- GiNaC-list - ginac.de

custom print function for GiNaC::add and GiNaC::mul for deterministic printing
by Diddens, Christian (UT-TNW) 12 Oct '21

12 Oct '21

Dear GiNaC guys, first of all: Thanks once more for your powerful library! I have written a neat C code generation for the just-in-time compilation of arbitrary multiphysics finite element equations. The equations are defined in python, can be coupled and combined, and are finally monolithically solved via the library oomph-lib. While code generation is fast, the compilation of the generated code takes some time (~10 seconds for complicated couplings of Navier- Stokes, advection-diffusion, Marangoni flow, mesh node positions for ALE methods, etc.) Now, if the code generation was deterministic, I could check whether the same code has already been compiled in a previous run and skip the compilation in that case. I could also use the tool "ccache" for this. As proposed in the manual, I therefore wanted to write my own print implementation and make sure that in particular "add" and "mul" are written in an e.g. lexiographical order. However, it is a bit problematic since the properties "epvector seq" and "ex overall_coeff" from the class "expairseq" are protected without any public getter functions. Of course, I could iterate over the "::op(index)", but this would convert each "expair" first in an "ex" and I think the "expair" together with the "overall_coeff" is definitely the better way to go. So is there any neat way to access there properties from outside of the class? I understand your decision for using hashes and do not focus on a determinstic order, but in my case, it would definitely help to same some redundant JIT compilation time. Thanks for any help, you are doing a perfect job with GiNaC. Best regards, Christian

2 6

GiNaC 1.8.1 released
by Richard B. Kreckel 09 Aug '21

09 Aug '21

Hi, GiNaC 1.8.1 has just been tagged. Here's what's new: * Add method relational::canonical() and improve conversion of relational to Boolean (it now works on many simple symbolic cases). * Improve normalization of negative exponents. * Fix indexing multiply referenced objects with ex::operator[]. * Make functions evalf() their arguments before doing own evalf(). * Fix bugs in H_evalf() and in evaluation of iterated integrals. * Several portability improvements and compiler warning fixes. If you aren't vaccinated, consider getting, and keep hacking! -richy. -- Richard B. Kreckel <https://in.terlu.de/~kreckel/>

1 0

Cross-compiling on Linux for Windows via x86_64-w64-mingw32-g++
by c.diddens＠utwente.nl 28 Jul '21

28 Jul '21

Dear all, first of all, I want to thank you all for this very nice piece of software. I am currently developing a Python package to wrap the finite element library oomph-lib in Python. GiNaC is required here to generate C code which is compiled on the fly to quickly assemble the weak forms which are defined in Python. In some sense, this is similar to the finite element library FEniCS, which also uses GiNaC, but oomph-lib provides better capabilities for multi-physics and ALE methods. My Python code works perfect in Linux, but as a long time goal, I also want to produce Python Wheels for Windows. However, I have a hard time to cross-compile GiNaC and CLN for Windows (64 bit). Does anyone has some experience with that? I am aware of the precompiled Windows binary, but they seem to be a bit out of date... I am grateful for any input. Best regards, Christian

3 3

Question
by Aimad Chabounia 22 Jul '21

22 Jul '21

Hello, I am currently doing an internship and we used GiNaC in our project to simplify expressions. my question is how I can generate a balanced expression tree from a ginac expression ?? (now I create expressions in form of a binary expression tree) Thanks and regards, Aimad

2 2

Parse error involving factorial symbol
by Roberto Bagnara 15 May '21

15 May '21

Hi there. I am reviving some 10-years old software that is based on GiNaC. When I started the work I was pessimistic, given so much time has passed. But now the thing compiles again. The problem is that I cannot parse the test-suite that was used for validation. I get: on line 392: parse error: GiNaC: parse error at line 0, column 0: expected ')' or ',' in argument list, got: "!" [GiNaC::ex GiNaC::parser::parse_identifier_expr()(parser/parser.cpp:88)] I guess something has changed in the parser, or maybe my reviving effort introduced a bug. Line 392 is the last one in this block: y (n+3)! factorial(3+n) factorial(3+n) factorial(n)*(3+n)*(1+n)*(2+n) factorial(3+n) factorial(3+n) This is the first block that contains an "!". This and all the subsequent blocks of tests that contain an "!" symbol fail in the same way. None of the blocks before fails. Any hint? Thanks, Roberto -- Prof. Roberto Bagnara Applied Formal Methods Laboratory Department of Mathematical, Physical and Computer Sciences University of Parma, Italy http://www.cs.unipr.it/~bagnara/ mailto:bagnara@cs.unipr.it

3 5

Re: [GiNaC-list] Parse error involving factorial symbol
by Vladimir V. Kisil 15 May '21

15 May '21

Dear Roberto, Yes, normalisation methods in GiNaC undergone some modification over that period as well. Thus particular form of the expression can be different now. Best wishes, Vladimir -- Vladimir V. Kisil http://www.maths.leeds.ac.uk/~kisilv/ Book: Geometry of Mobius Maps https://doi.org/10.1142/p835 Soft: Geometry of cycles http://moebinv.sourceforge.net/ Jupyter notebooks: https://github.com/vvkisil/MoebInv-notebooks >>>>> On Fri, 14 May 2021 22:36:13 +0200, Roberto Bagnara <bagnara(a)cs.unipr.it> said: RB> On 5/14/21 3:33 PM, Vladimir V. Kisil wrote: >>>>>>> On Fri, 14 May 2021 15:08:09 +0200, Roberto Bagnara >>>>>>> <bagnara(a)cs.unipr.it> said: RB> On 5/14/21 9:52 AM, Vladimir V. Kisil wrote: >> >> Dear Roberto! It seems that Ginsh and parser in GiNaC are >> >> implemented differently. Ginsh understands postfix factorial >> >> notation like "3!" but GiNaC parser is not. GiNaC parser is >> >> still happy with "factorial(3)". Best wishes, Vladimir >> >> RB> Thanks Vladimir! But please help me understand: I did not RB> change anything in the part of the code invoking the GiNaC RB> parser, and I did not change the tests. So the situation you RB> are describing, i.e., GiNaC parser not understanding postfix RB> factorial notation, is something that changed from, say, 10 RB> years ago. In other words, do you agree that, say, 10 years RB> ago, the GiNaC parser was accepting that notation? >> It seems that before 2008-08-21 GiNaC and Ginsh had used the same >> parser, which Ginsh is using till now. After a patch they >> diverged in this respect. So it is quite well possible that your >> code was running with the old version of GiNaC but cannot do this >> now without some alteration. RB> Dear Vladimir, RB> This explains everything. Now I am investigating failures in RB> the regression test-suite. One of the most promising examples RB> for my diagnosis efforts is this: RB> simplification for output of RB> -1+1/12*sqrt(sqrt(3)*sqrt(12))*sqrt(12)+1/12*sqrt(-sqrt(3)*sqrt(12))*sqrt(12) RB> was expected to be -1+(1/2+1/2*I)*sqrt(2) but resulted in RB> -1+(1/2+1/2*I)*4^(1/4) RB> From what you write I gather that back in 2008, or even before, RB> we obtained sqrt(2) where we now obtain 4^(1/4). RB> Thanks, RB> Roberto RB> -- Prof. Roberto Bagnara Applied Formal Methods Laboratory RB> Department of Mathematical, Physical and Computer Sciences RB> University of Parma, Italy http://www.cs.unipr.it/~bagnara/ RB> mailto:bagnara@cs.unipr.it

1 0

to_long() of an imaginary unit is 9
by Vitaly Magerya 14 Apr '21

14 Apr '21

Hi, all. It seems that numeric conversion routines like to_long() don't check for imaginary numbers, and happily return random results. Here's an example program: #include <ginac/ginac.h> #include <ginac/parser.h> #include <fstream> #include <sstream> using namespace std; using namespace GiNaC; int main() { parser reader; istringstream ifs("I"); ex expr = reader(ifs); if (is_a<numeric>(expr)) { numeric x = ex_to<numeric>(expr); cout << "Numeric: " << x << "\n"; cout << ".real(): " << x.real() << "\n"; cout << ".imag(): " << x.imag() << "\n"; cout << ".numer(): " << x.numer() << "\n"; cout << ".denom(): " << x.denom() << "\n"; cout << ".numer().to_long(): " << x.numer().to_long() << "\n"; cout << ".denom().to_long(): " << x.denom().to_long() << "\n"; cout << ".to_long(): " << x.to_long() << "\n"; } else { cout << "Not numeric: " << expr << "\n"; } } The result of it with GiNaC 1.7.8 (CLN 1.3.6) is: $ c++ -o example example.cpp -lginac -lcln && ./example Numeric: I .real(): 0 .imag(): 1 .numer(): I .denom(): 1 .numer().to_long(): 9 .denom().to_long(): 1 .to_long(): 9 Seeing that the documentation contains this phrase: Both to_int()/to_long() and to_double() discard the imaginary part of complex numbers. ... I'm guessing this is not the expected behavior. * * * As a related question, is there any way to prevent parser from interpreting "I" as an imaginary unit? I'd like it to be just a variable.

2 1

Collect() with a list of variables doesn't work in ginsh
by Vitaly Magerya 24 Mar '21

24 Mar '21

Hi, all. In section 5.7.2 of the docs (Expanding and collecting) there is this example of ginsh usage: > a=expand((sin(x)+sin(y))*(1+p+q)*(1+d)); > collect(a,{p,q}); The docs say I should get this: d*sin(x)+(d*sin(x)+sin(y)+d*sin(y)+sin(x))*p ... but actually: cannot modify multiply referenced object This is with GiNaC V1.7.6, V1.7.8, and the latest commit of V1.8.0. How do I make this work?

3 7

Differentiate matrix wrt vector
by Orxan Shibliyev 04 Mar '21

04 Mar '21

Suppose, I have a 5x5 matrix and I want to differentiate the matrix wrt to a 5x1 matrix in order to obtain a 3D matrix or tensor. In the following code, A.diff(uL); cannot compile. Is it possible to differentiate a matrix with symbolic elements wrt to a vector which also contains symbols? #include <iostream> #include <ginac/ginac.h> using namespace std; using namespace GiNaC; int main() { const int DIM = 5; symbol g("g"); symbol uL1("uL1"); symbol uL2("uL2"); symbol uL3("uL3"); symbol uL4("uL4"); symbol uL5("uL5"); matrix uL = { {uL1}, {uL2}, {uL3}, {uL4}, {uL5} }; auto qL = matrix(DIM, 1); qL(0,0) = sqrt(uL1); qL(1,0) = uL2 / qL(0,0); qL(2,0) = uL3 / qL(0,0); qL(3,0) = uL4 / qL(0,0); qL(4,0) = (uL5 + (g - 1) * (uL5 - 0.5 * (pow(uL2,2) + pow(uL3,2) + pow(uL4,2)) / uL1)) / uL1; symbol uR1("uR1"); symbol uR2("uR2"); symbol uR3("uR3"); symbol uR4("uR4"); symbol uR5("uR5"); matrix uR = { {uR1}, {uR2}, {uR3}, {uR4}, {uR5} }; auto qR = matrix(DIM, 1); qR(0,0) = sqrt(uR1); qR(1,0) = uR2 / qR(0,0); qR(2,0) = uR3 / qR(0,0); qR(3,0) = uR4 / qR(0,0); qR(4,0) = (uR5 + (g - 1) * (uR5 - 0.5 * (pow(uR2,2) + pow(uR3,2) + pow(uR4,2)) / uR1)) / uR1; auto q = qL.add(qR); q.mul_scalar(0.5); matrix B = { {2*q(0,0), 0, 0, 0, 0}, {q(1,0), q(0,0), 0, 0, 0}, {q(1,0), q(0,0), 0, 0, 0}, {q(2,0), 0, q(0,0), 0, 0}, {q(3,0), 0, 0, q(0,0), 0}, {q(4,0)/g, (g-1)*q(1,0)/g, (g-1)*q(2,0)/g, (g-1)*q(3,0)/g, q(0,0)/g} }; matrix C = { {q(1,0), q(0,0), 0, 0, 0}, {(g-1)*q(4,0)/g, (g+1)*q(1,0)/g, (1-g)*q(2,0)/g, (1-g)*q(3,0)/g, (g-1)*q(0,0)/g}, {0, q(2,0), q(1,0), 0, 0}, {0, q(3,0), 0, q(1,0), 0}, {0, q(4,0), 0, 0, q(1,0)} }; auto A = B.mul(C.inverse()); A.diff(uL); return 0; }

2 1

Some problem Installing GiNaC on Linux
by 이재욱 16 Jan '21

16 Jan '21

Hello GiNac-list, I have some problems installing the GiNaC on Linux. May I ask you for advice? I followed the tutorial PDF to install, and the linux computer showed a message that "See check/test-suite.log Please report to ginac-list(a)ginac.de ". So, I sent an email to get some advice. I have the below prerequisite software version. $ g++ --version : 4.8.5 $ git --version : 1.8.3.1 $ autoconf --version : 2.69 $ libtool --version : 2.4.6 $ automake --version : 1.13.4 These files are in the '/usr/bin/' and I will install the GiNaC in the '/home/{USERNAME}/Package/' (I think these points make some problems.) First, I install the CLN before installing the GiNaC. $ ./configure --prefix=/home/{USERNAME}/Package $ make $ make check (--> all Pass) $ make install And then, I try to install the GiNaC $ export PKG_CONFIG_PATH=/home/{USERNAME}/Package/lib/pkgconfig $ ./configure --prefix=/home/{USERNAME}/Package $ make $ make check (--> there are 7 fail, and 51 PASS; it shows me the above message) Related log file(test-suite.log) is attached, I do not know what is wrong. Please let me help to install GiNaC well Thanks, ljw.

2 1