[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: quadl test request for Matlab
From: |
Rik |
Subject: |
Re: quadl test request for Matlab |
Date: |
Thu, 25 Aug 2011 10:10:54 -0700 |
On 08/24/2011 06:50 PM, Dmitri A. Sergatskov wrote:
> On Wed, Aug 24, 2011 at 2:30 PM, Rik <address@hidden> wrote:
>> Well then. It looks like it is the Octave implementation of the algorithm
>> which fails for large tolerances. I suppose we could make this an xtest
>> and ignore the results for now. I don't think anyone is going to leap to
>> fix this and it does make people believe something is wrong with their
>> installed version of Octave.
>>
> With tolerance above 0.25 the recursion never happens because in the
> (is+(i1-i2) == is) condition "is" = 1/eps and condition is true.
> One can force the recursion to happen at least once by doing some dirty trick
> like that:
>
>
> I do not claim it to be a proper fix...
8/25/11
Dmitri,
I went ahead and implemented a solution along the lines you proposed. I
have also cleaned up the code to reflect more modern Octave syntax and code
conventions. The changeset is here
http://hg.savannah.gnu.org/hgweb/octave/rev/85dac13a911b.
Thanks,
Rik