Karim Belabas on Wed, 22 Dec 2004 19:35:42 +0100

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

pari-2.2.9 released

Dear PARI enthusiast,

I would like to announce the release of pari-2.2.9.ALPHA. The sources can be
obtained through the address



  - The old 'intnum' has been obsoleted [ renamed intnumromb ] and
    replaced by an array of more versatile and more powerfull functions,
    with a rather bewildering array of options, see ??intnum for a start.

    The algorithms are variants of standard "double exponential" methods
    and are quite capable of computing integrals to hundreds or thousands
    of decimal digits. They are sensitive to the accuracy of user input
    though [ order of singularities or oscillating behaviour at oo, for
    instance ], so read the documentation first.

  - faster algorithms for Pi, log, atan, acos, asin.

  - [ libpari ] gcoeff() & friends are now valid lvalues. Explicit typecasts
    between GENs and longs are no longer needed. See the libpari manual.

  - fixes various glitches in 2.2.8

GP's online documentation at


has been updated.

Have fun� and Merry Christmas !


P.S: The Changelog:

Done for version 2.2.9 (released 23/12/2004):
    1- ploth(,256). splines had stopped working
    2- gamma(-1.), psi(-1.) [ non-positive integers masquerading as t_REALs ]:
       wrong results (oo) or 'division by 0' error messages
    3- log(1. + 1e-20) --> print more decimals than are significant
    4- [Mac OS X + Fltk]: remove the BROKEN_FORK assertion --> graphic
       window becomes independent of gp as under other OSes.
    5- elllseries did not accept complex arguments
GN  6- [ix86 + Sun cc] build fails [ unsatisfied symbol reference to 'mulll' ]
    7- [libpari] gprec_w didn't behave well for real zero
    8- gp -p 51234567890 --> internal overflow and primelimit set to smaller
    9- [ GMP kernel ] incorrect resmod2n --> wrong 2-adic sqrt
   10- [hi-res plot: fltk] fix compilation with g++-3.4
BA 11- [hi-res plot: fltk] line plotting was broken
   12- oo loops in prodinf [ e.g prodinf(x=2, zeta(x)) ]
   13- [compatible=3] wrong prototype for isoncurve, lex, sign, thetanullk
   14- quadhilbert(D > 0) spent an unreasonable time to find a relative
       equation defined over Q [ use Galois theory instead of naive search ]
   15- typo in resmod2n [ missing (ulong) typecast ] -> issquare((2^64+1)^2)=0
BA 16- galoisinit() could fail to find a polynomial defining the fixed field.
   17- elllseries had an apparent singularity at s = 2 [ loss of accuracy ]
   18- gcoeff, gmael & their variants were not lvalues.
   19- typo in element_val(t_FRAC) [ "non invertible" errors in rnfpseudobasis ]
   20- added -fno-strict-aliasing to gcc flags [ PARI code doesn't follow
       strict aliasing rules. In any case, gcc 3.3 can't handle new mael ]
   21- 1.a --> 1.0000 [ should be error ].
       (x.a = x+1); 1.a --> 1.0000 [ should be 2 ]
   22- bnrdisclist(bnfinit(x,2),1) --> SEGV
GN 23- [Solaris cc] use safer optimization flag [ -xalias_level=any, fix a
       pb with 18 above ]

    1- increased maximum binary exponent for t_REAL, and maximum GEN length
       on 64 bit machines. Bumped BINARY_VERSION for writebin --> objects
       saved in writebin format between 2.2.1 and 2.2.8 are incompatible.
    2- use easier to differentiate colors for hi-res plots [ replace "sienna"
       by "violetred" and "cornsilk" by "green" ]
    3- GP interface of function intnum [ old flags not recognized, much more
       elaborate interface ]
    4- library interface of functions intnum, prodeuler, suminf, sumalt,
       sumalt2, sumpos, sumpos2, prodinf, prodinf1  [ GEN (*eval)(GEN,void*)
       everywhere instead of entree * ]
BA  5- allow Configure -l <directory>  [ in addition to <directory>/pari.cfg ]
    6- renamed gzero --> gen_0, gun --> gen_1, gdeux --> gen_2.
    7- replaced kludgy matexp.c example by straightforward extgcd

    1- new algorithm for log( t_COMPLEX ), log(2) [ AGM ]
    2- new algorithm for Pi [ AGM ]
    3- new algorithm for atan, acos, asin [ AGM ]
HC  4- "double exponential method" for numerical integration and sumation.
       See ??intnum. New functions intcirc, intfouriercos, intfourierexp,
       intfouriersin, intfuncinit, intlaplaceinv, intmellininv,
       intmellininvshort, intnuminit, intnuminitgen, intnumromb, intnumstep,
       sumnum, sumnumalt, sumnuminit
    5- new universal constant gen_m1
    6- [bench suite] test-intnum and test-stark toplevel targets
Karim Belabas                     Tel: (+33) (0)1 69 15 57 48
Dep. de Mathematiques, Bat. 425   Fax: (+33) (0)1 69 15 60 19
Universite Paris-Sud              http://www.math.u-psud.fr/~belabas/
F-91405 Orsay (France)            http://pari.math.u-bordeaux.fr/  [PARI/GP]