Algorithmic Optimal Control - CO2 Uptake of the Ocean


Here you find a list of software projects developed by our group or jointly with us.

Metos3D: A Marine Ecosystem Toolkit for Optimization and Simulation in 3-D

Metos3D is a free and portable software toolkit for the estimation of parameters of marine ecosystem models in 3-D. For the efficient off-line simulation of passive biogeochemical tracers transport matrices are used. To programmatically decouple the driver routines for the ocean circulation and the marine ecosystem models, a general biogeochemical model interface is defined. For the computation of periodic solutions, the simulation package implements a fixed point iteration (spin-up) and an Newton approach to solve the underlying periodic transport equations. The source code, an installation script, documentation and a comprehensive verification are available on GitHub.

Optimal Experimental Design Toolbox for Matlab

This toolbox allows to optimize model parameters and corresponding measurement conditions.

Often models contain roughly known model parameters. A common way to make such models more realistic is to optimize these parameters so that the model is more consistent with measurements of the modeled variables. These measurements are carried out under different conditions, which crucially affect the information content of the measurement results.

For spatial or temporal models the place and time of the measurements must be chosen. Also often is to decide which entities are measured and which measurement methods are applied. In temporal experiments the state of several values can be influenced over time, too.

All these measurement conditions can be optimized with this toolbox so that the information contain of the corresponding measurement results is maximized. As a result fewer measurements have to be carried out to gain sufficient accurate model parameters which saves time, money and effort.


Optimal PDE control using COMSOL MULTIPHYSICS®

COMSOL (formerly FEMLAB, registered trademark of COMSOL AB, Sweden) is a simulation software for differential equations using the finite element method. Using COMSOL we were able to solve optimal control problems.

Some COMSOL script files are available here.

The matrix-decomposition library

This library allows to approximate Hermitian (dense and sparse) matrices by positive definite matrices. Furthermore it allows to decompose (factorize) positive definite matrices and solve associated systems of linear equations.


The following software projects are also developed by us, but are still in beta stage:


Simulation library:

Python functions for simulating mathematical models, estimating model parameters, quantifying uncertainties and visualizing model results.


Measurements library:

Python functions to handle, statistically analyze and plot measurement data.


Util library:

Python functions used in several other projects.