Tolerance parameters

[nishnash54]

Solves a number of recent cases regarding ftol, ftol_iter, their implementation, testing and usage.

Description

• Add assertion for ftol_iter
  makes sure ftol_iter >= 0 and int
• Optimize code using deque to replace list
  append() and popleft() methods integrated
• Improvements in unit test
  improve unit test coverage by adding test for assertion, effects on optimizer without ftol, with ftol and with both ftol and ftol_iter based on 0-1 knapsack problem.

TODO

• Improve documentation for ftol and ftol_iter parameters
• Add examples and use cases (demo notebook)
• Implement testing for BinaryPSO

Related Issue

#397 #373

Motivation and Context

Improves documentation, covers testing of tolerance based parameters and improves understanding of the effects of tolerance parameters.

How Has This Been Tested?

Additional testing script for all tolerance related parameters.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

[nishnash54]

Have to update the documentation. Currently the parameters are documented as follows.

• ftol: relative error in objective_func(best_pos) acceptable for convergence. Default is -inf
• ftol_iter: number of iterations over which the relative error in objective_func(best_pos) is acceptable for convergence. Default is 1

@ljvmiranda921 and @whzup would like your inputs on improving the documentation, making it clear as to how the features work independently as well as together.

@hakan-gecili I am aware of that. This is just a basic test bed. Later I will have to integrate this with the TestOptimizer base class, so just kept it simple for now.

[hakan-gecili]

With a very small tweak you can use the Knapsack Problem code for testing BinaryPSO

[ljvmiranda921]

Hey @nishnash54 thanks for taking point on this. This is noted, I'll be checking this again on the weekend! Let me know if this PR is already good for review!

[nishnash54]

Hey @ljvmiranda921 I think most of the stuff is complete and you can review the changes now.
I guess the documentation can be updated later or also moved to the documentation update milestone.

[ljvmiranda921]

Yeah, let's do the documentation after. The notebook suffices for now in this PR. Will wait for @whzup 's review before we proceed! Thanks for this, @nishnash54 !

[nishnash54]

@ljvmiranda921 @whzup
Are these changes fine?

[whzup]

Looks good to me! If @ljvmiranda921 is ok with the formatting changes it should be good to go 😄.

[ljvmiranda921]

Let's merge this then let's resolve msat59's PR and see how we can integrate it