Bill Allombert on Mon, 11 May 2009 16:10:06 +0200

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: Another problem with matrix inversion

On Mon, May 11, 2009 at 09:57:53AM +0200, Jeroen Demeyer wrote:
> Lorenz Minder wrote:
>> So while A has an inverse in this case, it is not found.  The workaround
>> using chinese remaindering works if the factorization is known, so this
>> is going to be a problem if the modulus is an RSA modulus, for example.
>> It would be better to try to run the algorithm as is, and as soon as a
>> nonzero non-invertible remainder is found, to split into two instances
>> with the now known partial factorization and continue.
> This brings up another question regarding GP: is it possible to trap  
> "impossible inverse modulo" errors and actually recover the element  
> which caused the impossible inverse?  I think it would be nice to have  
> that functionality in GP, but I don't really know what would be the  
> correct way to do it.

This is only possible in libpari mode currently, using the variable

For GP, this is point 5) of bug #329.

Maybe we should add a function errdata() that return the current value
of global_err_data.