Consider expanding testing for analyze_unsteady_convergence

[camUrban]

Problem Statement

analyze_unsteady_convergence()'s level of integration test coverage is unclear.

Location(s): pterasoftware/convergence.py

Proposed Solution

Assess the existing integration test coverage and supplement if necessary.

Additional Context

This issue is referenced by a TEST comment above analyze_unsteady_convergence().

[IlyasFardaouix]

I think one potential cause of unclear test coverage for analyze_unsteady_convergence() is the varying configuration of wing panels across the scripts in unsteady_benchmark.py and unsteady_benchmark_converge.py. Specifically, the num_spanwise_panels parameter is set to 6 in the former but 20 in the latter. Have you considered testing the function with a range of panel configurations to ensure it's robust?

[camUrban]

Hi @IlyasFardaouix. All of this project's tests live in tests. The scripts you are referencing aren't used for testing analyze_unsteady_convergence(). They are used for benchmarking the solver's performance. Performance benchmarking is, for now, done manually, and not a part of the the test suite.

If you'd like to help analyze the test coverage for analyze_unsteady_convergence(), the place to look is tests/integration/test_unsteady_convergence.py 😊

[IlyasFardaouix]

You're right that the scripts in unsteady_benchmark.py and unsteady_benchmark_converge.py aren't directly used for testing analyze_unsteady_convergence(), but the varying panel configurations there could be a good starting point for creating test cases. Before diving into that, though, I'd like to review the current test cases in test_unsteady_convergence.py to see if they cover the function's behavior across different input scenarios. Can you confirm if the test_unsteady_convergence class in that file includes tests with different num_spanwise_panels values?