> I am currently working on
> http://hub.darcs.net/thielema/hmm-hmatrix
>
> It does not need log-numbers because it normalizes all temporary results.
> This way I can use fast hmatrix operations. Would normalization also be a
> solution for other probabilistic models?
I assume these are models with finite state spaces? FYI I uploaded a Kalman filter https://github.com/idontgetoutmuch/Kalman which is also a HMM. I feel we should be able to abstract over the Markov transition kernel to describe the model without caring whether the state space is continuous or discrete but that it is about as much thinking I have done on the subject.
Dominic Steinitz
dominic(a)steinitz.org
http://idontgetoutmuch.wordpress.com
FYI and please chip in
> Edward Kmett <ekmett <at> gmail.com> writes:
>
> > * random
> >
> > We've had some truly excellent work done over the last couple of
> > years on how to deal with "splitting" a random number generator in a
> > cryptographically sound manner. I spent some time cleaning up a few
> > outstanding issues for this package personally over the summer, but
> > have not had nearly enough time to devote to the issue of how to
> > integrate the outcome of the recent research on splitting, while
> > simultaneously caring about performance and soundness.
>
> I finally managed to get some time to start looking at this. I have
> created a ticket here
>
> https://github.com/haskell/random/issues/25
>
> Please add comments and any information you think Carter and I might
> find helpful.
>
> I discovered the package has some benchmarks that are 3+ years old. At
> the moment, I am not sure what they are telling me. If anyone has any
> background / views please chip in.
>
> Dominic.
Dominic Steinitz
dominic(a)steinitz.org
http://idontgetoutmuch.wordpress.com
