computation of the state transition matrix
For some special cases the computation of the state transition matrix returns NaN. This seems to be the case when it is computed using the symbolic matrix exponential. An example is
A = [... 0 0 1 0; ... 0 0 0 1; ... 3 -1 0 0; ... 1 3 0 0]
Quick workaround: Use the ode-solver to compute the transition matrix.
Better solution: Do not compute the transition matrix by the symbolic matrix exponential, use the numeric version of it. The disadvantage of this solution is that the transition matrix can not be returned as a quantity.Symbolic.