpc_nthroot
Fredrik Johansson
2008-07-29 18:33:34 +00:00
dd9451ac39
Fixed bug in fsqrt for down rounding (see issue 14); added nthroot; added cmp_ae in test_mpmath.py to test separately the real and imaginary part.
Mario Pernici
2008-07-29 08:38:32 +00:00
dc784518ce
Faster mpc_cbrt using Newton method.
Mario Pernici
2008-07-14 17:51:15 +00:00
9d76668e6a
Added cbrt.
Mario Pernici
2008-07-07 16:09:32 +00:00
0ca738e61c
Fixed gmpy support to work with Python 2.4. Python 2.4 does not use the __index__ method of an mpz when indexing a list so added int() around mpz when indexing a list.
Case Van Horsen
2008-07-06 03:35:09 +00:00
a07cf48e8f
Whitespace fixed.
Ondrej Certik
2008-07-05 20:26:38 +00:00
f73e77e391
updated searchtools.js
Fredrik Johansson
2008-07-05 14:49:57 +00:00
37f30a23f3
rebuilt searchindex.json
Fredrik Johansson
2008-07-05 14:17:25 +00:00
3c3282e66c
add option in to_str not to strip trailing zeros
Fredrik Johansson
2008-07-05 13:38:24 +00:00
98383b5663
made aliases log10, degrees, radians, frexp, modf, ln available
Fredrik Johansson
2008-07-05 13:31:20 +00:00
04466ffe45
possible fix for documentation search in IE
Fredrik Johansson
2008-07-05 11:00:37 +00:00
2e5677e488
Added additional sanity checks to -strict mode. Fixed a few comments.
Case Van Horsen
2008-07-03 15:31:45 +00:00
474607a84b
update docs
Fredrik Johansson
2008-07-03 11:40:39 +00:00
1864d1509c
New numeral_gmpy that can handle extremely large values.
Case Van Horsen
2008-07-03 06:44:41 +00:00
1baed73c9e
Update documentation for -nogmpy and -strict options to runtests.py.
Case Van Horsen
2008-07-03 06:22:29 +00:00
f33566fc0f
make to_str return something resembling a scaled zero at prec = 0
Fredrik Johansson
2008-07-02 15:06:07 +00:00
4cc40bdf58
use chudnovsky's algorithm to compute pi
Fredrik Johansson
2008-07-02 10:45:36 +00:00
7118f6a130
Added support for -nogmpy and -strict to runtests.py.
Case Van Horsen
2008-07-02 06:00:25 +00:00
adc3dc5f7b
Minor cleanup of gmpy support.
Case Van Horsen
2008-07-01 06:25:43 +00:00
ca824e3b6d
fix atan(x) for ginormous x
Fredrik Johansson
2008-06-30 11:23:41 +00:00
bfd55bb6f3
use binary splitting to compute exp(1). this is faster at high precision in both python mode and gmpy mode
Fredrik Johansson
2008-06-28 22:42:02 +00:00
1ba58c0ff4
fix for issue 45, patch contributed by ntelang
Fredrik Johansson
2008-06-28 21:43:13 +00:00
acdca16420
use gmpy in a few more places in specfun.py
Fredrik Johansson
2008-06-26 16:33:58 +00:00
3803309ec8
fixes to string conversion to take advantage of gmpy / fix pidigits.py
Fredrik Johansson
2008-06-26 15:52:37 +00:00
32b597dc00
apply casevh's patch for gmpy support
Fredrik Johansson
2008-06-26 15:15:43 +00:00
e171fb0a6d
slightly faster exp, cos, sin at high precision
Mario Pernici
2008-06-25 09:07:19 +00:00
0cfa540935
fix triple quote in docstring
Fredrik Johansson
2008-06-24 03:39:07 +00:00
cbd49c5676
commit Mike Taschuk's code for elliptic functions, with only slight modifications for now
Fredrik Johansson
2008-06-21 19:10:14 +00:00
a88e8ae17f
Slightly faster log for dps > 30 using Halley method. Added test for bug in previous implementation of log.
Mario Pernici
2008-06-17 16:17:14 +00:00
5c1a334895
faster inverse trigonometric functions at high precision using Newton method
Mario Pernici
2008-06-04 17:26:32 +00:00
c9bd346795
faster trigonometric functions at high precision using Brent's trick
Mario Pernici
2008-05-26 15:03:25 +00:00
3c51d33c57
Fixed bug in TS_estimate_error.
Mario Pernici
2008-05-23 09:46:17 +00:00
7bc5be1c0b
small speed-up for ODE
Mario Pernici
2008-05-14 16:21:52 +00:00
74b5869c5f
fix issue 38: remove obsolete imports
Fredrik Johansson
2008-05-08 15:44:33 +00:00
a28b976f00
fix shebangs
Fredrik Johansson
2008-04-25 14:20:30 +00:00
a737822ad3
improvements to arange and tests for it, contributed by Vinzent Steinberg
Fredrik Johansson
2008-04-23 19:57:19 +00:00
051dbabccd
added shebangs to script files
Fredrik Johansson
2008-04-23 19:52:19 +00:00
f1b172f6fc
add version information to __init__.py. fixing the imports in specfun.py
Fredrik Johansson
2008-04-21 17:27:12 +00:00
7e00c0444e
documentation tweaks
Fredrik Johansson
2008-04-20 15:19:21 +00:00
03d14d0c04
some more documentation updates
Fredrik Johansson
2008-04-20 14:43:28 +00:00
be5d7d7c54
attempt to fix issue 22
Fredrik Johansson
2008-04-20 14:21:52 +00:00
44fc62b834
some additional documentation and cleanup. changing quadosc to use alt=1 by default
Fredrik Johansson
2008-04-20 12:33:55 +00:00
ead1fb2373
more work on docs. renamed ODE_integrate to odeint and made its interface more similar to scipy's odeint. made arange more accurate
Fredrik Johansson
2008-04-19 18:08:56 +00:00
9c6a8ac2fe
updates to documentation and setup files, in anticipation of 0.8 release
Fredrik Johansson
2008-04-19 15:26:53 +00:00
2c6e2e696f
possible fix for issue 26
Fredrik Johansson
2008-04-19 11:10:38 +00:00
9e3ea181ad
remove specfun subdirectory. adding a possible fix for issue 22
Fredrik Johansson
2008-04-19 11:03:43 +00:00
21834f715d
reciprocal elementary functions implemented. permit integration endpoints in quadts to be complex.
Fredrik Johansson
2008-04-18 18:06:41 +00:00
5c34141f39
modified interface to some functions not to return an error estimate by default. improved polyroots by making it clean up pure real roots and sorting the list more cleverly. documentation updates.
Fredrik Johansson
2008-04-17 18:44:31 +00:00
75f2b26cd2
fix bug in mpf raised to a negative half-integer power
Fredrik Johansson
2008-04-16 10:16:29 +00:00
66d2982086
text/html type
Fredrik Johansson
2008-04-14 21:38:23 +00:00
c5da31214c
updated documentation (mainly to describe oscillatory integration)
Fredrik Johansson
2008-04-14 21:34:48 +00:00
243756a4fc
implemented oscillatory quadrature
Fredrik Johansson
2008-04-14 20:01:16 +00:00
6438d89936
more unified implementation of some of the special functions, with bessel and incomplete gamma functions computed from hypergeometric series in the general case (jv now supports complex arguments). fixed mpf comparison methods to work with float(nan)
Fredrik Johansson
2008-04-13 18:32:40 +00:00
6e530f2d2f
try changing css file mime types to text/css
Fredrik Johansson
2008-04-12 22:33:52 +00:00
b1de6daae8
second try to get css to work
Fredrik Johansson
2008-04-12 22:23:50 +00:00
d0bfd24ddf
fix css loading (hopefully)
Fredrik Johansson
2008-04-12 22:08:37 +00:00
053e1c7b7b
uploading in-progress sphinx-based documentation for 0.8. slight weaks to identify.
Fredrik Johansson
2008-04-12 21:36:05 +00:00
8777002268
improvements to identify. a few other small bugfixes
Fredrik Johansson
2008-04-12 17:44:49 +00:00
97e9444257
implemented PSLQ + additional heuristics for constant recognition
Fredrik Johansson
2008-04-10 18:25:15 +00:00
f7c6771de1
Fixed bug in fexp; added test for it.
Mario Pernici
2008-04-09 11:03:48 +00:00
48a12a39f0
adding missing tests for constants. slightly increasing working precision for khinchin and glaisher to make tests pass
Fredrik Johansson
2008-04-08 13:58:29 +00:00
adc5dbc124
implemented the Lambert W function
Fredrik Johansson
2008-04-08 13:01:09 +00:00
5b09ae21ac
complex acosh and asinh more precise
Mario Pernici
2008-04-03 15:43:34 +00:00
f85bbc94de
fixed bug in test committed in previous version
Mario Pernici
2008-04-02 04:11:10 +00:00
1c08787474
more precise and faster complex acos and asin; added tests for them
Mario Pernici
2008-04-01 21:20:20 +00:00
0bf9884093
fsqrt a little more precise; more work to be done, especially for exact squares.
Mario Pernici
2008-03-29 18:09:53 +00:00
a238976e70
implemented richardson and shanks transformations for computing limits and series
Fredrik Johansson
2008-03-29 13:49:30 +00:00
59bcd8fe5d
improve accuracy of exp. this still needs some tuning, especially in the way log calls exp_series
Fredrik Johansson
2008-03-29 13:01:16 +00:00
eec1ece08d
implement chebyshev approximation (evaluation of the polynomial in chebyshev form to be done)
Fredrik Johansson
2008-03-28 15:55:46 +00:00
7df1bc8358
slight tidying of mpc_sqrt. adding mario's extended accuracy tests. give runtests.py a -profile parameter.
Fredrik Johansson
2008-03-27 12:22:54 +00:00
27c72baf98
cmp_sqrt with improved precision
Mario Pernici
2008-03-26 21:10:59 +00:00
55515cf865
allow fmul to be used without rounding the result, saving a call to normalize
Fredrik Johansson
2008-03-26 16:34:35 +00:00
25ebf44a16
avoiding temporary roundings in complex mul to ensure correct rounding of the final value. this also gives a 40% speedup
Fredrik Johansson
2008-03-26 15:09:56 +00:00
fb20b2dbbb
better algorithm for complex tan and tanh
Fredrik Johansson
2008-03-26 12:34:44 +00:00
03b4d21b8b
implement various factorial related functions and some orthogonal polynomials. implement a function gammaquot to deal with cancellation of gamma poles in the analytic continuation of hyp2f1 and elsewhere
Fredrik Johansson
2008-03-25 20:44:46 +00:00
c5eebde994
fix the bug in the previous bugfix
Fredrik Johansson
2008-03-25 10:26:48 +00:00
7a12b09f55
fix for issue 33
Fredrik Johansson
2008-03-25 10:13:29 +00:00
ce25e91067
use 1/z transformation to calculate hyp2f1 for |z| > 1
Fredrik Johansson
2008-03-24 13:38:08 +00:00
Fredrik Johansson