Property level important

Depending on the approximation goal different error functions are suitable.

Determines the strategy used to select the approximation training data

Another property, but "only" important. This comment is extracted to the end of the comment section or until the next tag.

This description is about the importance of the SubClass property.

The larger the maximal order, the better approximation quality can be achieved, for paying the cost of higher evaluation time.

Specifying a higher tolerance results in sparser but less precise approximations

Choices sign and eps can lead to long computation times if the input samples are large as the full decomposition must be computed.

Determines the size of the reduced space. Has different effects depending on the choice of the general.POD.Mode property.

The flags for the matlab builtin quadprog solver.

If the data is badly conditioned increasing lambda can help.

Influences the performance and time needed for solving.

The sampling strategy affects the quality of the parameter samples used for training.

Choose an appropriate ODE solver for your system.

Defines the end time \(T\) up to which the dynamical system has to be simulated.

If available, supply a jacobian function evaluation handle to improve speed and reliability of implicit solvers.

Providing a sparsity pattern to implicit solvers might be crucial for the performance of the solver due to memory restrictions.

Some odes require a certain (small) initial step.

Including this data in the subspace can reduce projection errors of nonlinearities

Too little minimal improvement causes a too large subspace