search

modelica / fmi-standard

Specification of the Functional Mock-up Interface (FMI)
https://fmi-standard.org/
Other
274 stars 85 forks source link

Feature request: Second order derivatives #372

Open modelica-trac-importer opened 6 years ago

modelica-trac-importer commented 6 years ago

Reported by jorn.baayen on 9 Mar 2016 15:57 UTC Currently, FMI 2.0 ME supports first order directional derivatives. This is useful for gradient-based optimization and parameter estimation.

Yet a typical non-linear optimization algorithm also requires the availability of second-order derivative information. While it is possible to estimate the second-order derivatives using iterative methods, exact derivatives typically improve optimization performance significantly. This is especially true if the second-order derivatives are provided as a sparse Hessian matrix.

This is a feature request for an (optional) method to supply second-order derivative information.

Migrated-From: https://trac.fmi-standard.org/ticket/372

modelica-trac-importer commented 6 years ago

Modified by andreas.junghanns on 6 Feb 2017 10:48 UTC

modelica-trac-importer commented 6 years ago

Comment by aviel on 13 Feb 2017 16:19 UTC Changed the milestone to 2.1, since it seems to be a request for a new feature.

modelica-trac-importer commented 6 years ago

Comment by cbertsch on 1 Jun 2018 14:46 UTC This is big issue and would need its own FCP.

modelica-trac-importer commented 6 years ago

Modified by cbertsch on 18 Jun 2018 10:16 UTC

t-sommer commented 3 years ago

Can't this be solved by passing the order as an additional argument like we do in fmi3GetOutputDerivatives()?