Modularization Uniformity for ZNE, PEC, DDD, and LRE

[bdg221]

Looking at the different techniques that have been modularized, there are inconsistencies.
ZNE, PEC, and DDD have one function that generates the circuits and another that provides the final expectation value.

LRE has a function that generates the circuits and one that generates the coefficients used in generating the final expectation value.

ZNE: scaled_circuits and combine_results
PEC: generate_sampled_circuits and combine_results
DDD: generate_circuits_with_ddd and combine_results
LRE: multivariate_layer_scaling and multivariate_richardson_coefficients

I think it would be useful to have the function that generates the circuits to have the same naming convention. Also, I think adding a combine_results to LRE makes sense for consistency.

[natestemen]

Agreed. In fact, that was the original intention I believe, but PRs started going up before anything was agreed upon. This is a good place to brainstorm.

Circuit generation

1. intermediary_circuits
2. get_circuits / get_circuits_from_config
3. construct_circuits

Result combination

1. combine_results / process_results
2. compute_expectation_value / compute_ev / compute_expval

---

Of my limited brainstorm, construct_circuits with process_results is my favorite.

[natestemen]

This paper¹ calls the "intermediary circuits" response circuits. We could use this if we wanted to align with something in the literature, but it is my first time hearing the term.

Footnotes

1. Page 2, first column, second paragraph. ↩

[natestemen]

Lets go with construct_circuits and combine_results.

@bdg221 to turn this into an issue.

[bdg221]

Closing the discussion as it is being addressed in #2692