Join GitHub today
GitHub is home to over 36 million developers working together to host and review code, manage projects, and build software together.Sign up
[REVIEW]: fmcmc: A friendly MCMC framework #1427
Status badge code:
Reviewers and authors:
Please avoid lengthy details of difficulties in the review thread. Instead, please create a new issue in the target repository and link to those issues (especially acceptance-blockers) by leaving comments in the review thread below. (For completists: if the target issue tracker is also on GitHub, linking the review thread in the issue or vice versa will create corresponding breadcrumb trails in the link target.)
Reviewer instructions & questions
@fabian-s, please carry out your review in this issue by updating the checklist below. If you cannot edit the checklist please:
The reviewer guidelines are available here: https://joss.readthedocs.io/en/latest/reviewer_guidelines.html. Any questions/concerns please let @arokem know.
Review checklist for @fabian-s
Conflict of interest
Code of Conduct
referenced this issue
May 3, 2019
If you haven't already, you should seriously consider unsubscribing from GitHub notifications for this (https://github.com/openjournals/joss-reviews) repository. As a reviewer, you're probably currently watching this repository which means for GitHub's default behaviour you will receive notifications (emails) for all reviews
To fix this do the following two things:
For a list of things I can do to help you, just type:
This was referenced
May 10, 2019
LICENSE is a placeholder: USCbiostats/fmcmc#5
tagged release version is missing: USCbiostats/fmcmc#6
I don't see a clear "statement of need and audience" in the repo.
Documentation issue for proposal: USCbiostats/fmcmc#7
Another "documentation" issue I can't pin on any specific function is the more general question on how to specify any priors?
DOIs are missing in references: USCbiostats/fmcmc#8
You have a "statement of need" in your paper which I find rather hyperbolic, TBH. The ultimate flexibility you claim comes at the cost of users being required to 1) implement the (log-)likelihood/posterior themselves, and 2) figuring out a suitable scaling of the proposal (kernel) for reasonable accceptance rates, which will be challenging in almost any interesting application. In addition, it seems like a stretch to claim a "pure R" implementation if some of the heavy lifting is actually being done in C++. Could the last paragraph be phrased more accurately?
Typos: "this alows" --> allows, "convergance" --> convergence, "existing mcmc packages" --> MCMC packages
I've addressed most of your comments on USCbiostats/fmcmc#9 (thanks for them!). Regarding the rest of the comments:
There is no need to have a special argument for that. Any set of particular priors can be added directly in the objective function. For example, suppose that your log-likelihood function looks like this:
If you want to add priors to the third parameter, say you think it is distributed Beta with parameters 2 and 8, you can do it by setting:
So there is no need to do anything else than that. The algorithm will just work.
I've added a statement of need as an introduction in the README.md file in which I explicitly state who should be using this R package. This, as other MCMC packages, are for those who want to implement their own model via a log-likelihood function (for example). In particular:
That was a very good point. We actually figured out how to rewrite a function (which was in C++) to be as efficient using only base R. Right now the package doesn't need to be compiled anymore.
Please let me know if you have any additional comments :).
Done, and added another MCMC R package/paper that was relevant for the literature review. @fabian-s I will close the issues you started. Thanks!
No, it was something related to the version of OSX I was using. Using a more recent one fixes the issue (confirmed here).
@fabian-s I've added you to the list of authors of the R package with the role "rev" (Reviewer). LMK if that's OK with you (pulled your ORCID from one of your papers).
@gvegayon : looks like all is well here. I have one small comment on the manuscript. The word "Gaussian" is derived from a name and should be capitalized (appears twice, as far as I can see).
Once you correct this, could you please create an archive of the software (e.g., using https://zenodo.org) and post the DOI on this thread?
If you would like to include a link to your paper from your README use the following code snippets:
This is how it will look in your documentation:
We need your help!
Journal of Open Source Software is a community-run journal and relies upon volunteer effort. If you'd like to support us please consider doing either one (or both) of the the following: