commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sébastien Brisard <sebastien.bris...@m4x.org>
Subject Re: [Math] Question about "AbstractLeastSquaresOptimizer" code
Date Tue, 24 Apr 2012 19:42:54 GMT
Ooops, I hit the wrong key... So here is the rest of my previous message...

Hi Gilles,
>
> Does anyone object to the estimated error be defined as
>
>  errors[i] = FastMath.sqrt(covar[i][i]);
>
> ?
>
>
> Best regards,
> Gilles
>
I'm not an expert on parameter estimation, but I'm not sure about the
removal of the c prefactor in the above expression. See for example
NR, chapter 15. In particular, equation 15.6.4 might be of interest.
I also found this thread
http://www.physicsforums.com/showthread.php?t=47534
which certainly does not provide a definite answer, but the formula
suggests that c should stay.

Again, I'm not claiming your suggestion is wrong, I would just like to
find a reference which clearly states the right answer...
One way to check the formula would be to test against another
software. I know that gnuplot provides error estimates on the
parameters, I'm happy to run it with the data set of your choice.

Best regards,
Sébastien


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org


Mime
View raw message