c0pep0d / otto Goto Github PK

Hi @auroreloisy
I just followed the procedure to install the software in a Conda environment.
The tests generate an error: ModuleNotFoundError: No module named 'google.protobuf'
I didn't have any problems with the installation (only some deprecation warnings) and Python 3.8.13 is installed in the env.
(NB: I'm on macOS Big Sur.)
Do you have any idea what's going on?
R.
Review thread: #4266

Overall the submission (#4266) seems very nice. Here's a list of minor comments:

• When running visualise.py -i myparam with N_DIMS = 3 I get the following ValueError 👇 at the end of the run. All .png files are saved in the frames folder but the video is not created. The video is properly created for N_DIMS = 1 and N_DIMS = 2. Maybe document this issue.

Traceback (most recent call last):
  File "/home/michalis/PhD/code/otto/otto/classes/visualization.py", line 423, in _draw_3D
    sc0 = ax[0].scatter(
  File "/home/michalis/anaconda3/envs/ottoenv/lib/python3.8/site-packages/matplotlib-3.5.1-py3.8-linux-
ValueError: s must be a scalar, or float array-like with the same size as x and y

• Consider using python -m pytest tests at the testing section of the README.md file instead of pytest tests. I couldn't run pytest tests because my terminal was looking at another version of pytest. Using the -m switch is generally good practice.
• Minor typo in the Visualizing and evaluating a learned policy section of the README.md file: I believe YYmmdd-HHMMSS_value_model_bkp_i should have been YYmmdd-HHMMSS_model_bkp_i
• Maybe include the contents of the README.md file in the documentation as well. The README.md file did a brilliant job on walking the user through the software but I feel like some of the sections should also be listed in the docs (Installation, How to use OTTO?).
• Line 8 of the draft paper: Not sure if it's a grammatical error or if I don't understand the sentence:
  Far from a “toy” POMDP, it incorporates physical models...

In general this was easy to follow and it was a pleasure to review 👏👏

Hi there!

Thanks a lot for a great software and documentation!

I am currently working on information socialtaxis (similar to this paper). I also plan to explore scenarios where the source moves, and where the source produces different hit distributions.

A the moment I was thinking to integrate OTTO with more general agent-based modeling toolbox Mesa.

Please let me know if you have any recommendations on how to better address these tasks! Or if you have any comments or suggestions!

Valerii

hello! Respected Authors：
when I run the evaluate.py,“It does so essentially by running thousands of episodes in parallel and averaging over those”.So my question is：Do these parallel episodes search the same source？, I mean whether these sources' location in the space are the same?because the source is setting random in the code,right? If not the same,does it still make sense to evaluate different methods to search for sources with varying locations?
I hope my question can be understood and answered. Thank you very much！