-
-
Notifications
You must be signed in to change notification settings - Fork 368
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
Analysis API: compute effective sample size #2569
Comments
Looking at |
Not sure if this answers your question, but the number of draws before thinning isn't relevant for n_eff calculations.
We should be using "draws" not "samples" in the code if you want to fix that while you're at it---we'd like to use "sample" to mean a sequence of draws.
|
It seems like `draws` is the preferred label. Do we want to change the
parameter num_samples (CmdStan) to num_draws? Or at least add a note in
the documentation that "num_samples" uses deprecated terminology?
…On 07/06/2018 12:05 AM, Bob Carpenter wrote:
Not sure if this answers your question, but the number of draws before
thinning isn't relevant for n_eff calculations.
We should be using "draws" not "samples" in the code if you want to fix
that while you're at it---we'd like to use "sample" to mean a sequence
of draws.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#2569 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AmFA1oPCkVQlqcF4O4oJ6M7aLnrbxVznks5uDuIdgaJpZM4VEgDy>.
|
I believe the answer to my question is, no, PR #2575 is now posted. I took some liberties on the code, and tried to justify them in the post. Apologies for jumping this conversation across three different pages: Discourse, this issue, and the above PR. |
I hope this is the right place for this issue, apologies in advance if not. |
@kedartal: This is not the right place. This is a specific issue. I don't want to hijack this issue with a discussion of If you have a reproducible example of a bug, you can create a new issue. Otherwise, we have a forum for general discussion if you have a question: |
Can this be closed now? |
It seems this should be closed because of PR #2575 |
Summary:
Offer interfaces a unified calculation of effective sample size.
Description:
This issue is solely for the calculation of ess for a single parameter across multiple chains.
Many details were discussed on the Discourse: Analysis API thread. Design details:
stan-dev/stan:src/stan/analyze/mcmc/compute_ess.hpp
compute_effective_sample_size(std::vector<double*> draws, std::vector<size_t> sizes)
where each chain's draws are stored in a contiguous block of memorycompute_effective_sample_size(std::vector<double*> draws, size_t size)
Current Version:
v2.17.1
The text was updated successfully, but these errors were encountered: