[OpendTect_Developers] On variances.

[Bruno Chaintreuil]

Dear Fidel, All,

Thanks for sharing your experience!

It might indeed be related to a precision issue, but it might also be 
that the algorithm they use (the one I gave previously) is not the best 
one.

See http://en.wikipedia.org/wiki/Algorithms_for_calculating_variance

We also implemented the Two-pass algorithm in OpendTect, which is often 
more accurate.

Best Regards,
Bruno

-- 
Bruno Chaintreuil, MSc.
Software Engineer
dGB Earth Sciences B.V.
Nijverheidstraat 11-2, 7511 JM Enschede, The Netherlands
bruno.chaintreuil at dgbes.com
Tel: +31 534315155 , Fax: +31 534315104
http://www.dgbes.com



FIDEL REYES RAMOS wrote:
>
> Hi everyone,
>
>  
>
> Let me write you my own experiences:
>
> I  have been working with variances also, and machines can calculate 
> negative values.
>
>  
>
> Depending of the tails of the distribution, if they are not Gaussian, 
> and one tail is longer than the other, this distribution may have 
> negative variances. You can measure the asymmetry of the tails with 
> skewness.
>
>  
>
> Some spectra can give negative variances, for example from 
> instantaneous spectra of some time frequency distributions, and this 
> is a problem because negative variances does not have a physical meaning.
>
>  
>
> What do you think?
>
> Best regards.
>
> Fidel.
>
>  
>
>  
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Developers mailing list
> Developers at opendtect.org
> http://lists.opendtect.org/mailman/listinfo/developers
>