[GiNaC-devel] infinite loop in simplify_indexed()

[PG CLARK]

I can confirm that I was able to successfully compile and run my test  
code with this patch applied. took about 10 seconds to execute.

Thanks for the help guys

peter

Quoting Alexei Sheplyakov <alexei.sheplyakov at gmail.com>:

> The patch below (drastically) improves the run time at the expense of using
> more RAM in some situations. Please note: it doesn't improve the actual
> algorithm (iteration over all permutations).
>
> diff --git a/ginac/symmetry.cpp b/ginac/symmetry.cpp
> index 7d1ff97..d93e34c 100644
> --- a/ginac/symmetry.cpp
> +++ b/ginac/symmetry.cpp
> @@ -22,6 +22,7 @@
>
>  #include "symmetry.h"
>  #include "lst.h"
> +#include "add.h"
>  #include "numeric.h" // for factorial()
>  #include "operators.h"
>  #include "archive.h"
> @@ -495,7 +496,8 @@ static ex symm(const ex & e,  
> exvector::const_iterator first, exvector::const_ite
>
>  	// Loop over all permutations (the first permutation, which is the
>  	// identity, is unrolled)
> -	ex sum = e;
> +	exvector sum_v;
> +	sum_v.push_back(e);
>  	while (std::next_permutation(iv, iv + num)) {
>  		lst new_lst;
>  		for (unsigned i=0; i<num; i++)
> @@ -505,8 +507,9 @@ static ex symm(const ex & e,  
> exvector::const_iterator first, exvector::const_ite
>  			memcpy(iv2, iv, num * sizeof(unsigned));
>  			term *= permutation_sign(iv2, iv2 + num);
>  		}
> -		sum += term;
> +		sum_v.push_back(term);
>  	}
> +	ex sum = (new add(sum_v))->setflag(status_flags::dynallocated);
>
>  	delete[] iv;
>  	delete[] iv2;
>
>
> Best regards,
> 	Alexei
>
> P.S. The bug has nothing to do with infinite loops.


----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.