cached the algorithms in pytest

[oliverchampion]

Describe the changes you have made in this PR

For speed, I've cached the Pytest algorithms such that they are not re-initiated every run. Testing is faster now. Especially good for Matlab-based testing where the Matlab engine needs initiating (different PR)

Link this PR to an issue [optional]

none

Checklist

• [ X ] Self-review of changed code
• [ X ] Added automated tests where applicable
• Update Docs & Guides

[etpeterson]

Two thoughts about this.

• There already is a python cache decorator. Maybe better to use that? https://docs.python.org/3/library/functools.html
• I'm not sure if we actually want to cache. If an algorithm has some kind of memory or learning then reusing it might bias the results. I doubt it, but we don't know how they are implemented. How much time does it save to cache in this way? Could we run more in parallel instead?

[oliverchampion]

@etpeterson, with regards to (1) I will look into this.
With regards to (2), this doesn't make a huge difference with just python code. But Matlab takes 20 seconds to initiate. If this is done for every single test point, this takes ages. Note that I only store algorithms with identical settings.
I would imagine that one would like the algorithm to remain in memory. If it does, we can fix that once it occurs?

[etpeterson]

That is slow. I'm wondering what part of it is taking time. Do you have an example run before and after the caching? Like is it the initialization of Matlab or the algorithm itself?

[oliverchampion]

It is the initiation of Matlab within python. I think it starts some sort of Matlab kernel. I now intimate the python-matlab connection whenever the algorithm is initiated (before it was on a per-fit basis).

…

On Fri, 30 May 2025, 17:25 Eric Peterson, ***@***.***> wrote: *etpeterson* left a comment (OSIPI/TF2.4_IVIM-MRI_CodeCollection#104) <#104 (comment)> That is slow. I'm wondering what part of it is taking time. Do you have an example run before and after the caching? Like is it the initialization of Matlab or the algorithm itself? — Reply to this email directly, view it on GitHub <#104 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB33S7JFPBBEXG6BTTG5RPD3BBZ5HAVCNFSM6AAAAAB6CCG4NSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDSMRSGY4TMOJSGA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[etpeterson]

Does your other ticket make this one obsolete?