perf: exploration of better matmul algorithms

[steven-murray]

This adds a notebook that explores different ways to do the V = Z.dot(Z.T.conj()) calculation.

Ideas that I haven't yet explored:

• [ ] Using MAGMA (see eg. https://github.com/PSims/BayesEoR)
• [ ] Using PyTorch
• [ ] Proper NVIDIA profiling of all the functions

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Comparison is base (d45e9bc) 100.00% compared to head (af6f273) 65.07%. Report is 28 commits behind head on main.

:exclamation: Current head af6f273 differs from pull request most recent head 48d83f7. Consider uploading reports for the commit 48d83f7 to get more accurate results

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #69 +/- ## ============================================ - Coverage 100.00% 65.07% -34.93% ============================================ Files 8 8 Lines 567 567 Branches 88 88 ============================================ - Hits 567 369 -198 - Misses 0 193 +193 - Partials 0 5 +5 ``` | [Flag](https://app.codecov.io/gh/HERA-Team/matvis/pull/69/flags?src=pr&el=flags&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HERA-Team) | Coverage Δ | | |---|---|---| | [unittests](https://app.codecov.io/gh/HERA-Team/matvis/pull/69/flags?src=pr&el=flag&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HERA-Team) | `64.72% <ø> (-34.57%)` | :arrow_down: | Flags with carried forward coverage won't be shown. [Click here](https://docs.codecov.io/docs/carryforward-flags?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=HERA-Team#carryforward-flags-in-the-pull-request-comment) to find out more.

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[mkolopanis]

If you're going through all the efforts of writing a cublas implementation, might investigated compiled extensions like cython or a rust extension. Both have interop with numpy which is pretty straightforward