- GiNaC-list - ginac.de

Symbolic integration of trigonometric functions
by Jan 27 Dec '13

27 Dec '13

Hi, the Ginac documentation says: "GiNaC is not very good (yet?) at symbolically evaluating integrals" so I though I might help it along. Code for sin() and cos() is in this branch which is based on the current master: https://github.com/jrheinlaender/ginac/tree/jrheinlaender/integral and if you like it I might add more functions. Jan Rheinländer

1 0

Fwd: GiNaC error when trying to evaluate sin(x)^2 expression
by José Pontes 22 Dec '13

22 Dec '13

Hi, When subsiting x by a float in a sin(x)^2 expression, I get the following error: libc++abi.dylib: terminating with uncaught exception of type cln::notreached_exception: Internal error: statement in file float/elem/cl_F_square.cc, line 21 has been reached!! using this code: symbol x; ex exTest = sin(x)*sin(x); exmap map; map[x] = 1.0; cout << exTest.subs(map) << endl; I need to do similar substitutions in a matrix, and they allways fail for expressions that contain cos(x)^2 or sin(x)^2, but no for ones that contain cos(x)*sin(x), for example. Can you help me understanding this error and/or getting around it? Regards

2 1

Compiling GiNaC on OS X
by David Seery 30 Oct '13

30 Oct '13

Hello, >> From the error message I can conclude that there are two constructors >> applyable - which one is the correct one? E.g. casting 0 to ex (changing >> the line to: H(lst( ex(0) ),1/arg).hold() solves the error - is this the >> correct cast or is the other cast the correct one? > > According to the manual, the harmonic polylogarithm H expects a general > lst as first argument. So, lst(ex(0)) is correct and lst(int(0)) would > be incorrect. > > But I wonder why this clarification is needed. IMHO, the first candidate > shouldn't be considered, if the std::list<T>(T) ctor is explicit, or am > I missing something? > > Does the attached patch work for you? I am also experiencing this problem, and the patch fixes it for me. David

1 0

Compiling GiNaC on OS X
by Stefan Schupp 23 Oct '13

23 Oct '13

Hi, I am trying to compile GiNaC under OS X with clang++ and the usage of libc++ (this is required, as my project, where I want to include GiNaC requires the c++11 standard). Therefore I added the following configuration options: ./configure CXX=/usr/bin/clang++ CXXFLAGS="-std=c++11 -stdlib=libc++" LDFLAGS="-stdlib=libc++" Upon invoking make I get an error regarding an ambiguous conversion in multiple lines: inifcns_nstdsums.cpp:2604:18: error: ambiguous conversion for functional-style cast from 'int' to 'lst' (aka 'container<std::list>') return e * (-H(lst(0),1/arg).hold()); ^~~~~ ./container.h:145:2: note: candidate constructor container(STLT const & s, bool discardable = false) ^ ./container.h:167:11: note: candidate constructor explicit container(const ex & p1) All errors are in the above file inifcns_nstdsums.cpp and occur with the same cast: H(lst(0),1/arg).hold() >From the error message I can conclude that there are two constructors applyable - which one is the correct one? E.g. casting 0 to ex (changing the line to: H(lst( ex(0) ),1/arg).hold() solves the error - is this the correct cast or is the other cast the correct one? Best Stefan

2 1

Re: [GiNaC-list] GiNaC-list Digest, Vol 93, Issue 4
by Gunnar Birke 30 Sep '13

30 Sep '13

On 26.09.2013 12:00, ginac-list-request(a)ginac.de wrote: > Send GiNaC-list mailing list submissions to > ginac-list(a)ginac.de > > To subscribe or unsubscribe via the World Wide Web, visit > https://www.cebix.net/mailman/listinfo/ginac-list > or, via email, send a message with subject or body 'help' to > ginac-list-request(a)ginac.de > > You can reach the person managing the list at > ginac-list-owner(a)ginac.de > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of GiNaC-list digest..." > > > Today's Topics: > > 1. Re: user defined functions and constants (Vladimir V. Kisil) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Thu, 26 Sep 2013 09:34:40 +0100 > From: "Vladimir V. Kisil" <kisilv(a)maths.leeds.ac.uk> > To: GiNaC discussion list <ginac-list(a)ginac.de>, Gunnar Birke > <gunnar.birke(a)online.de> > Cc: "Vladimir V. Kisil" <kisilv(a)maths.leeds.ac.uk> > Subject: Re: [GiNaC-list] user defined functions and constants > Message-ID: <18176.1380184480(a)krein.leeds.ac.uk> > > Hi, > >>>>>> On Sat, 21 Sep 2013 20:07:12 +0200, Gunnar Birke <gunnar.birke(a)online.de> said: > GB> the map, then call subs() with it. Enable the same support for > GB> functions however seems very difficult, as functions can only > GB> defined at compile time with a macro. And using subs() seems a > GB> bit silly with functions: > > Unfortunately, I did not understand the difficulty which you were > talking about. May be you was mislead by name of GiNaC::function > class? For your example, you do not need o define f(x) and g(x,y) as > GiNaC::function. They may be happily defined as an GiNaC::ex containing > GiNaC::realsymbol x and y. I think, this will be sufficient for most of > your purposes. > > Best wishes, > Vladimir Well, my problem is that I don't understand how I turn an GiNaC::ex containing f(2) into an GiNaC::ex 2*3+2 in the case that f(x) is some GiNaC::ex 3*x + 2 where x is a symbol object. How do I tell GiNaC, that f(x) means 3*x+2 at runtime from the moment on the user defined f(x) in this manner. What I (think I) understood is that I can define functions with DECLARE_FUNCTION and REGISTER_FUNCTION at compile time and that I can have an expression GiNaC::ex f, which contains something like 3*x+2 where x is a symbol. So far so good, but how can I handle things like f(2)?

2 3

user defined functions and constants
by Gunnar Birke 26 Sep '13

26 Sep '13

Hello guys, I'm working on a CAS for school mathematics in C++ and want to use GiNaC as a backend. In a CAS, the user has the option to define constants and functions, for example def a = 5 def f(x) = 2 * x + 3 def g(x, y) = x + 2y and then use these definitions like this f(3) + a * 4 - g(4, 5) Implement this for constants is pretty straight forward in GiNaC with the subs() method and an exmap which contains all defined constants. For the example above, it would be enough to insert a symbol with the value a together with the expression 5 into the map, then call subs() with it. Enable the same support for functions however seems very difficult, as functions can only defined at compile time with a macro. And using subs() seems a bit silly with functions: GiNaC::ex ex(f(2)); ex.subs(f(2) == f(2)); //what? This problem seems pretty common to me and I really wondered why GiNaC has no options to archive this. Did I miss something?

2 1

Multiplication of Boolean variables
by tklam 02 Sep '13

02 Sep '13

Hi, I am implementing an arithmetic system for only Boolean variables based on GiNaC. May I know how I can modify the behaviour of symbolic multiplication such that: x*x == x (instead of x*x==x^2), and the multiplication of numeric remains unchanged (eg 3*4 == 12) ? Using subs(x^2 == x) works but I found it is slow if my expressions are huge and there are many symbols. Thank you very much. Best regards, argama

3 2

GiNaC Match Documentation
by Mark Round 15 Aug '13

15 Aug '13

I have been encountering problems with the GiNaC function match. Basically I can not understand the documentation. I wanted to detail my story in the hope that perhaps my problems are solved and in the future someone else can avoid my pitfalls. The documentation informs me that I need an associative array. I can not find how to construct such an array elsewhere in the documentation so I assume that I am supposed to use exhashmap. That was a point of confusion for me. Then I designed the following code snipet, symbol x("x"), y("y"); ex tmp=x+y; exhashmap <ex> array; tmp.match(wild(1)+y); which apparently works fine. Now I wish to fill 'array' with the replacement table. So I modified to, tmp.match(wild(1)+y,array); which returned the following compiler messages, momentum_integration.cpp: In function ‘GiNaC::ex bring_to_parameterisable_form(GiNaC::ex)’: momentum_integration.cpp:13:33: error: no matching function for call to ‘GiNaC::ex::match(const GiNaC::ex, GiNaC::exhashmap<GiNaC::ex>&)’ momentum_integration.cpp:13:33: note: candidates are: In file included from /usr/include/ginac/ginac.h:30:0, from momentum_integration.h:14, from momentum_integration.cpp:1: /usr/include/ginac/ex.h:150:7: note: bool GiNaC::ex::match(const GiNaC::ex&) const /usr/include/ginac/ex.h:150:7: note: candidate expects 1 argument, 2 provided /usr/include/ginac/ex.h:151:7: note: bool GiNaC::ex::match(const GiNaC::ex&, GiNaC::exmap&) const /usr/include/ginac/ex.h:151:7: note: no known conversion for argument 2 from ‘GiNaC::exhashmap<GiNaC::ex>’ to ‘GiNaC::exmap& {aka std::map<GiNaC::ex, GiNaC::ex, GiNaC::ex_is_less>&}’ make: *** [momentum_integration.o] Error 1 I attempted to use the command exmap but I can not find anything on that in the documentation. I tried several permutations around, exmap <ex,ex> array; which made the compiler return, momentum_integration.cpp: In function ‘GiNaC::ex bring_to_parameterisable_form(GiNaC::ex)’: momentum_integration.cpp:11:8: error: ‘GiNaC::exmap {aka std::map<GiNaC::ex, GiNaC::ex, GiNaC::ex_is_less>}’ is not a template make: *** [momentum_integration.o] Error 1 At which point it is unclear to me what else I should try. The code was ran on a 64-bit Linux Mint-Debian distribution with ginac-1.6.2.

2 1

ginac 1.6.2-2 on mingw-32.
by testhandle 07 Aug '13

07 Aug '13

(most of this text was written using google translate. english is not good at me. so there's a possibility that the meaning is not transmitted.) there may be unnecessary operation and defects. however, it is forced by this operation, but it have confirmed the operation of the tutorial examples. * step 0, dependencies. gmp 5.0.1 cln (gmp) 1.3.3 gcc 4.7.2 ld 2.23.1 m4 1.4.16 autoconf 2.68 autoreconf 2.68 automake 1.11 libtool 2.4 python 2.7.3 bison 2.4.2 flex 2.5.35 * step 1. $ git clone git://www.ginac.de/ginac.git * step 2, modify the 'ginac/config/config.rpath'. line 67: replace mingw* | pw32* | os2*) -> pw32* | os2*) line 57: insert mingw*) wl='Wl,' ;; line 135: replace cygwin* | mingw* | pw32*) -> cygwin* | pw32*) line 185: replace cygwin* | mingw* | pw32*) -> cygwin* | pw32*) line 199: replace linux*) -> linux* | mingw*) * step 3. $ cd ginac $ autoreconf -i * step 4, modify the 'ginac/configure'. line 15539: delete * step 5, modify the 'ginsh/ginsh_lexer.lpp'. line 37: replace #include "ginsh_parser.hpp" -> #include "ginsh_parser.h" * step 6. $ export CLN_LIBS="-L/usr/local/lib -lcln -lgmp" $ export CLN_CFLAGS="-I/usr/local/include" $ ./configure $ make $ make install * test 1. // hello.cpp ---------------- #include <iostream> #include <ginac/ginac.h> using namespace std; using namespace GiNaC; int main() { symbol x("x"), y("y"); ex poly; for (int i=0; i<3; ++i) poly += factorial(i+16)*pow(x,i)*pow(y,2-i); cout << poly << endl; return 0; } $ g++ -c -O3 -I/usr/local/include -L/usr/local/bin hello.cpp $ g++ -o hello hello.o -lginac -lcln -lgmp &> result.txt $ ./hello 355687428096000*x*y+20922789888000*y^2+6402373705728000*x^2 * test 2. // hello2.cpp ---------------- #include <iostream> #include <ginac/ginac.h> 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() { symbol z("z"); for (int i=0; i<6; ++i) cout << "H_" << i << "(z) == " << HermitePoly(z,i) << endl; return 0; } $ g++ -c -O3 -I/usr/local/include -L/usr/local/bin hello2.cpp $ g++ -o hello2 hello2.o -lginac -lcln -lgmp &> result.txt $ ./hello2 H_0(z) == 1 H_1(z) == 2*z H_2(z) == -2+4*z^2 H_3(z) == -12*z+8*z^3 H_4(z) == 12+16*z^4-48*z^2 H_5(z) == 120*z+32*z^5-160*z^3

1 0

Numeric and Constants
by David Narvaez 13 Jul '13

13 Jul '13

Hi, While implementing a GiNaC based backend for my software I decided to replace doubles by GiNaC::exs. That seemed to work fine until I actually needed the final native value of the expression, so I changed my mind to use GiNaC::numerics instead, which works fine except for constants, since I'm unable to do things like GiNaC::numeric angle = 2 * GiNaC::Pi; is there a quick way to pass from a constant to a numeric? (or alternatively a way to turn expressions into doubles and I should go back to GiNaC::ex). Thanks in advance. David E. Narvaez

1 0