Numerical linearization and data recovery

[crupp@…]

I use the linearization feature a lot, but I often forget to build with the +generateSymbolicLinearization flag. Also, building with symbolic generation can take a lot longer and sometimes fails for various reasons. Being able to do numerical linearization would help a lot.

Additionally, being able to recover the algebraic variables after linearization is important. As far as I know, the only way to do this is to do a restart with overrides of the state variables. This, however, is slow and doesn't always work, especially with dynamic states. So, a data recovery option for the linearization would be great. This would essentially be augmenting the state-space system with output variables for each of the algebraic variables.

I have already locally implemented numeric linearization in the runtime code that is used if the symbolic Jacobian is not available. I've also added data recovery code that dumps the data to the same linearization file if requested. I'd like to contribute this code. What is best way of doing this? I could send some patches, but it seems more appropriate to create a fork on git and then somehow get it pushed/merged/pulled (I don't know the correct term)? I haven't yet done any development with git and it is still a bit of a mystery to me, so any pointers would be great.

[Lennart Ochel]

Doing a pull request – as you did - is the best way to contribute. I will review and merge it if there is no issue.

[Lennart Ochel]

Thanks for your contribution.