Least-Squares Fitting with asymetric errors

[Erik Ø. Sørensen]

On 5. mai. 2004, at 17.05, Rodney Sparapani wrote:
>>
>> What is the best way of allowing for asymmetric errors?  Is the way to
>> do it redefine my "fi" function in both the function (to fit) and the
>> calculation of its derivative to be a step function?
>> i.e.   if ( yi <= Yi ) use the ErrorBar+ for sigma
>>       if ( yi >  Yi ) use the ErrorBar- for sigma

I am not sure of what you mean, I think a more complete description
would be necessary. But it seems to me that here you are setting
yourself up for a discontinous problem (imagina a yi that is close to 
Yi,
such that a marginal change in parameters will push an observation over 
the edge and give rise to a discrete change in the objective as one 
sigma is replaced by the other).

If I understand what you are trying to do, the likelihood should be a 
weighted sum of the expressions with the two different sigmas, with the 
weights being determined by the probability of that regime given data 
and parameters.

Erik
--
Erik Ø. Sørensen
-------------- next part --------------
A non-text attachment was scrubbed...
Name: S?rensen Erik ?..vcf
Type: text/directory
Size: 437 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/gsl-discuss/attachments/20040506/ad6cb05a/attachment.bin>
-------------- next part --------------