This could improve both speed and RAM usage, but could result in a major version change.

Hi @m-d-grunnill, I noticed the following potential typos in the example notebook and the MetaCast package:

• The MultnomialSeeder class is missing the "i" in "Multi"?
• The metacast.event_handling module provides an EventQueue class, but "queue" is spelt "que" in several places, such as:
  • The metacast.event_handling.event_que module (unlikely to be an issue, since the EventQueue class is typically imported from metacast.event_handling)
  • Throughout the demonstration notebook
• "Compliance" is spelt "complaince" in several times in the demonstration notebook.

The last two should be simple to address, while correcting the first one would introduce a breaking change unless the class is exported under both the original and corrected names.

None of these are critical, and they do not detract from the value of the package, the examples presented in the notebook, or the JOSS paper. So please resolve these typos as you see fit, or not at all.

Hi @m-d-grunnill

I cannot run the test cases because they contain import statements that refer to src.metacast, and while src/ is a directory it isn't a valid Python module:

$ python3 tests/test_event_que.py 
Traceback (most recent call last):
  File "/home/.../MetaCast/tests/test_event_que.py", line 9, in <module>
    from src.metacast import MetaCaster
ModuleNotFoundError: No module named 'src'

If I replace src.metacast with metacast, I can run all of the test cases in a virtual environment where I have installed MetaCast, and they all succeed.

Creating a virtual environment (e.g., in ./venv) and installing MetaCast is relatively straightforward, and the following could be saved as a shell script (e.g., run_tests.sh):

$ python3 -m venv ./venv
$ . venv/bin/activate
$ pip install .
$ for test_file in tests/test_*.py; do python3 "${test_file}"; done
....
----------------------------------------------------------------------
Ran 4 tests in 0.553s

OK
.
----------------------------------------------------------------------
Ran 1 test in 0.000s

OK
                                                                                              .
----------------------------------------------------------------------                         
Ran 1 test in 0.960s

OK
........
----------------------------------------------------------------------
Ran 8 tests in 0.259s

OK

NOTE: In the rest of this comment, I provide a suggestion about how to automate the test cases. I do not expect you to implement this kind of automation as a necessary step in the JOSS review process. This is just a suggestion for your consideration, entirely independent of my comments for the JOSS review.

Alternatively, you could consider using an automation tool. For example, I use nox for running test cases, building and publishing online documentation, running code checks, etc. To run the MetaCast test cases with nox, you can create a file called noxfile.py with the following content:

import nox
import pathlib

@nox.session(reuse_venv=True)
def tests(session):
    """Run test cases."""
    session.install('.')
    test_dir = pathlib.Path('tests')
    for test_file in test_dir.glob('test_*.py'):
        session.run('python3', test_file)

If you install nox, you can then run it from the root of the repository to run the test cases:

$ nox
nox > Running session tests
nox > Creating virtual environment (virtualenv) using python3 in .nox/tests
nox > python -m pip install .
nox > python3 tests/test_event_que.py
....
----------------------------------------------------------------------
Ran 4 tests in 0.610s

OK
nox > python3 tests/test_metacaster.py
........
----------------------------------------------------------------------
Ran 8 tests in 0.263s

OK
nox > python3 tests/test_lhs_and_prcc.py
                                                                                              .
----------------------------------------------------------------------                         
Ran 1 test in 1.014s

OK
nox > python3 tests/test_infection_seeding.py
.
----------------------------------------------------------------------
Ran 1 test in 0.000s

OK
nox > Session tests was successful.

You could then add a GitHub workflow that installs Python, then installs and runs nox.

Hi @m-d-grunnill,

Overall, this paper looks good, but I noticed some grammatical and structural issues:

• Under Summary, on line 18, the comma after "models" should be moved to after "flexibly." Also, the acronym "ODE" should be defined before using it.
• Under Statement of need:
  • The third word in the title of this section should be capitalized to match the capitalization pattern shown in the other sections' titles.
  • On line 33: The word "to" should be added between "order" and "expedite."
  • On line 37: The phrase that starts with "The dimensions (groupings)..." is a fragment and should be merged with one of the sentences around it.
  • On line 37: The word "cluster" should be "clusters."
  • On lines 38-40: The sentence that starts with "MetaCast builds..." could be rewritten for greater clarity.
  • On lines 42-43: I'd suggest rewriting the phrase "makes MetaCast an ideal package" in the last sentence of this paragraph, as this phrase was already used in the last sentence of the paragraph right above it (under Summary).
• Under State of Field:
  • On line 46: A comma should be added after "knowledge."
  • On lines 49 and 58: To stay consistent with previous descriptions of the package in the paper, the phrase "for Epidemiology" should be expanded to "for Epidemiology and Ecology."
  • On line 60: Covasim should have the same formatting as Epipack and the other package names.
  • On line 61: The word "a" should be removed.
  • On line 67: The comma at the end of the line should be a colon.
• Under Acknowledgements and Funding, the bullet points are not rendering correctly in the PDF.

Currently the lhs_prcc function only supports sampling from the uniform distribution.

In a previous issue @robmoss mentioned his package lhs. This package may provide features for overcoming this issue.

Hi Martin,

I have some questions about your code, just for educating. Are you coming to the lab next Wednesday?

Hi @m-d-grunnill I noticed the following comment in §5.2 of the demonstration notebook:

"Note currently metacast only support latin hypercube sampling with uniform distributions."

I feel that this is a relatively important limitation, and it should probably be mentioned elsewhere (e.g., README.md).

End users can circumvent this limitation by sampling from the unit interval, and using inverse-CDF functions to then sample from their desired distributions (e.g., using the ppf() method of scipy.stats distributions). So while this is a limitation of the MetaCast package, it still allows users to sample from arbitrary distributions. You could highlight this in §5.2 of the notebook (and maybe in README.md).

Would it be relatively simple to allow users to provide ppf() methods for each parameter? (I'm asking out of interest, not in th expectation that you'll do this as part of the JOSS review).

Aside: I maintain a Latin hypercube sampling package (imaginatively named lhs) that supports sampling from arbitrary distributions, and defining "dependent distributions" whose shape parameters can be functions of the samples drawn for other parameters. Distributions can be defined in a number of ways.