- GiNaC-list - ginac.de

compilation problem
by chrisd＠catv7250.extern.kun.nl 03 Sep '01

03 Sep '01

Hello everybody, I have just compiled the new GiNaC at my place. This took a bit more time than I expected because of the following compiler error c++ -DHAVE_CONFIG_H -I. -I. -I.. -I/usr/local/include -g -O2 -c numeric.cpp -fPIC -DPIC -o .libs/numeric.lo /usr/include/g++-v3/bits/stl_construct.h: In function `void std::_Construct(_T1*, const _T2&) [with _T1 = cln::cl_RA, _T2 = cln::cl_RA]': numeric.cpp:128: instantiated from here /usr/include/g++-v3/bits/stl_construct.h:78: no matching function for call to `cln::cl_RA::operator new(unsigned int, void*&)' /usr/local/include/cln/rational_class.h:35: candidates are: static void* cln::cl_RA::operator new(unsigned int) /usr/local/include/cln/rational_class.h:36: static void* cln::cl_RA::operator new(unsigned int, cln::cl_RA*) make[2]: *** [numeric.lo] Error 1 make[2]: Leaving directory `/home/chrisd/download/GiNaC-0.9.3/ginac' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/home/chrisd/download/GiNaC-0.9.3' make: *** [all-recursive-am] Error 2 So, the file numeric.cpp does not compile. I also found the source of the error. If I comment out all calls to vector<cln::cl_RA>::push_back(cln::cl_RA) the error messages disappear. All calls to this function happen inside the function const numeric bernoulli(const numeric &nn) and there are two of these calls. I use gcc 3.0.1 and cln 1.1. Hopefully maintainers find this information usefull to correct the error or come up with a work around. Greetings, Chris Dams

2 1

None
by Richard B. Kreckel 16 Aug '01

16 Aug '01

We are psyched to announce GiNaC 0.9.3. This release is even more fat-, meta-tag-, cholesterol- and bug-free than earlier versions: * Series expansion is now much more consistent for very small order expansion, giving some users unprecedented expressiveness and * lsolve() accepts an algorithmic hint as parameter so one can squeeze out ultimate performance. Get your copy as long as our supply of this groundbreaking product lasts!

1 0

ginsh compilation with gcc2.96 -- readline 4.2
by Vladimir Dyadichev 12 Aug '01

12 Aug '01

Hello I would like to compile GiNaC on my Linux system with gcc 2.96 . I have libreadline v 4.2 installed on my system but it seems that g++ do not like readline header files. Consequently the ginsh fails to compile (library itself compiles OK). Any suggestions ??? Thanks in advance. Vladimir Here is part of configure output: checking for readline/readline.h... yes checking for readline/history.h... yes checking for tgetent... no checking for tgetent in -lncurses... yes checking for readline in -lreadline... yes checking for version of libreadline... unknown The part of config.log corresponding to that failure configure:5993: c++ -o conftest -g -O0 conftest.C -lreadline -lncurses 1>&5 configure:6022: checking for version of libreadline configure:6050: c++ -o conftest -g -O0 conftest.C -lreadline -lncurses 1>&5 In file included from /usr/include/readline/keymaps.h:37, from /usr/include/readline/readline.h:36, from configure:6038: /usr/include/readline/rltypedefs.h:48: `rl_command_func_t' declared with an exception specification/usr/include/readline/rltypedefs.h:51: `rl_compentry_func_t' declared with an exception specificatio /usr/include/readline/rltypedefs.h:52: `rl_completion_func_t' declared with an exception specificatin /usr/include/readline/rltypedefs.h:54: `rl_quote_func_t' declared with an exception specification /usr/include/readline/rltypedefs.h:55: `rl_dequote_func_t' declared with an exception specification/usr/include/readline/rltypedefs.h:57: `rl_compignore_func_t' declared with an exception specificatin /usr/include/readline/rltypedefs.h:59: `rl_compdisp_func_t' declared with an exception specificationusr/include/readline/rltypedefs.h:62: `rl_hook_func_t' declared with an exception specification /usr/include/readline/rltypedefs.h:65: `rl_getc_func_t' declared with an exception specification /usr/include/readline/rltypedefs.h:70: `rl_linebuf_func_t' declared with an exception specification/usr/include/readline/rltypedefs.h:73: `rl_intfunc_t' declared with an exception specification /usr/include/readline/rltypedefs.h:75: `rl_icpfunc_t' declared with an exception specification /usr/include/readline/rltypedefs.h:76: `rl_icppfunc_t' declared with an exception specification /usr/include/readline/rltypedefs.h:78: `rl_voidfunc_t' declared with an exception specification /usr/include/readline/rltypedefs.h:79: `rl_vintfunc_t' declared with an exception specification /usr/include/readline/rltypedefs.h:80: `rl_vcpfunc_t' declared with an exception specification /usr/include/readline/rltypedefs.h:81: `rl_vcppfunc_t' declared with an exception specification In file included from /usr/include/readline/readline.h:37, from configure:6038: /usr/include/readline/tilde.h:55: `tilde_hook_func_t' declared with an exception specification configure: failed program was: #line 6031 "configure" #include "confdefs.h" #ifdef __cplusplus extern "C" void exit(int); #endif #include <stdio.h> #include <sys/types.h> #include <readline/readline.h> main() { FILE *fd; fd = fopen("conftest.out", "w"); fprintf(fd, "%s\n", rl_library_version); fclose(fd); return 0; } The compilation of ginsh fails with exactly the same error messages: make[2]: Вход в каталог `/home/vlad/src/GiNaC-0.9.2/ginsh' sed -n -f ./ginsh_fcn_help.sed <ginsh.1 >ginsh_fcn_help.c sed -n -f ./ginsh_op_help.sed <ginsh.1 >ginsh_op_help.c c++ -DHAVE_CONFIG_H -I. -I. -I.. -I./../ginac -I../ginac -DIN_GINAC -g -O2 -c ginsh_parser.cc In file included from /usr/include/readline/keymaps.h:37, from /usr/include/readline/readline.h:36, from ginsh.h:36, from ginsh_parser.yy:41: /usr/include/readline/rltypedefs.h:48: `rl_command_func_t' declared with an exception specification/usr/include/readline/rltypedefs.h:51: `rl_compentry_func_t' declared with an exception specification <etc>

3 2

GiNaC memory leak?
by Marko Riedel 08 Aug '01

08 Aug '01

Greetings. I am sending some code that I wrote in order to solve a differential equation. You run the compiled program with a command-line argument telling it what terms to use. Interesting values are e.g. 10, 14, 16 and 19. (The higher values take some time to compute.) Warning! When you run the program on the higher values, it starts consuming huge amounts of memory. On a SuSE Linux machine the OS eventually kills the process. What is going on here? The program solves a system of equations with a 1000 x 1000 approx. coefficient matrix. What is all the memory being used for? Take, say 32 bytes per entry; the system should then allocate about 32M of memory. In fact it tries to allocate more than 512M of memory! Best regards, Marko Riedel ---------------------------------------------------------------------- #include <stdio.h> #include <fstream> #include <ginac/ginac.h> using namespace std; using namespace GiNaC; static symbol u("u"); static symbol z("z"); static symbol l1("l1"); static symbol l2("l2"); static symbol phi("phi"); static void extracteqs(ex eq, lst *l) { int i, j; ex c1, c2; for(i=eq.ldegree(z); i<=eq.degree(z); i++){ c1=eq.coeff(z, i); for(j=eq.ldegree(u); j<=eq.degree(u); j++){ c2=c1.coeff(u, j); if(is_ex_of_type(c2, numeric)){ if(c2!=0){ cerr << "inconsistency: " << i << " " << j << endl; exit(-1); } } else{ (*l).append(c2==0); cerr << (*l).nops() << endl; } } } } int main(int argc, char **argv) { int maxcoeff, i, j; if(argc!=2){ cout << argv[0] << " <maxcoeff>" << endl; exit(-1); } sscanf(argv[1], "%d", &maxcoeff); if(maxcoeff<1){ cout << "argument " << maxcoeff << " out of range" << endl; exit(-2); } ex rt, lf1, lf2; lst vs; rt=0; lf1=0; lf2=0; for(i=0; i<maxcoeff; i++){ for(j=0; j<maxcoeff; j++){ symbol a; rt+=a*pow(z, i)*pow(u, j); vs.append(a); if(i<maxcoeff-6 && j<maxcoeff-6){ symbol b, c; lf1+=b*pow(z, i)*pow(u, j); lf2+=c*pow(z, i)*pow(u, j); vs.append(b); vs.append(c); } } } cerr << rt << endl; cerr << lf1 << endl; cerr << lf2 << endl; cerr << vs << endl; ex s, sf, sf1, sf2; sf=pow(1-u, 11)*pow(1-z, 7)*pow(1-z*u, 7); sf1=pow(1-u, 9)*pow(1-z, 5)*pow(1-z*u, 5); sf2=pow(1-u, 10)*pow(1-z, 2)*pow(1-z*u, 2); s=rt/sf1+ lf1*u*log(1/(1-z))/sf2+ lf2*log(1/(1-z*u))/sf2; ex f, rz, rzv, rzu, rzvu; f=(u-u*u*z*z)/pow(1-z, 2)/pow(1-z*u, 2); rz=1/pow(1-z, 2); rzv=1/pow(1-z, 2); rzu=rz.subs(z==z*u); rzvu=rzv.subs(z==z*u); ex t1, t2, t3, t4, t5, t6, t7, t8, t9; t1=6*u*u*rzu *f; t2=6*u*u*rzvu*f; t3=6*u*u*rzu *phi; t4=6*u*u*rzu *rz; t5=6*u*u*rzvu*rz; t6=6*u*u*rzu *rzv; t7=6* rz *f; t8=6* rzv *f; t9=6* rz *phi; ex rhs1, rhs2; rhs1=(t1+t2+t4+t5+t6+t7+t8).normal(); rhs2=(t3+t9).normal(); ex eq, eq1; eq=(s.diff(z, 2)-rhs1-rhs2.subs(phi==s)) .subs(lst(log(1/(1-z))==l1, log(1/(1-z*u))==l2)) .expand(); eq1=0; for(i=0; i<eq.nops(); i++){ eq1+=(eq.op(i)*sf).normal().expand(); cerr << "expand: " << i << " " << eq.nops() << endl; } ex leq1=eq1.coeff(l1, 1), leq2=eq1.coeff(l2, 1); eq1=eq1.subs(lst(l1==0, l2==0)); lst ceqs; ex s0=(s.subs(z==0)*pow(1-u, 9)).normal(); ex sz0=(s.diff(z).subs(z==0)*pow(1-u, 9)).normal(); ex i0=(u*pow(1-u, 9)).expand(); ex iz0=((4*u+4*u*u)*pow(1-u, 9)).expand(); cerr << "extract initial: 1" << endl; extracteqs(s0-i0, &ceqs); cerr << "extract initial: 2" << endl; extracteqs(sz0-iz0, &ceqs); cerr << "extract: 1" << endl; extracteqs(eq1, &ceqs); cerr << "eqs: " << ceqs.nops() << endl; cerr << "extract: 2" << endl; extracteqs(leq1, &ceqs); cerr << "eqs: " << ceqs.nops() << endl; cerr << "extract: 3" << endl; extracteqs(leq2, &ceqs); cerr << "eqs: " << ceqs.nops() << endl; ex sl; cerr << "solving: " << ceqs.nops() << " " << vs.nops() << " ..." << endl; try{ sl=lsolve(ceqs, vs); } catch(exception &e){ cerr << e.what() << endl; exit(-3); } cout << sl << endl; lst zlist; for(i=0; i<vs.nops(); i++){ zlist.append(vs.op(i)==0); } ex result; result=s.subs(sl).subs(zlist); cout << result << endl; archive a; a.archive_ex(result, "result"); char fname[80]; sprintf(fname, "fixedpoints%d.gar", maxcoeff); ofstream out(fname); out << a; out.close(); }

2 2

GiNaC 0.9.2 around
by Alexander Frink 31 Jul '01

31 Jul '01

Hi all, I apologize for the laziness of my co-workers who produced no more than 3 entries in the news section for GiNaC 0.9.2. But ok, it is summer time and very hot in Germany, so we hope to be more productive again in the future... Forced to write something before they let me go to the next beer-garden Alex * Epsilon tensor is more functional. * simplify_indexed() is better at detecting expressions that vanish for symmetry reasons. * Several little bugfixes and consistency enhancements. -- Alexander Frink E-Mail: Alexander.Frink(a)Uni-Mainz.DE Institut fuer Physik Phone: +49-6131-3923391 Johannes-Gutenberg-Universitaet D-55099 Mainz, Germany

1 0

Is GiNaC really that dirty?
by Chris Dams 17 Jul '01

17 Jul '01

Hello everybody, I have been playing around with ginac a bit, but the impression that I get from the tutorial is that one has to write really dirty code to get something done. Since it is not my intention to start a flame war based on vague sentences like the previous one, I will give an example of a function I want to write. The answer I am hoping for is something like: ``No, you stupid, you don't need to write things like that. It can be done much more beautifull, namely like this: ......'' To be more concrete, I want to write the function ex checkdeg(ex f,ex var,int n) which expects that *f* is a polynomial in expanded form in a variable in *var* and perhaps other variables. The return value is the same polynomial except that all terms in which *var* occurs in a power higher than *n* are discarded. Maybe you want to compare the function I have written to accomplish this with the FORM statement id var^m?{>n}=0; which does the same thing. ``Comparing'' especially with respect to its length and elegance. I want to make it a *map_function*. Now I have a problem since *map_function*s only have one argument. The solution I came up with looks like struct Checkdeg:public map_function { ex var; int n; ex operator()(const ex&f,ex v,int nint) { var=v; n=nint; return (*this)(f); } ex operator()(const ex&f) { ... Here comes the code that actually does something ... } } checkdeg; which looks somewhat queer. I really would like to be able to say something like return f.map(*this)(var,n) or maybe like this return f.map((*this)(var,n)) Well, you get the idea. I want to map a function with more than just one argument. Now after having ``solved'' the problem of writing the *map_function*, now let us consider the the function ex operator()(const ex&f) This function first checks whether *f* is an addition and then uses the *map* function to tell ginac that it should treat each term seperately. No problem here anymore. Then we have the case were *f* is not an addition. I really would want to write someting like lst repls; return f.match(wild(0)*pow(var,wild(1)),repls) if(repls.op(1)>n) f.subs(var==0); but then I discover that *pow(var,m)* is not mathched since it is not a product. I really cannot imagine any reason whatsoever why not to automatically consider something that is not a product as a product with a single factor in patern matching against a product. It would make code so much more elegant, and cannot be really difficult or inefficient. The same thing with the power. The manual mentions something like ten times that one may not use the pattern *pow(var,wild(1))* to match the expression *var*. Why on earth not? The cynical remark would be that if something needs to be said ten times, it can hardly be something you want to hear. A solution to this problem would seem to be using the *.has* method, but then again, this one does not return the thing that matches the wildcard. So I have to use the *.find* method. Then I have to interpret the list it returns. This I have to do by writing an *if* like if(repls.op(0).op(1)>n) to check the first element in the list, which contains an expression like *var^m* and get the *m* out of this. This is may be only slightly inelegant, but I think it is inelegant. Why does the *.find* method return the entire match expression, why not simply the thing that is equal to the wildcard. It would change my *if* into if(repls.op(0)>n) which is somewhat shorter (but in complicated expressions could be very much sorter). In my opinion the user can generally be expected to know the expression againts which he matches, so why does he have to dissemine that one again when he reads the result from the match? It simply does not give any new information. Furthermore it is inconsistent since the match method indeed only returns the values of the wildcards. The only objection I could imagine is that the *find* function could return more than one match. This is true, but in this case one could return {{value of $0,value of $1},{value of $0,value of $1}} Indeed, now I would have to write down exactly the same test, but the advantage is that I don't need to know that the exponent of a power is stored in *.op(1)*, which I really don't want to and shouldn't need to know as an ordinary user. After these considerations I felt I had to write something like ex operator()(const ex&f) { if(is_a<add>(f)) return f.map((*this)(var,n)); switch(n) { case 0: return f.subs(var==0); case 1: return f.subs(pow(var,wild())==0); default: lst repls; if(f.find(pow(var,wild()),repls)) { if(repls.op(0).op(1)>n) return f.subs(var==0); else return f; } else return f; } } for my function. Now I guess the whole *struct Checkdeg* is at least thrice as long as the thing I saw before me in my dreams. Does it really have to be that dirty? Greetings, Chris Dams

3 4

configure cannot find headers for libreadline
by Chris Dams 12 Jul '01

12 Jul '01

Hello everybody, Configure cannot find the headers for libreadline, although they are definately present. I tried setting the CPPFLAGS as in setenv CPPFLAGS -I/usr/locacal/gnu/include ./configure --prefix=/home/thef/chrisd/c++ but still configure does not find them. Does somebody know how to solve this problem? Note: I am on a unix where I am not allowd to set symlinks in system include dirs. (If this were allowed, I would probably not be posting anything here). Greetings, Chris Dams

2 3

error while loading shared libraries: libcln.so.2
by Wolfgang 06 Jul '01

06 Jul '01

I've done a standard installation (no flags, no special directories) of cln 1.1.1 with gcc 2.96. After 'make check' it said 'All tests passed'. Now I can compile a cln example file, but when I want to run the binary I get the above error message. Do I have to set certain gcc flags when compiling the example file, or how can solve that problem? Thank you for your help! Wolfgang - To UNSUBSCRIBE, email to ginac-list(a)ginac.de with a subject of "unsubscribe".

4 3

Polynomials
by Wolfgang Abele 04 Jul '01

04 Jul '01

How can I set the coefficient domain of polynomials? Say, I want my coefficients to be in Z mod 17, i.e. Galois Field 17, what do I do? Also, as a more general question, I'd like to know if GiNac poses any overhead on the polynomial classes provided by CLN. In other words, it is less efficient to write polynomials routines in Normal.cpp (the polynomial module in GiNac) than directly work with the CLN classes? Thanks in advance for your answers! Wolfgang - To UNSUBSCRIBE, email to ginac-list(a)ginac.de with a subject of "unsubscribe".

2 2

Ginsh installation problem
by Wolfgang Abele 03 Jul '01

03 Jul '01

> Another solution would be to set the LD_LIBRARY_PATH environment > variable to include the directory where CLN library is located. This > would allow you to move CLN easier. Thank you, Ben, for your quick answer! I set that variable, and my cln program runs fine now Here's my next problem: I cannot install Ginac: After I enter 'make', gcc 2.96 outputs numerous warnings such as /usr/local/include/readline/rltypedefs.h:73: `rl_intfunc_t` declared with an exception specification and eventually stops with error messages; ginsh_parser.yy:769: 'filename_completion_function' undeclared ginsh_parser.yy:769: 'completion_matches' undeclared ginsh_parser.yy:824: 'cannot convert char **(+) () 'to char **(*) in (const char *, int, int) assignment I have readline 4.2. Wolfgang - To UNSUBSCRIBE, email to ginac-list(a)ginac.de with a subject of "unsubscribe".

1 0