Weighted Torque Bug

[harryswift01]

Summary

This PR addresses issues related to the weighted torque calculation by reverting to the original method, optimizing the calculation to prevent crashes, and improving documentation for clarity.

Changes

Revert weighted torque calculation and improve documentation:

• Restored the original method for computing weighted torque.
• Clarified handling of small moment of inertia values in the docstring, explaining why values below machine precision are treated as zero.
• Also included a ValueError check guard on the square root calculation to make sure that the square root does not have any negative numbers within it.
• No functional changes beyond reverting to the previous calculation method.

Optimize weighted torque calculation to reduce crashes:

• Added a check to skip the calculation if the torque is already zero, as highlighted in issue Cannot Compute Weighted Torque #55, reducing unnecessary calculations.
• Retained the divide-by-zero check to ensure numerical stability when computing weighted torques.
• Improved docstring to reflect the updated logic.

Impact

• The reverted method restores the original weighted torque calculation, which maintains the standards that were followed in previous versions of the code but with an explanation as to why this is being calculated this way.
• The optimization prevents unnecessary calculations when the torque is zero, potentially improving performance and preventing crashes in certain scenarios.
• The improved documentation provides better clarity on the treatment of small moment of inertia values and the updated logic for weighted torque calculations.

[skfegan]

In the get_weighted_torques function (in GeometricFunctions.py), there is a check for if the moment of inertia is zero (line 359). I think that if the torques is not zero and the moment of inertia is zero, it should raise a divide by zero error and not just ignore the moment of inertia weighting while carrying on as if nothing unusual happened. As far as I can tell, the physics of the torques and moment of inertia mean that this case should not happen, but I think it is better to have error handling than just trusting it isn't an issue.

[skfegan]

All the logic I expected is there with appropriate error messages.

[jimboid]

The code looks to me like it reflects the output of the discussion with Richard this morning. We basically have one scenario that the program can carry on by setting the torque to zero and two situations that would lead to a graceful crash with informative error messages.

@skfegan I'm not going to hit approve on this review, can you check that you agree and then approve in your review for Harry to merge.

[jimboid]

@harryswift01 oddly this approved with a review ended in "comment"! Can you hold off merging until Sarah has looked too.

[jimboid]

@harryswift01 oddly this approved with a review ended in "comment"! Can you hold off merging until Sarah has looked too.

Ah no Sarah was just fast!