joss-reviews
joss-reviews copied to clipboard
openjournals
[REVIEW]: graphenv: a Python library for reinforcement learning on graph search spaces
Submitting author: @pstjohn (Peter C. St. John) Repository: https://github.com/NREL/graph-env/ Branch with paper.md (empty if default branch): Version: v0.0.5 Editor: @osorensen Reviewers: @iammix, @vwxyzjn, @Viech, @idoby Archive: Pending
Status
Status badge code:
HTML: <a href="https://joss.theoj.org/papers/1d1427f9c72a4a7fb9ac2a53506e0733"><img src="https://joss.theoj.org/papers/1d1427f9c72a4a7fb9ac2a53506e0733/status.svg"></a>
Markdown: [](https://joss.theoj.org/papers/1d1427f9c72a4a7fb9ac2a53506e0733)
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
@iammix & @vwxyzjn & @Viech & @idoby, your review will be checklist based. Each of you will have a separate checklist that you should update when carrying out your review. First of all you need to run this command in a separate comment to create the checklist:
@editorialbot generate my checklist
The reviewer guidelines are available here: https://joss.readthedocs.io/en/latest/reviewer_guidelines.html. Any questions/concerns please let @osorensen know.
✨ Please start on your review when you are able, and be sure to complete your review in the next six weeks, at the very latest ✨
Checklists
Hello humans, I'm @editorialbot, a robot that can help you with some common editorial tasks.
For a list of things I can do to help you, just type:
@editorialbot commands
For example, to regenerate the paper pdf after making changes in the paper's md or bib files, type:
@editorialbot generate pdf
Software report:
github.com/AlDanial/cloc v 1.88 T=0.07 s (818.7 files/s, 128621.4 lines/s)
-------------------------------------------------------------------------------
Language files blank comment code
-------------------------------------------------------------------------------
Python 28 967 1577 2962
Markdown 6 108 0 319
Jupyter Notebook 7 0 2307 279
YAML 6 36 115 216
TeX 1 12 0 134
reStructuredText 7 57 131 62
JSON 1 0 0 58
DOS Batch 1 8 1 26
Bourne Shell 1 1 7 16
make 2 5 7 15
-------------------------------------------------------------------------------
SUM: 60 1194 4145 4087
-------------------------------------------------------------------------------
gitinspector failed to run statistical information for the repository
Reference check summary (note 'MISSING' DOIs are suggestions that need verification):
OK DOIs
- 10.1016/j.patter.2021.100361 is OK
- 10.1039/d1sc02770k is OK
- 10.1038/s41597-020-00588-x is OK
- 10.1038/s41467-020-16201-z is OK
- 10.1007/978-3-030-50426-7_33 is OK
- 10.1038/s41598-019-47148-x is OK
MISSING DOIs
- None
INVALID DOIs
- 10.1038/s42256-022-00506-3 is INVALID
:point_right::page_facing_up: Download article proof :page_facing_up: View article proof on GitHub :page_facing_up: :point_left:
Review checklist for @idoby
Conflict of interest
- [x] I confirm that I have read the JOSS conflict of interest (COI) policy and that: I have no COIs with reviewing this work or that any perceived COIs have been waived by JOSS for the purpose of this review.
Code of Conduct
- [x] I confirm that I read and will adhere to the JOSS code of conduct.
General checks
- [x] Repository: Is the source code for this software available at the https://github.com/NREL/graph-env/?
- [x] License: Does the repository contain a plain-text LICENSE file with the contents of an OSI approved software license?
- [x] Contribution and authorship: Has the submitting author (@pstjohn) made major contributions to the software? Does the full list of paper authors seem appropriate and complete?
- [x] Substantial scholarly effort: Does this submission meet the scope eligibility described in the JOSS guidelines
Functionality
- [x] Installation: Does installation proceed as outlined in the documentation?
- [x] Functionality: Have the functional claims of the software been confirmed?
- [x] Performance: If there are any performance claims of the software, have they been confirmed? (If there are no claims, please check off this item.)
Documentation
- [x] A statement of need: Do the authors clearly state what problems the software is designed to solve and who the target audience is?
- [x] Installation instructions: Is there a clearly-stated list of dependencies? Ideally these should be handled with an automated package management solution.
- [x] Example usage: Do the authors include examples of how to use the software (ideally to solve real-world analysis problems).
- [x] Functionality documentation: Is the core functionality of the software documented to a satisfactory level (e.g., API method documentation)?
- [x] Automated tests: Are there automated tests or manual steps described so that the functionality of the software can be verified?
- [x] Community guidelines: Are there clear guidelines for third parties wishing to 1) Contribute to the software 2) Report issues or problems with the software 3) Seek support
Software paper
- [x] Summary: Has a clear description of the high-level functionality and purpose of the software for a diverse, non-specialist audience been provided?
- [x] A statement of need: Does the paper have a section titled 'Statement of need' that clearly states what problems the software is designed to solve, who the target audience is, and its relation to other work?
- [x] State of the field: Do the authors describe how this software compares to other commonly-used packages?
- [x] Quality of writing: Is the paper well written (i.e., it does not require editing for structure, language, or writing quality)?
- [x] References: Is the list of references complete, and is everything cited appropriately that should be cited (e.g., papers, datasets, software)? Do references in the text use the proper citation syntax?
Review checklist for @iammix
Conflict of interest
- [x] I confirm that I have read the JOSS conflict of interest (COI) policy and that: I have no COIs with reviewing this work or that any perceived COIs have been waived by JOSS for the purpose of this review.
Code of Conduct
- [x] I confirm that I read and will adhere to the JOSS code of conduct.
General checks
- [x] Repository: Is the source code for this software available at the https://github.com/NREL/graph-env/?
- [x] License: Does the repository contain a plain-text LICENSE file with the contents of an OSI approved software license?
- [x] Contribution and authorship: Has the submitting author (@pstjohn) made major contributions to the software? Does the full list of paper authors seem appropriate and complete?
- [x] Substantial scholarly effort: Does this submission meet the scope eligibility described in the JOSS guidelines
Functionality
- [x] Installation: Does installation proceed as outlined in the documentation?
- [x] Functionality: Have the functional claims of the software been confirmed?
- [x] Performance: If there are any performance claims of the software, have they been confirmed? (If there are no claims, please check off this item.)
Documentation
- [x] A statement of need: Do the authors clearly state what problems the software is designed to solve and who the target audience is?
- [x] Installation instructions: Is there a clearly-stated list of dependencies? Ideally these should be handled with an automated package management solution.
- [x] Example usage: Do the authors include examples of how to use the software (ideally to solve real-world analysis problems).
- [x] Functionality documentation: Is the core functionality of the software documented to a satisfactory level (e.g., API method documentation)?
- [x] Automated tests: Are there automated tests or manual steps described so that the functionality of the software can be verified?
- [x] Community guidelines: Are there clear guidelines for third parties wishing to 1) Contribute to the software 2) Report issues or problems with the software 3) Seek support
Software paper
- [x] Summary: Has a clear description of the high-level functionality and purpose of the software for a diverse, non-specialist audience been provided?
- [x] A statement of need: Does the paper have a section titled 'Statement of need' that clearly states what problems the software is designed to solve, who the target audience is, and its relation to other work?
- [x] State of the field: Do the authors describe how this software compares to other commonly-used packages?
- [x] Quality of writing: Is the paper well written (i.e., it does not require editing for structure, language, or writing quality)?
- [x] References: Is the list of references complete, and is everything cited appropriately that should be cited (e.g., papers, datasets, software)? Do references in the text use the proper citation syntax?
This seems like a very interesting package! I will have the time to dive in more in-depth soon, but in the meanwhile, can the authors fix the citation on line 17 of the paper to be consistent with the other citations, and resolve the invalid citation issue?
Also, please clarify the contributions of the two extra authors who are not listed as collaborators on the GitHub repo.
Thanks!
:point_right::page_facing_up: Download article proof :page_facing_up: View article proof on GitHub :page_facing_up: :point_left:
Hello @idoby, here are the things you can ask me to do:
# List all available commands
@editorialbot commands
# Get a list of all editors's GitHub handles
@editorialbot list editors
# Check the references of the paper for missing DOIs
@editorialbot check references
# Perform checks on the repository
@editorialbot check repository
# Adds a checklist for the reviewer using this command
@editorialbot generate my checklist
# Set a value for branch
@editorialbot set joss-paper as branch
# Generates the pdf paper
@editorialbot generate pdf
# Get a link to the complete list of reviewers
@editorialbot list reviewers
Reference check summary (note 'MISSING' DOIs are suggestions that need verification):
OK DOIs
- 10.1016/j.patter.2021.100361 is OK
- 10.1039/d1sc02770k is OK
- 10.1038/s41597-020-00588-x is OK
- 10.1038/s41467-020-16201-z is OK
- 10.1007/978-3-030-50426-7_33 is OK
- 10.1038/s41598-019-47148-x is OK
MISSING DOIs
- None
INVALID DOIs
- 10.1038/s42256-022-00506-3 is INVALID
Apologies if these contribution edits should have been made in the manuscript?
- Dmitry Duplyakin contributed documentation for running graph-env on AWS, and debugged a lot of that workflow (NREL/graph-env#22).
- Jeffrey Law has contributed extensively to downstream repositories NREL/rlmolecule, and his feedback has been helpful in the direction of this more general-purpose library
That invalid DOI, 10.1038/s42256-022-00506-3, should be live starting 04 August 2022 at 11:00 (US Eastern Time)
Apologies if these contribution edits should have been made in the manuscript?
- Dmitry Duplyakin contributed documentation for running graph-env on AWS, and debugged a lot of that workflow (Add config and doc for TSP on Ray at AWS NREL/graph-env#22).
- Jeffrey Law has contributed extensively to downstream repositories NREL/rlmolecule, and his feedback has been helpful in the direction of this more general-purpose library
I'm pretty new at reviewing for JOSS, but the documentation does not state this information has to be specified in the manuscript. So I think just your comment should be fine.
That invalid DOI, 10.1038/s42256-022-00506-3, should be live starting 04 August 2022 at 11:00 (US Eastern Time)
OK, LGTM then. Appreciate your work on the library and thank you for contributing it. :)
@pstjohn Looking at it now, it seems like the merge commit for Dmitry's PR prevented him being listed as a contributor by GitHub. I'm not sure why, since it usually works fine.
Apologies if these contribution edits should have been made in the manuscript?
- Dmitry Duplyakin contributed documentation for running graph-env on AWS, and debugged a lot of that workflow (Add config and doc for TSP on Ray at AWS NREL/graph-env#22).
- Jeffrey Law has contributed extensively to downstream repositories NREL/rlmolecule, and his feedback has been helpful in the direction of this more general-purpose library
I'm pretty new at reviewing for JOSS, but the documentation does not state this information has to be specified in the manuscript. So I think just your comment should be fine.
That invalid DOI, 10.1038/s42256-022-00506-3, should be live starting 04 August 2022 at 11:00 (US Eastern Time)
OK, LGTM then. Appreciate your work on the library and thank you for contributing it. :)
Wanted to chime in as editor that I agree with you. Comments in this thread are sufficient.
Review checklist for @vwxyzjn
Conflict of interest
- [x] I confirm that I have read the JOSS conflict of interest (COI) policy and that: I have no COIs with reviewing this work or that any perceived COIs have been waived by JOSS for the purpose of this review.
Code of Conduct
- [x] I confirm that I read and will adhere to the JOSS code of conduct.
General checks
- [x] Repository: Is the source code for this software available at the https://github.com/NREL/graph-env/?
- [x] License: Does the repository contain a plain-text LICENSE file with the contents of an OSI approved software license?
- [x] Contribution and authorship: Has the submitting author (@pstjohn) made major contributions to the software? Does the full list of paper authors seem appropriate and complete?
- [x] Substantial scholarly effort: Does this submission meet the scope eligibility described in the JOSS guidelines
Functionality
- [x] Installation: Does installation proceed as outlined in the documentation?
- [x] Functionality: Have the functional claims of the software been confirmed?
- [x] Performance: If there are any performance claims of the software, have they been confirmed? (If there are no claims, please check off this item.)
Documentation
- [x] A statement of need: Do the authors clearly state what problems the software is designed to solve and who the target audience is?
- [x] Installation instructions: Is there a clearly-stated list of dependencies? Ideally these should be handled with an automated package management solution.
- Yes, but there are no locked versions. This means the software could break in the future when dependencies introduce breaking changes.
- [x] Example usage: Do the authors include examples of how to use the software (ideally to solve real-world analysis problems).
- [x] Functionality documentation: Is the core functionality of the software documented to a satisfactory level (e.g., API method documentation)?
- [x] Automated tests: Are there automated tests or manual steps described so that the functionality of the software can be verified?
- [x] Community guidelines: Are there clear guidelines for third parties wishing to 1) Contribute to the software 2) Report issues or problems with the software 3) Seek support
Software paper
- [x] Summary: Has a clear description of the high-level functionality and purpose of the software for a diverse, non-specialist audience been provided?
- [x] A statement of need: Does the paper have a section titled 'Statement of need' that clearly states what problems the software is designed to solve, who the target audience is, and its relation to other work?
- [x] State of the field: Do the authors describe how this software compares to other commonly-used packages?
- [x] Quality of writing: Is the paper well written (i.e., it does not require editing for structure, language, or writing quality)?
- [x] References: Is the list of references complete, and is everything cited appropriately that should be cited (e.g., papers, datasets, software)? Do references in the text use the proper citation syntax?
Reference check summary (note 'MISSING' DOIs are suggestions that need verification):
OK DOIs
- 10.1016/j.patter.2021.100361 is OK
- 10.1039/d1sc02770k is OK
- 10.1038/s41597-020-00588-x is OK
- 10.1038/s41467-020-16201-z is OK
- 10.1038/s42256-022-00506-3 is OK
- 10.1007/978-3-030-50426-7_33 is OK
- 10.1038/s41598-019-47148-x is OK
MISSING DOIs
- None
INVALID DOIs
- None
The Paper and repo look good! Opened a couple of issues:
https://github.com/NREL/graph-env/issues/33 https://github.com/NREL/graph-env/issues/34
Also, please consider filling out the github repo's description and keywords. See the screenshot below as an example.
Review checklist for @Viech
Conflict of interest
- [x] I confirm that I have read the JOSS conflict of interest (COI) policy and that: I have no COIs with reviewing this work or that any perceived COIs have been waived by JOSS for the purpose of this review.
Code of Conduct
- [x] I confirm that I read and will adhere to the JOSS code of conduct.
General checks
- [x] Repository: Is the source code for this software available at the https://github.com/NREL/graph-env/?
- [x] License: Does the repository contain a plain-text LICENSE file with the contents of an OSI approved software license?
- [x] Contribution and authorship: Has the submitting author (@pstjohn) made major contributions to the software? Does the full list of paper authors seem appropriate and complete?
- [x] Substantial scholarly effort: Does this submission meet the scope eligibility described in the JOSS guidelines
Functionality
- [x] Installation: Does installation proceed as outlined in the documentation?
- [x] Functionality: Have the functional claims of the software been confirmed?
- [x] Performance: If there are any performance claims of the software, have they been confirmed? (If there are no claims, please check off this item.)
Documentation
- [x] A statement of need: Do the authors clearly state what problems the software is designed to solve and who the target audience is?
- [x] Installation instructions: Is there a clearly-stated list of dependencies? Ideally these should be handled with an automated package management solution.
- [x] Example usage: Do the authors include examples of how to use the software (ideally to solve real-world analysis problems).
- [x] Functionality documentation: Is the core functionality of the software documented to a satisfactory level (e.g., API method documentation)?
- [x] Automated tests: Are there automated tests or manual steps described so that the functionality of the software can be verified?
- [x] Community guidelines: Are there clear guidelines for third parties wishing to 1) Contribute to the software 2) Report issues or problems with the software 3) Seek support
Software paper
- [x] Summary: Has a clear description of the high-level functionality and purpose of the software for a diverse, non-specialist audience been provided?
- [x] A statement of need: Does the paper have a section titled 'Statement of need' that clearly states what problems the software is designed to solve, who the target audience is, and its relation to other work?
- [x] State of the field: Do the authors describe how this software compares to other commonly-used packages?
- [x] Quality of writing: Is the paper well written (i.e., it does not require editing for structure, language, or writing quality)?
- [x] References: Is the list of references complete, and is everything cited appropriately that should be cited (e.g., papers, datasets, software)? Do references in the text use the proper citation syntax?
I have completed my review with some remarks below and will tick my remaining boxes when these points were addressed or discussed. Overall, I find that this is a small but nicely maintained and documented package whose scope could be described a bit better to a non-specialist audience.
General checks
Substantial scholarly effort
Functionally this is a very small library: Excluding the examples subfolder, there are significantly less than 1000 lines of code in the graphenv folder that appears to contain the library as such, so I should flag this to @osorensen. This code gives the impression of a joint wrapper around two existing libraries. On the other hand, a substantial effort is apparent in the presentation and integration of this code in the form of tests, examples, and deploymet scripts. I can also see that the functionality offered is useful and not accessible to users not familiar with the backend libraries used.
Functionality
Installation
- I have opened #36.
Documentation
A statement of need
- JOSS asks for the target audience to be mentioned in the documentation.
Installation instructions
- The dependencies are handled by
pipbut it would be good to have them listed also in the documentation (for working with the source and for external package managers). - It appears that one can choose between
tensorflowandtorch, with one of them being required. This is not documented in the "Installation" section.
Manuscript
I find that the manuscript is well-written and concise but the distinction between graph search problems and other problem representations could be made more clear as this seems to be the selling point for the software.
Author list
- If possible with the template, you should use inter-word spacing between "St." and "John".
- If possible, the digit 2 and the pilcrow in the author affiliations would look better on the respective next line, before their associated text. (Also, I would expect some contact info for the corresponding author; was this missed or is it not intended?)
Summary
- The concept of "strong relational inductive biases" is not made clear to a general audience.
- It would be nice to get an intuition on why certain problems are graph search representable but do not have an algebraic form (beyond mentioning molecular optimization as an example) as this seems to be the main motivation to use your package over classical optimization techniques.
- "or integer program" -> "or integer programs"?
- I don't understand why you mention "with well defined linear objective function and linear constraints": If linearity is important here, you could just write "integer linear programs". Otherwise, mentioning linearity does not support your argument that graph search problems are more expressive than classical representations.
- You use both a "traveling salesman problem" and "Traveling Salesman Problem" capitalization.
- You write "RLLib" while the associated paper writes it as "RLlib" in the title.
Statement of need
- JOSS asks for the target audience to be mentioned here.
- In the RLLib paragraph, it is not clear how the concepts of parametrically defined and invalid actions, flattening, masking, and successor state relate to what we learned in the summary. If these are central concepts, you should consider introducing them briefly in the summary. Otherwise this paragraph might be too technical for a statement of need.
Thanks for your review @Viech!
Regarding scholarly scope, the guidelines state as some factors to consider, which I will consider point-by-point below, while trying to "think out loud".
- Age of software (is this a well-established software project) / length of commit history.
Looking at the repository, I find that the software seems rather young, with the first commit made on February 11 2022.
- Number of commits.
This is rather large, 241 in total.
- Number of authors.
There have been four contributors, which suggests that this is a joint effort.
- Total lines of code (LOC). Submissions under 1000 LOC will usually be flagged, those under 300 LOC will be desk rejected.
This one has already been commented on in @Viech's review.
- Whether the software has already been cited in academic papers.
Could you please comment on this @pstjohn? Is the software used in academic paper, or are you aware of papers in preparation which will use the package.
- Whether the software is sufficiently useful that it is likely to be cited by your peer group.
This is an important point. Making a joint wrapper around two existing libraries can in itself be very useful for users not familiar with those libraries.
In sum, I understand your concern about scholarly scope @Viech, and see that this is a difficult case. I would also be interested to hear what @iammix thinks of this, as the "Substantial scholarly effort" button has not been ticked off in your review checklist.
Also @Viech, could you think of any suggested extensions of the package that would make it more clearly within scope for JOSS?
You're of course also welcome to comment on this @pstjohn.
@osorensen Just adding a comment here since I did check the respective checkbox. Following the discussion here I went back and checked the commit history in greater depth and found that a large number of commits are just one-liners or seem to be purely about cosmetics/experimenting with settings.
So I would suggest the authors
- document these successful (or less successful) configurations and experiments to convince the reader that their library is indeed robust and stands for more than just integration between various libraries (which is great but personally not something I would read a journal to find out about), and
- go back and clean up some of the history so that important development milestones stand out more clearly.
Personally, I believe git history to be more than just a storage space for old versions of the code, it also serves to document the design and development process for a project, document decisions made, etc.
(also, apologies, this is my first review and I didn't realize reviews could extend beyond the checklist)
Thanks @idoby! And for the record, you're also free to update your checklist, so that items are unchecked until certain issues have been fixed.
Also @Viech, could you think of any suggested extensions of the package that would make it more clearly within scope for JOSS?
I'm more into "classical" optimization than machine learning, so I would not know about suited core features to be added. However one could certainly add more interface features: Default implementations of the while not done main loop with options for detailed output/logging of what is going on and control over stopping criteria, ability to stop, save, load, and continue a search, plotting of (parts of) the search graph, statistics concerning this graph, or additional ways to define a problem (e.g. by assigning callback functions to properties of some GraphSearchProblem instance so that the library can be used in an interactive console) are some potential features that come to my mind. I'm also just thinking out loud here; some of these ideas might not apply to graphenv. The general direction of these suggestions is that, ideally,
- users of
graphenvshould be able to get basic functionality out of it without importinggymorray.rllibthemselves and/or - they should be offered advanced "quality of life" features if they are already comfortable using these backend libraries directly.
With either of these directions pursued, the package would be clearly within scope as I understand it.
Thanks @Viech. Then I suggest that the authors (@pstjohn) work along the lines proposed in your comment above, in addition to addressing the points raised in the review.
Hello, sorry for my late response. I will finish my review by the end of the week.
Hi all,
Thanks for the excellent comments (so far) on this effort! We've made a few changes to the repo and manuscript (see issues NREL/graph-env#33, NREL/graph-env#34, NREL/graph-env#36, NREL/graph-env#37, NREL/graph-env#39, and NREL/graph-env#40 for additional discussion), and here I'll respond generally to some of the bigger picture comments on the size, age, and git history that have been raised.
This library was originally conceived as a refactor to our NREL/rlmolecule library, which used significantly more code to accomplish a similar task. We felt the amount of effort to integrate a tree search gym environment with a custom policy model in RLlib was nontrivial and might be of interest to the growing field of researchers looking at applying RL to new optimization problems. This in part explains the age of the library, our original work on these types of problems began in 2019, but we only recently refactored the code in this more general library. Our recent publication cites that rlmolecule repository, although a lot of that work was ported here.
We subsequently spent a long time attempting to reduce the size of the library, trying to optimize speed and code maintainability. For instance, we removed the 177 line space_util.py functions after finding ways to leverage the built-in repeated action space within rllib.
This also explains in part the large number of "small" commits; debugging RLlib can be challenging (as most of the work occurs in forked worker processes) and as a result we took a rather circuitous route to our current implementation. We're documenting current "gotcha's" in our online documentation, but could add a "Lessons Learned" section to our documentation moving forward to keep track of knowledge gained through failed experiments.
On re-writing the git history, I'm not confident enough in my git abilities to force-push main on a repo with PyPI integration and automated docs tied directly to SCM version tags :cold_sweat:. But moving forward, we will certainly package major development milestones into pull requests and merge those as a single commit.
We appreciate everyone's involvement so far! Please let us know if any of our responses could use additional detail. We put a response together around project scope in NREL/graph-env#37, and would be happy to continue those discussions there or in this thread.
(signed all authors)
I've reviewed the changes to the manuscript and documentation and they look good to me!
With respect to scope I want to stress that I can see this package being used and cited in research and that the code and repository quality look pretty good; so I've ticked that box now even though not much new substance was added to the core code at this point. I also understand that using compact constructs can improve code quality while reducing size and that the efficient use of a technical backend library can be a feature in itself.
If I may chime in concerning a history rewrite: I think this should not be relevant for publication. I see the value that a nice history has for understanding the ideas behind pieces of code but rewriting git history is typically considered bad style as it breaks forks, branches, and leaves important tags detached from the new history. I would suggest this in extreme cases but not here.