Catalyst ensemble simulation doesn't seem to work correctly in Pluto #3159
Comments
|
Can you also include the script/steps that you run in the repl? |
|
I ran the exact same file in the REPL block by block. Thanks for looking into this! |
|
I suspect that the use of an async plotting library is related. Indeed one difference between the Pluto runtime and the REPL is threading: each cell execution happens in a new thread. I also see that you are modifying the plot in one cell and displaying it in another, which means that the order of execution also plays a role (which makes it hard to debug). It's also possible that different seeding of Random is the issue. To narrow this down: Can you try to write a notebook using a different (synchronous) plotting library like Plots.jl that demonstrates the same issue? Ideally you would send me:
|
|
Will CairoMakie work? In that case, here is the file. I saved the plots to two files called I always have trouble installing Plots with my other packages. It would be helpful if you could give me some pointers about how to identify synchronous vs. async packages. I don't entirely think it's an issue with plotting because it also seems to mess up calculating quantiles at each timestep. Happy to add that if you think it'd be helpful. Somewhat unrelated question, but is there somewhere I can read more about how Pluto is implemented? I'm doing a lot of these simulations, possibly using multithreading, and I'm still trying to figure out the best workflow to run them correctly, avoid re-running them etc. Is it generally better to run simulations on the REPL and use Pluto just for the more lightweight analysis/plotting? |
Hi, I'm having trouble running an ensemble simulation using Catalyst.jl.
Environment:
Thanks for the help!
The text was updated successfully, but these errors were encountered: