|Re: [eigen] Precision in Cholesky Decomposition|
[ Thread Index |
| More lists.tuxfamily.org/eigen Archives
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] Precision in Cholesky Decomposition
- From: Benoit Jacob <jacob.benoit.1@xxxxxxxxx>
- Date: Mon, 8 Jun 2009 17:45:55 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:in-reply-to:references :date:message-id:subject:from:to:content-type :content-transfer-encoding; bh=iOmoF/yCSbhStSpvCwvDdNWEh1nLc6UYYvz1Ng+Kruk=; b=bit3MVeyTeDSYCHDppFAaTx5RFDt5Iqg5qw2Nl/m0K/USKr0XxUtw1ptw6Nprw7Pa6 n0m46k+nhy+793hChElaTfhs+euSPEqy8MKcfnwSS/QbuwkmNr2s0ymhRr0KErR0aEo6 yI3bGAArj920KpHV5M1szeI6a4Or3KEqRTT4w=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type:content-transfer-encoding; b=ZHqUZtq5wRgAWcfTVSP3+YfLhlRNPhPxPkNm6jRUhQJTGGVW+TWPEcNss3jVgWuDnJ KyCHKebxJk7KCCFULbNxZ9kYpGTFNmDgIXfYEvkdw/Bjn2qWicq8hIfS9I1XYo4E45gJ 9mriEkgPynYCzn/75f4ngBpbjO/sGv/XZoh8Y=
OK, cool! I hope you can stay on the development branch because that
would be quite heavy to backport.
2009/6/8 Scott Stephens <stephens.js@xxxxxxxxx>:
> The development branch works! Thanks for the tip Benoit.
> On Mon, Jun 8, 2009 at 9:44 AM, Benoit Jacob<jacob.benoit.1@xxxxxxxxx> wrote:
>> 2009/6/8 Scott Stephens <stephens.js@xxxxxxxxx>:
>>> I am using Eigen 2.0.2; I will try the development branch.
>>> I just double checked the precision necessary, and 1e-16 works, but
>>> 1e-15 does not. The condition number of the matrix is
>>> 2525609.87787008 (computed in MATLAB), so it would seem to be okay by
>>> your 1e15 criterion.
>> Indeed, that's a very reasonable condition number for double
>> precision, so it just seems that you hit the problems that existed in
>> the LLT implementation in Eigen 2.0.
>> Here's hoping that the development branch works for you! Keep us posted.