-
Notifications
You must be signed in to change notification settings - Fork 0
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feature request: gp_sampler and opt_gp_sampler #61
Comments
Hi, Some GPU accelerated ones were made for cobrexa1 here https://github.com/LCSB-BioCore/CuFluxSampler.jl and I still have to finish the port to cobrexa2, should be available soon. I'll be adding one extra such package with CPU counterparts so this might fit there nicely. (Anyway, TBH, I somehow got to the opinion that there isn't much point in implementing more complicated things than ACHR (or the faster-mixing partial-mean ACHR that we have there too), because most of the benefits of the more complex algorithms are completely outweighted on literally any GPU...) I'll keep this open here as a reminder for later. PRs accepted (if the code is sufficiently small). |
@stelmo is it just me or (opt)gpSampler improvements are so negligible that our ACHR impl already does them? In the paper I can't detect anything that we don't do already. |
Once that update/port is done, updating the docs on that page to reference that package. |
Yep, I am not sure there is much benefit in extending sampling functionality beyond what we have right now. |
Target functionality
Currently Cobrexa implements 1 method for flux sampling:
ACHR.
This is set by passing the
method
argument toflux_sample
but there is no other options provid)ed in the package other thansample_chain_achr
.There are two methods which sucessively build on ACHR:
gp_sampler
opt_gp_sampler
(https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0086587Desired output
flux_sample(model, method=gp_sampler)
flux_sample(model, method=opt_gp_sampler)
The text was updated successfully, but these errors were encountered: