February 2023 - GiNaC-list

feature request
by Marko Riedel 22 Feb '23

22 Feb '23

Greetings to all. Continuing with my GiNaC project I have noticed that there was some functionality missing that other software offers. In particular, I require **factorization of integers** and a function to compute the **set of divisors** of an integer. I have implemented these using basic math such as might be used in a freshman course. It passes all my tests, but it really needs an implementation from a professional mathematician. Maybe you can file this with other feature requests you have received. An excerpt from my code follows below. I hope I have understood lists correctly that they are doubly linked and a list append of one element runs in constant time. Best regards, Marko lst factorizeposint(numeric n) { lst result = {}, ent; int pot = 0; while(irem(n, numeric(2)).is_zero()){ pot++; n = iquo(n, numeric(2)); } if(pot > 0){ ent = {numeric(2), pot}; result.append(ent); } numeric q = 3; while(q*q <= n){ pot = 0; while(irem(n, q).is_zero()){ pot++; n = iquo(n, q); } if(pot > 0){ ent = {q, pot}; result.append(ent); } q += 2; } if(n > 1){ ent = {n, numeric(1)}; result.append(ent); } return result; } numeric totient(numeric n) { if(n==1) return 1; lst pfact = factorizeposint(n); numeric result = n; for(lst::const_iterator it = pfact.begin(); it != pfact.end(); ++it){ lst ent = ex_to<lst>(*it); result *= numeric(1) -numeric(1)/ex_to<numeric>(ent.op(0)); } return result; } lst divisors(numeric n) { if (n==1) return {1}; lst pfact = factorizeposint(n), ent; numeric tau(1); lst::const_iterator it; for(it = pfact.begin(); it != pfact.end(); ++it){ ent = ex_to<lst>(*it); tau *= numeric(1)+ex_to<numeric>(ent.op(1)); } exvector res {}; res.resize(tau.to_int(), numeric(1)); int sofar = 1; int pos; for(it = pfact.begin(); it != pfact.end(); ++it){ pos = sofar; ent = ex_to<lst>(*it); numeric p = ex_to<numeric>(ent.op(0)); numeric v = ex_to<numeric>(ent.op(1)); for(numeric vv(1); vv <= v; vv++){ for(int q = 0; q < sofar; q++){ res[pos++] = res[q]*pow(p,vv); } } sofar = pos; } lst result = {}; for(pos=0; pos<sofar; pos++) result.append(res[pos]); return result; }

2 2

Adding custom functions to the parser
by Vitaly Magerya 20 Feb '23

20 Feb '23

Hi, all. I'm trying to add custom functions to GiNaC::parser in a way similar to how it's done in default_reader.cpp: #include <ginac/ginac.h> #include <ginac/parser.h> #include <iostream> #include <sstream> #include <string> static GiNaC::ex sqrt_reader(const GiNaC::exvector& ev) { return GiNaC::sqrt(ev[0]); } int main() { std::string input = "sqrt(4)"; std::istringstream instream(input); auto proto = GiNaC::prototype_table(); proto[std::make_pair("sqrt", 1)] = sqrt_reader; auto symtab = GiNaC::symtab(); auto reader = GiNaC::parser(symtab, false, proto); GiNaC::ex expr = reader(instream); std::cout << input << " = " << expr << "\n"; } This unfortunately crashes before entering sqrt_reader. Here's a backtrace: #0 GiNaC::function::eval (this=0x7fffffffcd10) at function.cpp:1450 #1 0x000000000045c212 in GiNaC::ex::construct_from_basic (other=...) at ex.cpp:309 #2 0x000000000059e0d5 in GiNaC::ex::ex (other=..., this=0x7fffffffcc50) at ./ex.h:268 #3 GiNaC::dispatch_reader_fcn (args=std::vector of length 1, capacity 1 = {...}, ptr=<optimized out>) at parser/parser.cpp:62 #4 GiNaC::parser::parse_identifier_expr (this=<optimized out>) at parser/parser.cpp:103 #5 0x000000000059db7d in GiNaC::parser::parse_primary (this=this@entry=0x7fffffffd310) at parser/parser.cpp:173 #6 0x000000000059dc3f in GiNaC::parser::parse_expression (this=this@entry=0x7fffffffd310) at parser/parser.cpp:194 #7 0x000000000059e919 in GiNaC::parser::operator() (this=0x7fffffffd310, input=...) at parser/parser.cpp:226 #8 0x0000000000423e35 in main () at parsingcrash.cpp:21 Can anyone tell me how to do this correctly?

1 0

GiNaC::parser parses "3^+2+1" in the wrong order
by Vitaly Magerya 20 Feb '23

20 Feb '23

Hi, all. When parsing expressions with sign in the exponent, GiNaC's parses confuses the priorities somehow. Here are some examples: 3^2+1 is parsed as (3^2)+1, which is correct; 3^+2+1 is parsed as 3^(+2+1), which is wrong; 3^-2+1 is parsed as 3^(-2+1), which is wrong; 3^+2/3 is parsed as 3^(+2/3), which is wrong; etc. Here's a test case, it will print 27, not the correct 10: #include <ginac/ginac.h> #include <ginac/parser.h> #include <iostream> #include <sstream> #include <string> int main() { std::string input = "3^+2+1"; std::istringstream instream(input); GiNaC::parser reader; GiNaC::ex expr = reader(instream); std::cout << input << " = " << expr << "\n"; }

1 0

type casting
by Marko Riedel 12 Feb '23

12 Feb '23

Hi all, I have a set of custom objects that is declared like this: struct cindsym_is_less { bool operator() (const cindsymbol &lh, const cindsymbol &rh) const { return lh.compare(rh) < 0; } }; typedef std::set<cindsymbol, cindsym_is_less> cindsymset; Now I want to collect all custom objects that appear in an expression tree into this kind of set. I tried this: void indets_recurse(ex src, cindsymset & coll) { if(is_a<add>(src) || is_a<mul>(src) || is_a<power>(src)){ for(size_t idx = 0; idx < src.nops(); idx++) indets_recurse(src.op(idx), coll); return; } if(is_a<cindsymbol>(src)){ coll.insert(src); } } However this will produce the error "no matching function call" because src is an expression and not a cindsymbol. I know that I have my kind of symbol because is_a<cindsymbol> returned true. How can I get at the symbol wrapped in the expression? Best regards, Marko

2 4

symbol equality
by Marko Riedel 09 Feb '23

09 Feb '23

Greetings to all. I have been subclassing "symbol" for my project, where I need to attach an integer value to a symbol. I am not sure this is the best way to go about it but I would like to try. Everything works so far but now I have moved on to substituting variables in polynomials. Here is my sample program. #include <iostream> #include <ginac/ginac.h> #include "cindsymbol.h" #include "util.h" using namespace std; using namespace GiNaC; int main() { cindsymbol R = cindsymbol("R",0), G = cindsymbol("G",0), B = cindsymbol("B",0), X = cindsymbol("X",0); ex pl = X + pow(X, 3) + R + G + B; cout << R << G << B << " " << pl << endl; cout << pl.subs({X}, {pow(X,4)}) << endl; return 0; } The output of this program is as follows (the print functions from the subclass are being invoked correctly and do not print the zero value): RGB X+R+G+B+X^3 4*X^4+X^12 Obviously GiNaC thinks that R=X, B=X and G=X. How do I get this comparison to work? Do I need to override some function in my subclass? The function "cindsymbol" goes as follows (copied from the tutorial): static map<string, cindsymbol> directory; cindsymbol cindsym(const std::string & initname, unsigned cl) { std::string key = initname + "|" + std::to_string(cl); map<std::string, cindsymbol>::iterator it = directory.find(key); if (it != directory.end()) return it->second; else{ std::pair<std::string, cindsymbol> ent = make_pair(key, cindsymbol(initname, cl)); return directory.insert(ent).first->second; } } Can you advise me here? Thanks! Best regards, Marko Riedel

2 1

GiNaC 1.8.6 released
by Richard B. Kreckel 09 Feb '23

09 Feb '23

Hi, GiNaC 1.8.6 has been tagged and released. The News: * Fix wrong numeric info on transcendental functions. * Fix crash of evaluation of binomial(n, k) with negative integer n, k. Happy hacking! -richy. PS: If anyone knows how to push a commit together with the tags using git, do let me know. Yes, I've googled it up and tried several things but it never worked. -- Richard B. Kreckel <https://in.terlu.de/~kreckel/>

2 1

binomial coefficients
by Marko Riedel 07 Feb '23

07 Feb '23

Greetings. I would like to point out that binomial(-5, -7) will produce a segmentation fault. Compare to Maple which says that when binomial(n,k) has k<0 but n-k>=0 we should use binomial(n, n-k) which gives 15 in this case. Best regards, Marko

2 1

Segmentation Fault when calculate cos(cos(1/2))
by 翁湛阳 07 Feb '23

07 Feb '23

Dear GiNaC Team, When I tried to use GiNaC to calculate cos(cos(1/2)),I got a Segmentation Fault. My environment is as following: CPU: Intel i7-10750H Windows10 21H2 gcc version 12.2.0 (MinGW-W64 x86_64-ucrt-posix-seh) GiNaC 1.8.5 CLN 1.3.6 GMP 6.2.1 ArchLinux(in VMwave) Kernel: 6.1.9-arch1-1 GiNaC 1.8.4 installed by pacman CLN 1.3.6 installed by pacman gcc version 12.2.1 20230111(GCC) Relative code file is attached The program crash at line 8.I tried to compile and run the same program at ArchLinux(GCC 12.2.1 GiNaC 1.8.4 CLN 1.3.6),and also got a segmentation fault. Besides , I tried to use ginsh to execute “cos(cos(1/2));” in ArchLinux ,it gave a error: Internal error: statement in file float/misc/cl_F_extendsqrt.cc, line 30 has been reached!! The same error also occurred at Windows,but it just exit without the information above. After that, I built some lower version of GiNaC in Windows10 and re-run my code.I found the bug disappeared at GiNaC 1.8.3 . After reading the commit history I rollback ginac\inifcns_trans.cpp to version 1.8.3 and rebuilt ,the bug disappeared. All my best, -Weng Zhanyang.

2 1

subscripted variables
by Marko Riedel 06 Feb '23

06 Feb '23

Greetings to all. I am at this time considering to use GiNaC for some computations from group theory. I am not a C++ programmer. I would like to ask for assistance with the following task. I have a polynomial in several variables like a[1], a[2], a[5] etc. I understand I can initialize these by creating a symbol object with the variable name. The problem is, I need an easy and conformal means of accessing the integer in the variable name. This could be done by matching the name to a regular expression and extracting the value but I imagine there must be a better way. I am also looking for a means of extracting a set of variables that are present in a polynomial, or the degree these variables have. Furthermore, another question, does the C++ standard library have a means of memoizing functions? Thank you for your assistance. Marko Riedel

2 2