Bill Allombert on Wed, 26 Sep 2012 22:19:41 +0200

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

Re: forprime

On Sat, Sep 22, 2012 at 06:33:01AM +0200, Dirk Laurie wrote:
> 2012/9/21 Karim Belabas <>:
> > (We could apply transcendental functions to
> > square matrices with their usual mathematical meanings instead of
> > current componentwise computation, but nobody ever requested this... :-)
> That is a can of worms.  Even exp(A) when A is a matrix of floating-point
> numbers is notorious ("Nineteen dubious ways to compute the exponential
> of a matrix").

Interestingly, the method used in the old example file matexp.c is rated this way:

     Method 3. Scaling and squaring.
  When properly implemented, the resulting algorithm is one of the most effective we know.

But anyway, I only ever need to solve linear ODE over p-adic fields :).