October 2005 - GiNaC-list

ginac 1.3.3 fails make check under Mac OS X 10.4.x
by Marius Schamschula 08 Nov '05

08 Nov '05

Hello all, I've been running into a problem building ginac under Mac OS X 10.4.x (Tiger)/Darwin 8.x. I can compile w/o any problems, but when I run make check I get the following error: /bin/sh ../libtool --tag=CXX --mode=link g++ -g -O2 -o exams exam_paranoia.o exam_numeric.o exam_powerlaws.o exam_inifcns.o exam_inifcns_nstdsums.o exam_differentiation.o exam_polygcd.o exam_normalization.o exam_pseries.o exam_matrices.o exam_lsolve.o exam_indexed.o exam_color.o exam_clifford.o exam_archive.o exam_structure.o exam_hashmap.o exam_misc.o exams.o ../ginac/ libginac.la -L/usr/local/lib -lcln -lgmp g++ -g -O2 -o .libs/exams exam_paranoia.o exam_numeric.o exam_powerlaws.o exam_inifcns.o exam_inifcns_nstdsums.o exam_differentiation.o exam_polygcd.o exam_normalization.o exam_pseries.o exam_matrices.o exam_lsolve.o exam_indexed.o exam_color.o exam_clifford.o exam_archive.o exam_structure.o exam_hashmap.o exam_misc.o exams.o -Wl,-bind_at_load ../ginac/.libs/ libginac.dylib -L/usr/local/lib /usr/local/lib/libcln.dylib /usr/ local/lib/libgmp.dylib /usr/bin/ld: Undefined symbols: GiNaC::structure<sprod_s, GiNaC::compare_std_less>::reg_info collect2: ld returned 1 exit status make[3]: *** [exams] Error 1 make[2]: *** [check-am] Error 2 make[1]: *** [check-recursive] Error 1 make: *** [check] Error 2 Note: I don't see this problem under Mac OS 10.3.x (Panther)/Darwin 7.x. I tried building using both gcc 3.3 and gcc 4.0. gcc 3.3 dies with some linking errors. Any ideas? TIA Marius -- Marius Schamschula, Alabama A & M University, Department of Physics The Center for Hydrology Soil Climatology and Remote Sensing http://optics.physics.aamu.edu/ - http://www.physics.aamu.edu/ http://wx.aamu.edu/ - http://www.aamu.edu/hscars/

2 4

bug(?) in clifford_moebius_map
by varg＠theor.jinr.ru 31 Oct '05

31 Oct '05

Hello! I've noticed that example code from cs.MS/0410044 ("An Example of Clifford Algebras Calculations with GiNaC") does not work. $ g++ -O0 -g -o parabolic parabolic1-bin.cpp `ginac-config --libs` $ ./parabolic Metric is: e. *** Got problem in vector fields: Expression is a non-scalar Clifford number! *** Got problem in vector fields: Expression is a non-scalar Clifford number! *** Got problem in vector fields: Expression is a non-scalar Clifford number! Vect field Direct In Cayley In Cayley1 Segmentation fault (core dumped) $ ginac-config --version 1.3.3 $ g++ --version g++ (GCC) 4.0.3 20051023 (prerelease) (Debian 4.0.2-3) Copyright (C) 2005 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. I think the reason is that clifford_moebius_map mixes matrices and indexed expressions. $ gdb --silent ./parabolic [some boring stuff snipped] (gdb) break clifford.cpp:1023 (gdb) bt #0 GiNaC::clifford_moebius_map (a=@0x8088548, b=@0x808854c, c=@0x8088550, d=@0x8088554, v=@0x5ca926d4, G=@0x5ca926d0, rl=0 '\0') at clifford.cpp:1203 #1 0x2e89292e in GiNaC::clifford_moebius_map (M=@0x5ca9205c, v=@0x5ca926d4, G=@0x5ca926d0, rl=0 '\0') at clifford.cpp:1210 #2 0x2e8929fb in GiNaC::clifford_moebius_map (M=@0x5ca9205c, v=@0x5ca926d4, G=@0x5ca926d0) at clifford.cpp:1217 #3 0x08052acf in main (argc=1, argv=0x5ca94034) at parabolic1-bin.cpp:604 (gdb) frame 0 #0 GiNaC::clifford_moebius_map (a=@0x8088548, b=@0x808854c, c=@0x8088550, d=@0x8088554, v=@0x5ca926d4, G=@0x5ca926d0, rl=0 '\0') at clifford.cpp:1203 1203 return clifford_to_lst(e, cu, false); (gdb) call e.dbgprint() e~symbol13*[[x],[y]].symbol13*exp(t)*exp(-t)^(-1) (gdb) Due to this meaningless combination of matrix and indexed expression clifford_to_lst fails, and since the example code is not really bullet-proof, the end result is a segfault. Any ideas? Best regards, Alexei -- All science is either physics or stamp collecting.

2 3

Unsubscribe notifications
by Christian Bauer 25 Oct '05

25 Oct '05

Hi! If you've been getting "You have been unsubscribed from ginac-list@..." messages, these are from the old, defunct lists at thep.physik.uni-mainz.de. I noticed they were still active and, lacking the means to properly delete them, unsubscribed everyone (and configured them to discard all incoming messages). I guess I should have disabled the "Send good-bye message" option first... If you are reading this, you are still subscribed to the real GiNaC mailing lists at ginac.de, and you do not have to take any action. I'm sorry if I caused any confusion. Bye, Christian -- / Physics is an algorithm \/ www.cebix.net

1 0

GiNaC 1.3.3 release
by Richard B. Kreckel 24 Oct '05

24 Oct '05

Hi, GiNaC 1.3.3 is out and available. The changes are: - Fixed bug occuring in algebraic substitutions with expressions involving indices. - Fixed bug that selected the wrong algorithm for determinants of purely numeric matrices with at least one floating point element. - Fixed bug in function H that caused an infinite recursion for arguments around +-I. - Fixed const-correctness in printing handler for GCC 4.0.2. - Made lookup in adaptivesimpson precision-aware. - Added series expansion for functions (classical) Li and S around x==0. - Added fsolve() numerical univariate real-valued function solver. - Added functions sub_matrix() and reduced_matrix(). - Small cleanups. Less warnings with latest GCC. As always, this release can be downloaded from http://www.ginac.de/Download.html Enjoy, -richy. -- Richard B. Kreckel <http://www.ginac.de/~kreckel/>

1 0

ViewCVS frontend
by Christian Bauer 15 Oct '05

15 Oct '05

Hi! The GiNaC CVS can also be browsed at http://www.ginac.de/viewcvs/GiNaC/ Bye, Christian -- / Physics is an algorithm \/ www.cebix.net

1 0

CVS server moved
by Christian Bauer 15 Oct '05

15 Oct '05

Hi! As previously announced, the GiNaC CVS server has moved to the new location cvs.ginac.de. To do an anonymous checkout: $ cvs -d :pserver:anoncvs@cvs.ginac.de:/home/cvs/GiNaC login (password is 'anoncvs') $ cvs -d :pserver:anoncvs@cvs.ginac.de:/home/cvs/GiNaC co GiNaC If you are working with a copy checked out from the old CVS server, you need to do a fresh checkout from cvs.ginac.de. Happy hacking, Christian -- / Physics is an algorithm \/ www.cebix.net

1 0

We are moving...
by Richard B. Kreckel 08 Oct '05

08 Oct '05

Dear all, Due to ongoing technical problems at Mainz University (*), it was decided to move GiNaC's and CLN's web server / CVS server / mailing lists to a rented server. This should allow us to continue communicating via mailing lists, doing development and updating the web pages, things that have become difficult during the last 10 weeks or so. We are going to move the DNS entries ASAP. The hosted machine's IP-address is going to be 81.169.130.167. We'll certainly see some outage while the transfer occurs, but hopefully not for long time. Meanwhile, you can all make sure the address you use as mailing list addresses is ...(a)ginac.de, not ...(a)thep.physik.uni-mainz.de or ...(a)zino.physik.uni-mainz.de because these will cease to work. We apologize for the inconveninece -richy. (*) Case in point: right now the web server is dead. -- Richard B. Kreckel <http://www.ginac.de/~kreckel/>

1 0

Breaking News
by Richard B. Kreckel 07 Oct '05

07 Oct '05

Erlangen, October 6, 2005 (for immediate release) The GiNaC/CLN hackers, the leading providers of advanced mathematical software solutions for research and industrial applications, are proud to announce that their system scored second in the "Many Digits" Friendly Competition [0], held 3rd and 4th October at the Radboud University in Nijmegen, The Netherlands. Although Richard Kreckel, former vice president of GiNaC's PR department has retired into industry, he was available to compete against the world's finest exact real arithmetic hackers. "It was a thrilling experience" said Richard Kreckel at a meeting with press. "I basically started hacking on the competition problems Monday morning and stopped only an hour before our turn at the competition Tuesday afternoon. Little sleep compensated only by beer, coffee and lots of caffeinated peppermints. It was quite similar to those release parties at University." Maybe it wasn't enough caffeine, though. The winner at the competition was the MPFR team headed by Paul Zimmermann. Says Richard Kreckel: "Well, they certainly cut some corners. They discovered, coded and tested closed-form solutions for hard iterative problems, thus being better than anybody else by orders of magnitude. Their winning is well-deserved. Congratulations!" Other competing systems were iRRAM, Wolfram's Mathematica, Maple, RealLib, COMP, Few Digits and Bignum. Although only provisional rankings have been published so far [1], ground-shaking changes in the results are not expected. The "Many Digits" Friendly Competition was of purely numerical nature: none of the 24 competition problems called for symbolic manipulations. Hence, it really was a measure of CLN's numerical power, rather than GiNaC's symbolic flexibility. Bruno Haible, the original author of the CLN library could not be reached for a statement. It is assumed that he is busy hunting down bugs in several dozen GNU software packages at the same time. He has all the time. After all, he has retired to industry, too. [0] <http://www.cs.ru.nl/~milad/manydigits/> [1] <http://www.cs.ru.nl/~milad/manydigits/raw_provisional.html>

1 0

Problem compiling CLN on Darwin.
by Chris Dams 06 Oct '05

06 Oct '05

Dear all, I tried to compile CLN on Darwin. To be precise, uname -a says Darwin macfarland.local 7.9.0 Darwin Kernel Version 7.9.0: Wed Mar 30 20:11:17 PST 2005; root:xnu/xnu-517.12.7.obj~1/RELEASE_PPC Power Macintosh powerpc When running make check, the following error is produced cd tests; make check ./exam dyld: /Users/dams/download/cln-1.1.9/tests/.libs/exam Undefined symbols: ___gxx_personality_v0 __ZNSt8ios_base4InitC1Ev __ZNSt8ios_base4InitD1Ev __ZSt9terminatev __ZSt4cerr __ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc __ZNSo3putEc Apparently the dynamic library loader cannot find some symbols of the standard C++ library. However, that is very strange, as having the linker printing diagnostics reveals that the static library /usr/lib/gcc/darwin/3.3/libstdc++.a is used. Does anyone know more about this platform, and how I can solve this problem? Best wishes, Chris

2 2

normalization ambiguity
by Jonathan Brandmeyer 05 Oct '05

05 Oct '05

I recently discovered that expressions like pow(x - y, 2) can normalize to pow(y - x, 2) if y compares less than x (according to the ex_is_less comparator). Is this behavior deliberate? A particular test in the test suite (check/exam_normalization.cpp) normally only passes when pow(x - y, 2).normal() == pow(x-y, 2). Thanks, -Jonathan

2 2