How to write the structure for this particular problem using ceres solver?

Question

I have been trying to optimise a cost function that involves the inverse of a covariance matrix. The covariance matrix is computed recursively - meant to say if parameters to be varied in the opt problem are available at time instants - 0 5 10 15 20 25 Then the covariance matrix is updated recursively in [0,5] ,[5,6].....intervals. And again the elements inside the covariance matrix are dependent on the parameters to be varied in the optimisation problem. Can you suggest how to deal with this problem? Thanks