[Biojava-l] Can anyone help me with MarkovModels?

[Matthew Pocock]

 --- Oscar Lopes <oscar at gauss.inesc.pt> wrote: >
Thanks!!
> 
> My problem probabilly is somewhere in the
> transitions distributions.
> Instead of setting the weights myself, I set random
> values using DistributionTools. Now it works fine.
> Maybe I forgot
> something important and didn't noticed...

Great. The DP recursions will do some very strange
things if some of the parameters are not set. I guess
they should scream blue murder.

> 
> By the way, is there any tool to check if a
> MarkovModel is valid or if all
> the distributions have their weights set?

Don't know about valid, but a little "everything is
set to not NaN" or "every propability is non-zero"
checker could be added to a tools class.

> 
> 
>  Oscar Lopes

Matthew

__________________________________________________
Yahoo! Plus
For a better Internet experience
http://www.yahoo.co.uk/btoffer