feat: add benchmarks (skeleton)

[redeboer]

Closes #103

[redeboer]

@spflueger We'll need so think of a smart way to smear the initial parameters after data generation and before fit. Will also need to put the example recipe files -- created with benchmark/create_recipe.py -- under source control, so that we can things like omega mass width a bit more wide.

[codecov]

Codecov Report

Merging #106 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #106   +/-   ##
=======================================
  Coverage   78.73%   78.73%           
=======================================
  Files          14       14           
  Lines         649      649           
=======================================
  Hits          511      511           
  Misses        138      138           

[spflueger]

@spflueger We'll need so think of a smart way to smear the initial parameters after data generation and before fit. Will also need to put the example recipe files -- created with benchmark/create_recipe.py -- under source control, so that we can things like omega mass width a bit more wide.

That's true. Can you open an issue for this random initial parameter problem.

I think its totally ok uploading a fixed model recipe for testing purposes.

[redeboer]

That's true. Can you open an issue for this random initial parameter problem.

There's already this one, right? ComPWA/expertsystem#8

[redeboer]

@spflueger 2d6d300 still doesn't address the problem, because there's no clean way to fix or modify width and mass from the python interface.

[spflueger]

That's true. Can you open an issue for this random initial parameter problem.

There's already this one, right? ComPWA/expertsystem#8

This is something else. Or did I misunderstand you here? You are talking about some function which randomly picks a starting point in the optimization space right?