Codecov Report

❌ Patch coverage is 97.82609% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 98.49%. Comparing base (4ab5d3b) to head (0ac337c).
⚠️ Report is 58 commits behind head on main.

@@            Coverage Diff             @@
##             main     #322      +/-   ##
==========================================
+ Coverage   98.42%   98.49%   +0.06%     
==========================================
  Files          29       29              
  Lines        5397     5510     +113     
==========================================
+ Hits         5312     5427     +115     
+ Misses         85       83       -2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

Hello @baggepinnen, I still need to implement the tests, but do you have any comments on my points above and the documentation of the new OrthogonalCollocation. The preview of the documentation is available here : OrthogonalCollocation, the associated con_nonlinprogeq! method, and init_orthocolloc.

Thanks for your time!

I no longer have the numerics details in my head, but something I do know tend to be important is numerical scaling by the "typical magnitude" of the variables. A nice benefit of these methods is that they can be made to work for DAEs relatively easily

Benchmark Results (Julia v1)

I no longer have the numerics details in my head, but something I do know tend to be important is numerical scaling by the "typical magnitude" of the variables. A nice benefit of these methods is that they can be made to work for DAEs relatively easily

Yes, right. For now I do not support any scaling factors at all throughout the package. My experience is that most of the time it's possible to "re-scale" the states/outputs/inputs/etc. manually by modifying the f! and h! function, and it's generally a better idea than using the scaling factors available in optimization and nonlinear solver packages, since you have the full control on how you want to do it (e.g. linear or not, offset or not, etc.). And since you did it yourself, there is no weird surprised on the effect of the various parameters e.g. the covariance matrices in the kalman filter, etc.