- GiNaC-list - ginac.de

Re: Airy functions
by Jan Bos 24 Dec '07

24 Dec '07

Dear Alexei, Thank you for your interest and careful examination of the code. I made an effort to fit the functions in the GiNaC framework and got series expansions etc. more or less for free. I'm looking forward to seeing the result when it meets the GiNaC standards. To answer your questions a fundamental property of the algorithm is needed: Depending on the argument a so called "steepest descent" contour is determined along which the integrand is monotonically decreasing. This ensures that a) roundoff does not occur, b) the range over which the integrand has to be evaluated to obtain a certain accuracy is minimized and c) the dominant factor (possibly very larger or small) can be placed outside the integral. Furthermore all derivatives at +/- infinity are zero. For functions with this property the trapezoidal rule has exponential convergence, that is, halving the stepsize more than doubles the number of significant digits of the result. For an integrator with non-constant step size this property would be lost. > AiryContour* pcntr=createAiryContour(zn); > > Pointers are evil. What is the reason for explicit allocation/deallocation? > Again depending on the argument (zn) but more involved, 3 different types of contours are needed of which one is the truly steepest descent contour and the other 2 are approximations thereof. AiryContour is the base class of them. I hope this answers your questions Best regards, Jan

2 1

A differential geometry-specific extension
by Diego Conti 21 Dec '07

21 Dec '07

Hi, I have been working on an extension to GiNaC for differential geometry, and I have released a first version, available from <https://sourceforge.net/projects/libwedge/>. Current features include: - Vector spaces: determine a basis from a list of generators, and similar computations. - Manifolds and differential forms: exterior derivative; wedge product. - Lie groups: the general linear group; subgroups determined by the choice of a subalgebra; abstract Lie groups defined in abbreviated form, e.g. writing (0,0,12) for the Heisenberg group, characterized by the existence of a basis of left-invariant one-forms e1,e2,e3 such that de3=e1 ^ e2 and e1,e2 are closed. - Riemannian metrics and G-structures, defined on a Lie group or a coordinate patch of a generic manifold, and represented by an orthonormal basis of 1-forms (adapted frame); spinors, Clifford multiplication. - Connections: the Levi-Civita connection; curvature; covariant derivatives; define connections on generic manifolds and impose conditions on the Christoffel symbols, e.g. to obtain curvature conditions. I thought maybe someone in this list would be interested. Diego Conti

2 2

ginac installation
by alexander baker 05 Dec '07

05 Dec '07

Am getting the following error when I try and make GiNac, the configure step has worked ok, please see below. Regards Alex Baker. Configuration of GiNaC done. Now type "make". root@treebeard:/home/bakera/inglethorpe/ginac/GiNaC# make cd . \ && CONFIG_FILES= CONFIG_HEADERS=config.h \ /bin/bash ./config.status config.status: creating config.h config.status: config.h is unchanged make all-recursive make[1]: Entering directory `/home/bakera/inglethorpe/ginac/GiNaC' Making all in ginac make[2]: Entering directory `/home/bakera/inglethorpe/ginac/GiNaC/ginac' @LIBTOOL@ --mode=compile g++ -DHAVE_CONFIG_H -I. -I. -I.. /usr/local/lib -g -O2 -c add.cpp /bin/bash: line 1: @LIBTOOL@: command not found make[2]: *** [add.lo] Error 127 make[2]: Leaving directory `/home/bakera/inglethorpe/ginac/GiNaC/ginac' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/home/bakera/inglethorpe/ginac/GiNaC' make: *** [all-recursive-am] Error 2 -- Mobile: 07788 872118 Blog: www.alexfb.com

2 11

problem running ginac
by alexander baker 04 Dec '07

04 Dec '07

The configure, make and install worked fine, however compiling a simple example below gives the following: #include <iostream> #include <ginac/ginac.h> using namespace std; using namespace GiNaC; int main(){ numeric two = 2; } Running on Ubuntu dual core amd 64 bit. Alex /usr/local/lib/libginac.so: undefined reference to `GiNaC::get_symbol_name(GiNaC::ex const&)' /usr/local/lib/libginac.so: undefined reference to `GiNaC::is_lexer_symbol_predefined(GiNaC::ex const&)' /usr/local/lib/libginac.so: undefined reference to `ginac_yytext' /usr/local/lib/libginac.so: undefined reference to `GiNaC::set_lexer_symbols(GiNaC::ex)' /usr/local/lib/libginac.so: undefined reference to `ginac_yylex()' /usr/local/lib/libginac.so: undefined reference to `GiNaC::set_lexer_string(std::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)' /usr/local/lib/libginac.so: undefined reference to `ginac_yyrestart(_IO_FILE*)' collect2: ld returned 1 exit status -- Mobile: 07788 872118 Blog: www.alexfb.com

1 0

Trouble building example from manual
by Ken Smith 27 Nov '07

27 Nov '07

First my environment. Mac OS 10.5.1 (Darwin 9.1.0) gcc version 4.0.1 (Apple Inc. build 5465) readline-5.1 gnump-4.2.1 cln-1.1.9 GiNaC-1.4.1 I get a coredump when I run the first example from section 2.1 of the GiNaC 1.4.1 Tutorial, revision 5 September 2007. Here is the stack trace. Program received signal EXC_BAD_ACCESS, Could not access memory. Reason: KERN_PROTECTION_FAILURE at address: 0x00000002 0x00000002 in ?? () (gdb) bt #0 0x00000002 in ?? () #1 0x006d6480 in cl_module__cl_prin_globals__firstglobalfun () at output/cl_prin_globals.cc:6 Cannot access memory at address 0x6 I built my program like this: g++ -I/usr/local/include -L/usr/local/lib src/tut.cpp -lginac I also tried adding -lcln -lgmp and -lreadline to the link but the result was the same. This same error happens when I attempt to run ginsh. I was concerned during the build when configure said: === The following minor problems have been detected by configure. === Please check the messages below before running "make". === (see the section 'Common Problems' in the INSTALL file) == Unsupported version of libreadline. == I could not run a test of libreadline (needed for building ginsh). Configuration of GiNaC 1.4.1 done. Now type "make". I ignored this and built anyway since I don't plan to use ginsh, just the library. Perhaps the issues are unrelated. Please help me to get started with building and running the tutorial examples. Ken Smith

2 2

Disabling flattening of sums and products?
by Martin Sandve Alnæs 27 Nov '07

27 Nov '07

Is it possible to disable the automatic flattening of expressions like f = (x+y) + (x+y) into g = 2*y+2*x ? In other words, I want to be able to retain the additions in expression f as binary operators. The reason for this is that I want to examine some expression to detect subtrees that occur multiple times, and this automatic flattening destroys some opportunities for this. Example: a = x*y b = (x*y) * (x*y) Examining a I will see a mul with ops x,y, while examining b I will see a mul with ops power(y,2), power(x,2), requiring some kind of factorization to recognize the similarities. -- Martin

3 3

GiNaC releases 1.3.9 and 1.4.1
by Jens Vollinga 21 Nov '07

21 Nov '07

Hello everybody, GiNaC 1.4.1 is out and available. The changes are: * Fixed memory leak in ginac_yylex(). * Fixed memory leaks in mul::eval() and power::eval(). * Fixed macro checking for version of libreadline (Mac OS X). * Fixed broken preprocessor instruction in excompiler.cpp. * Fixed broken module loading in excompiler.cpp. * info(info_flags::has_indices) now works for sums and products. * Improved mul::expand(). * Improved CLN output. As always, this release can be downloaded from http://www.ginac.de/Download.html We also made a backporting bug-fix release for the GiNaC 1.3.x series, which contains the following changes: * Fixed memory leak in ginac_yylex(). * Fixed memory leaks in mul::eval() and power::eval(). * Fixed macro checking for version of libreadline (Mac OS X). * info(info_flags::has_indices) now works for sums and products. * Improved CLN output. This package can be downloaded from the ftp server: ftp://ftpthep.physik.uni-mainz.de/pub/GiNaC/ Best wishes, Jens

1 0

failing to compile expressions to C function pointers
by Marian Anghel 19 Nov '07

19 Nov '07

I'm trying to implement the example that shows how to define a GiNaC expression and to compile an associated C code. My implementation of the 5.15.3 example in ginac_tutorial.pdf is as follows: #include <iostream> #include <ginac/ginac.h> #include <iterator> #include <numeric> using namespace std; using namespace GiNaC; ex HermitePoly(const symbol & x, int n) { ex HKer=exp(-pow(x, 2)); // uses the identity H_n(x) == (-1)^n exp(x^2) (d/dx)^n exp(-x^2) return normal(pow(-1, n) * diff(HKer, x, n) / HKer); } int main() { try { symbol x("x"); ex myexpr = HermitePoly(x,2); ; FUNCP_1P fp; compile_ex(myexpr, x, fp); cout << fp(1) << endl; } catch(exception &p){ cerr << p.what() << endl; } return 0; } NOTES: 1) The code compiles but the attempt to execute it gives this error: excompiler::link_so_file: could not open compiled module! 2) I'm not sure that my installation of GiNaC went right. I'm not very skilled with software installation and I did it in my home directory using the prefix configuration option. All examples tested thus far have worked correctly except for this one. 3) I'm not sure about the relationship to the error I'm getting but I'll mention this anyway: The doc/examples directory does not have any of the compile[123].cpp examples mentioned in ginac-examples.pdf. It has instead the archive1.cpp example. I'll appreciate your help in fixing this problem. Thanks! Marian Anghel

2 1

I could not find libreadline
by bernd.speiser＠t-online.de 18 Nov '07

18 Nov '07

Hi all, I am installing ginac 1.4.0 on an openSuSE 10.3 machine. I have installed libreadline: # ll /lib/*readline* lrwxrwxrwx 1 root root 21 Nov 15 10:02 /lib/libreadline.so -> /lib/libreadline.so.5 lrwxrwxrwx 1 root root 23 Oct 13 09:59 /lib/libreadline.so.4 -> /lib/libreadline.so.5.2 lrwxrwxrwx 1 root root 18 Oct 8 20:55 /lib/libreadline.so.5 -> libreadline.so.5.2 -rwxr-xr-x 1 root root 198824 Sep 22 00:16 /lib/libreadline.so.5.2 Still, configuring ginac results in: === The following minor problems have been detected by configure. === Please check the messages below before running "make". === (see the section 'Common Problems' in the INSTALL file) == I could not find libreadline (needed by ginsh). In config.log, I find: configure:22950: checking for readline in -lreadline configure:22980: g++ -o conftest -g -O2 conftest.cc -lreadline >&5 /usr/local/lib/libreadline.so: undefined reference to `PC' /usr/local/lib/libreadline.so: undefined reference to `tgetflag' /usr/local/lib/libreadline.so: undefined reference to `tgetent' /usr/local/lib/libreadline.so: undefined reference to `UP' /usr/local/lib/libreadline.so: undefined reference to `tputs' /usr/local/lib/libreadline.so: undefined reference to `tgoto' /usr/local/lib/libreadline.so: undefined reference to `tgetnum' /usr/local/lib/libreadline.so: undefined reference to `BC' /usr/local/lib/libreadline.so: undefined reference to `tgetstr' collect2: ld returned 1 exit status which suggests that libreadline is looked for in /usr/local/lib, but not in /lib. Can this behavior be changed? Not even installing libreadline in /usr/local/lib like # ll /usr/local/lib/*readline* -rw-r--r-- 1 root root 659580 Nov 13 10:54 /usr/local/lib/libreadline.a lrwxrwxrwx 1 root root 16 Nov 13 10:54 /usr/local/lib/libreadline.so -> libreadline.so.5 lrwxrwxrwx 1 root root 18 Nov 13 10:54 /usr/local/lib/libreadline.so.5 -> libreadline.so.5.2 -r-xr-xr-x 1 root root 484002 Nov 13 10:54 /usr/local/lib/libreadline.so.5.2 does help. Any suggestion, what could be the reason? Best regards Bernd -- ======================================================================= Bernd Speiser Institut für Organische Chemie Auf der Morgenstelle 18 temporary address: Auf der Morgenstelle 15 D-72076 Tübingen Germany phone: +49-7071-2976205 (office) +49-7071-2976242 (laboratory) fax: +49-7071-295518 e-mail: bernd.speiser(a)uni-tuebingen.de Internet: http://www.uni-tuebingen.de/speiser =======================================================================

3 3

memory leak when constructing expression from a string
by Alexei Sheplyakov 16 Nov '07

16 Nov '07

Hello! This silly program #include <iostream> #include <stdexcept> #include <ginac/ginac.h> using namespace std; using namespace GiNaC; int main(int argc, char** argv) { ex e; const symbol x("x"), y("y"); const lst syms(x, y); static const char* str[] = { "x^2+2*x*y + cos(x)", "Li2(x/y) + log(y/x)" }; unsigned N=0; try { for (; ; N++) { e = ex(str[N & 1], syms); } } catch (...) { cerr << "N = " << N << endl; return 1; } return 0; } exhausts all the available memory. I don't think this is correct behaviour. Any suggestions how to fix this? Best regards, Alexei -- All science is either physics or stamp collecting.

1 0