October 2010 - GiNaC-devel

About a (poor) parity in expressions.
by Francois Maltey 18 Oct '10

18 Oct '10

Hello, I'm a ginac user because I use Sage, and I work around expressions. I discover that ginac remains function arguments (more than the other computer algebra system). So sin(x)+sin(-x) remains sin(x)+sin(-x), and I don't get 0. The only automatic rewrite rules I discover around signs and integers are : 1 / exp(x) ==> exp(-x) exp(x)^2 ==> exp(2*x) I don't speak about exp(log(x)) ==> x, sin(atan(x)) ==> x/sqrt(x^2+1), ... Are there other rewrite rules ? After some computations, I offen get a result as sin(x)+sin(-x) or cos(x)-cos(-x). I know it's almost impossible to choose between cos(a-b) and cos(b-a). But it's possible to choose only one expression between a-b and b-a : 1/ Look at the only numeric constant in a product, and test if it's a positive one. 2/ Look at the first term in a sum, and test this first term. Can I find it in ginac or must I use this sage function : def pseudoPositive (expr) : if expr._is_real() : return bool (RR(expr) >= 0) if expr._is_numeric () : return bool ((expr.real() > 0) or (expr.real() == 0 and expr.imag() > 0)) if expr._is_symbol() : return True opor = expr.operator() opands = expr.operands() if opor == operator.mul : return pseudoPositive (opands[-1]) if opor == operator.add : return pseudoPositive (opands[0]) return True Then I redefine the usual functions as sin by : def rewSin (expr) : if pseudoPositive (expr) : return sin (expr) else : return -sin(-expr) By this way I get all the easy simplifications. Where do ginac developpers see these rules ? inside-ginac or outside-ginac ? It's not a good idea to (re-)code these rules inside sage if I can get them by ginac. Many thanks for your advices. Francois (in France)

4 5

GiNaC for Mac OS X fink
by David Fang 13 Oct '10

13 Oct '10

Hi all, Quick announcement: A package for ginac-1.5.8 on Mac OS X has been recently added to fink (http://www.finkproject.org/) It builds against cln-1.3.1. It is currently only available in the unstable tree. Enjoy! Fang

1 0

Re: [GiNaC-devel] [GiNaC-list] archiving custom functions
by grey_earl＠web.de 13 Oct '10

13 Oct '10

Hi, there still seems a problem left - after unarchiving, substituting doesn't work anymore with realsymbols, while it works with symbols. I augmented the testcase, which now is #include <iostream> #include <fstream> #include <stdexcept> using namespace std; #include <ginac/ginac.h> using namespace GiNaC; int main() { archive aro, ari; realsymbol r("r"); symbol s("s"); ex test, test2; test = r+s; cerr << "writing archive\n"; ofstream aus("test.gar"); aro.archive_ex(test, "test"); aus << aro; aus.close(); cerr << "done writing archive\n"; cerr << "reading archive\n"; ifstream ein("test.gar"); ein >> ari; ein.close(); cerr << "done reading archive\n"; lst symlst; symlst = r,s; cerr << "unarchiving\n"; try { test2 = ari.unarchive_ex(symlst); } catch (std::exception &e) { cerr << e.what() << std::endl; } cerr << "recovered expression\n\n"; cerr << "expression = " << test2 << "\n"; test = test2.subs(lst(r==0,s==0)); cerr << "substituting (r=0,s=0) = " << test << "\n"; return 0; } Thanks! ___________________________________________________________ GRATIS! Movie-FLAT mit über 300 Videos. Jetzt freischalten unter http://movieflat.web.de

1 0

[SCM] GiNaC -- a C++ library for symbolic computations branch, master, updated. release_1-4-0-233-gcbb93fa
by git＠ginac.de 09 Oct '10

09 Oct '10

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "GiNaC -- a C++ library for symbolic computations". The branch, master has been updated via cbb93fadabbd56ba006902967b15b2b2aebb037c (commit) from e99d0d58c1bbaa8ee73e4a90a90aa1086f2f813d (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit cbb93fadabbd56ba006902967b15b2b2aebb037c Author: Alexei Sheplyakov <alexei.sheplyakov(a)gmail.com> Date: Sat Oct 9 18:39:41 2010 +0200 mul: algebraic_subs_mul(), has(): don't write beyond the end of array algebraic_match_mul_with_mul() iterates over operands of mul, that is for (size_t i=0; i<e.nops(); ++i) However, the size of arrays (`vectors' in STL speak) passed to this function is seq.size(), which is nops() - 1 for any mul object. Thus algebraic_match_mul_with_mul() accesses beyond the arrays limit. Usually it's not a problem, since any reasonable implementation of std::vector<bool> packs booleans into ints (or longs). However, some STL implementations (in particular, the one shipped with msvc) are more picky, and access beyond the vector<bool> limits results in a segfault. Therefore let's play safe and allocate proper number of elements (that is, nops()) for those arrays (subsed and currsubsed). ----------------------------------------------------------------------- Summary of changes: ginac/mul.cpp | 12 +++++++----- 1 files changed, 7 insertions(+), 5 deletions(-) hooks/post-receive -- GiNaC -- a C++ library for symbolic computations

1 0

[SCM] GiNaC -- a C++ library for symbolic computations branch, ginac_1-5, updated. release_1-4-0-252-g052f124
by git＠ginac.de 09 Oct '10

09 Oct '10

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "GiNaC -- a C++ library for symbolic computations". The branch, ginac_1-5 has been updated via 052f124551d2b78104b0de2008db6b8819713625 (commit) from c40c54f17b68f79ee42833adb3364970385e2caa (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 052f124551d2b78104b0de2008db6b8819713625 Author: Alexei Sheplyakov <alexei.sheplyakov(a)gmail.com> Date: Sat Oct 9 18:39:41 2010 +0200 mul: algebraic_subs_mul(), has(): don't write beyond the end of array algebraic_match_mul_with_mul() iterates over operands of mul, that is for (size_t i=0; i<e.nops(); ++i) However, the size of arrays (`vectors' in STL speak) passed to this function is seq.size(), which is nops() - 1 for any mul object. Thus algebraic_match_mul_with_mul() accesses beyond the arrays limit. Usually it's not a problem, since any reasonable implementation of std::vector<bool> packs booleans into ints (or longs). However, some STL implementations (in particular, the one shipped with msvc) are more picky, and access beyond the vector<bool> limits results in a segfault. Therefore let's play safe and allocate proper number of elements (that is, nops()) for those arrays (subsed and currsubsed). (cherry picked from commit cbb93fadabbd56ba006902967b15b2b2aebb037c) ----------------------------------------------------------------------- Summary of changes: ginac/mul.cpp | 12 +++++++----- 1 files changed, 7 insertions(+), 5 deletions(-) hooks/post-receive -- GiNaC -- a C++ library for symbolic computations

1 0

[PATCH] mul: algebraic_subs_mul(), has(): don't write beyond the end of array
by Alexei Sheplyakov 09 Oct '10

09 Oct '10

algebraic_match_mul_with_mul() iterates over operands of mul, that is for (size_t i=0; i<e.nops(); ++i) However, the size of arrays (`vectors' in STL speak) passed to this function is seq.size(), which is nops() - 1 for any mul object. Thus algebraic_match_mul_with_mul() accesses beyond the arrays limit. Usually it's not a problem, since any reasonable implementation of std::vector<bool> packs booleans into ints (or longs). However, some STL implementations (in particular, the one shipped with msvc) are more picky, and access beyond the vector<bool> limits results in a segfault. Therefore let's play safe and allocate proper number of elements (that is, nops()) for those arrays (subsed and currsubsed). Thanks to Jan Rheinländer for a bug report. --- ginac/mul.cpp | 12 +++++++----- 1 files changed, 7 insertions(+), 5 deletions(-) diff --git a/ginac/mul.cpp b/ginac/mul.cpp index dfda410..3733bc4 100644 --- a/ginac/mul.cpp +++ b/ginac/mul.cpp @@ -729,6 +729,9 @@ bool algebraic_match_mul_with_mul(const mul &e, const ex &pat, exmap& repls, int factor, int &nummatches, const std::vector<bool> &subsed, std::vector<bool> &matched) { + GINAC_ASSERT(subsed.size() == e.nops()); + GINAC_ASSERT(matched.size() == e.nops()); + if (factor == (int)pat.nops()) return true; @@ -760,8 +763,8 @@ bool mul::has(const ex & pattern, unsigned options) const if(is_a<mul>(pattern)) { exmap repls; int nummatches = std::numeric_limits<int>::max(); - std::vector<bool> subsed(seq.size(), false); - std::vector<bool> matched(seq.size(), false); + std::vector<bool> subsed(nops(), false); + std::vector<bool> matched(nops(), false); if(algebraic_match_mul_with_mul(*this, pattern, repls, 0, nummatches, subsed, matched)) return true; @@ -771,8 +774,7 @@ bool mul::has(const ex & pattern, unsigned options) const ex mul::algebraic_subs_mul(const exmap & m, unsigned options) const { - std::vector<bool> subsed(seq.size(), false); - exvector subsresult(seq.size()); + std::vector<bool> subsed(nops(), false); ex divide_by = 1; ex multiply_by = 1; @@ -781,7 +783,7 @@ ex mul::algebraic_subs_mul(const exmap & m, unsigned options) const if (is_exactly_a<mul>(it->first)) { retry1: int nummatches = std::numeric_limits<int>::max(); - std::vector<bool> currsubsed(seq.size(), false); + std::vector<bool> currsubsed(nops(), false); exmap repls; if(!algebraic_match_mul_with_mul(*this, it->first, repls, 0, nummatches, subsed, currsubsed)) -- 1.7.1

1 0

Is this a bug?
by Jan private 08 Oct '10

08 Oct '10

Hello Alexei, while compiling an application with MSVC, I stumbled on this (because the code crashed) File: mul.cpp Function: ex mul::algebraic_subs_mul(const exmap & m, unsigned options) These two vectors are sized to seq.size(): std::vector<bool> subsed(seq.size(), false); std::vector<bool> currsubsed(seq.size(), false); They are then used in: algebraic_match_mul_with_mul() This iterates over the nops() of the mul: for (size_t i=0; i<e.nops(); ++i) { Can it be that the nops() of the mul is (always) equal to seq.size() + 1? Output of one example: ----- Code inserted in mul::algebraic_subs_mul(): std::cout << "mul::algebraic_subs_mul: nops: " << this->nops() << ", seq.size(): " << seq.size() << std::endl; std::cout << "mul::algebraic_subs_mul: " << tree << *this << dflt << std::endl; ------------ Output: mul::algebraic_subs_mul: nops: 2, seq.size(): 1 mul::algebraic_subs_mul: mul @00769848, hash=0x769820, flags=0x3, nops=2 mm 1 (numeric) @00753360, hash=0x160af41d, flags=0xf ----- overall_coeff 1/10 (numeric) @00769828, hash=0x36fbefb6, flags=0xf ===== ------------- Because the code crashes on the last iteration of the for loop, and if I size the vectors to seq.size()+1 (or maybe better nops()) it does not crash any more. Another possibility is that there is another error earlier on which results in a mul with seq.size() = 1 and nops() = 2 (for my example). Regards, Jan

2 1

[SCM] GiNaC -- a C++ library for symbolic computations branch, master, updated. release_1-4-0-232-ge99d0d5
by git＠ginac.de 04 Oct '10

04 Oct '10

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "GiNaC -- a C++ library for symbolic computations". The branch, master has been updated via e99d0d58c1bbaa8ee73e4a90a90aa1086f2f813d (commit) from 8fa654f2637fa755e890499ece1a7bc498779ea3 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit e99d0d58c1bbaa8ee73e4a90a90aa1086f2f813d Author: Alexei Sheplyakov <alexei.sheplyakov(a)gmail.com> Date: Mon Oct 4 09:21:05 2010 +0200 Avoid infinite loop when unarchiving realsymbol and possymbol. symbol::read_archive(): explicitly set status_flags::evaluated (and status_flags::expanded) on object being unarchived. These flags get reset by basic::operator=(const basic&) for realsymbol and possymbol, and nothing sets (except symbol ctor), so automatic evaluation never terminates (or rather, terminates due to a stack overflow). Therefore it's necessary need to set status_flags::evaluated explicitly. Thanks to Markus Fröb for a bugreport and a test case. ----------------------------------------------------------------------- Summary of changes: ginac/symbol.cpp | 9 +++++++++ 1 files changed, 9 insertions(+), 0 deletions(-) hooks/post-receive -- GiNaC -- a C++ library for symbolic computations

1 0

[SCM] GiNaC -- a C++ library for symbolic computations branch, ginac_1-5, updated. release_1-4-0-251-gc40c54f
by git＠ginac.de 04 Oct '10

04 Oct '10

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "GiNaC -- a C++ library for symbolic computations". The branch, ginac_1-5 has been updated via c40c54f17b68f79ee42833adb3364970385e2caa (commit) from 89d5356b4aa33cb4481575f9453f36c3404b015b (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit c40c54f17b68f79ee42833adb3364970385e2caa Author: Alexei Sheplyakov <alexei.sheplyakov(a)gmail.com> Date: Mon Oct 4 09:21:05 2010 +0200 Avoid infinite loop when unarchiving realsymbol and possymbol. symbol::read_archive(): explicitly set status_flags::evaluated (and status_flags::expanded) on object being unarchived. These flags get reset by basic::operator=(const basic&) for realsymbol and possymbol, and nothing sets (except symbol ctor), so automatic evaluation never terminates (or rather, terminates due to a stack overflow). Therefore it's necessary need to set status_flags::evaluated explicitly. Thanks to Markus Fröb for a bugreport and a test case. (cherry picked from commit e99d0d58c1bbaa8ee73e4a90a90aa1086f2f813d) ----------------------------------------------------------------------- Summary of changes: ginac/symbol.cpp | 9 +++++++++ 1 files changed, 9 insertions(+), 0 deletions(-) hooks/post-receive -- GiNaC -- a C++ library for symbolic computations

1 0

[PATCH] [bugfix] Avoid infinite loop when unarchiving realsymbol and possymbol.
by Alexei Sheplyakov 02 Oct '10

02 Oct '10

symbol::read_archive(): explicitly set status_flags::evaluated (and status_flags::expanded) on object being unarchived. These flags get reset by basic::operator=(const basic&) for realsymbol and possymbol, and nothing sets (except symbol ctor), so automatic evaluation never terminates (or rather, terminates due to a stack overflow). Therefore it's necessary need to set status_flags::evaluated explicitly. Thanks to Markus Fröb for a bugreport and a test case. --- ginac/symbol.cpp | 9 +++++++++ 1 files changed, 9 insertions(+), 0 deletions(-) diff --git a/ginac/symbol.cpp b/ginac/symbol.cpp index 9897bb4..8d05238 100644 --- a/ginac/symbol.cpp +++ b/ginac/symbol.cpp @@ -109,6 +109,15 @@ void symbol::read_archive(const archive_node &n, lst &sym_lst) for (lst::const_iterator it = sym_lst.begin(); it != sym_lst.end(); ++it) { if (is_a<symbol>(*it) && (ex_to<symbol>(*it).name == tmp_name)) { *this = ex_to<symbol>(*it); + // XXX: this method is responsible for reading realsymbol + // and possymbol objects too. But + // basic::operator=(const basic& other) + // resets status_flags::evaluated if other and *this are + // of different types. Usually this is a good idea, but + // doing this for symbols is wrong (for one, nothing is + // going to set status_flags::evaluated, evaluation will + // loop forever). Therefore we need to restore flags. + setflag(status_flags::evaluated | status_flags::expanded); return; } } -- 1.7.1

1 0