Repository for an automatic scaling API for Dymos and OpenMDAO
An API for automated scaling of dynamically-constrained optimization problems in Dymos.
As of Version 1.0, the API supports the use of two automatic scaling techniques—-isoscaling (IS) and projected jacobian rows normalization (PJRN)—-for scaling the discretized aspects of small dynamically-constrained optimization problems in Dymos. The API also facilitates the implementation of custom automatic scaling techniques through inheritance from an AutoScaler base class.
For various algorithm implementation details and basic usage tutorials, please see the Autoscaling API Tutorial LaTeX file located in the docs folder. A similar tutorial on general scaling principles can be found on GitHub at https://github.com/hweyandtnasa/scaling-tutorial. (Note that these documents will be added to and improved over time, so check for new versions from time to time!)