The extract_sparse_data_from_model function was causing kalman filter tests to fail because of how I wrote the numba wrappers. I also got it to cause sigfaults, which is obviously bad.
The entire method of numbafying the model for estimation needs to be revisited. Preferably implement a function to compile a jitclass model object with relevant numerical methods (fit, impulse_response, check_bk, etc).
The
extract_sparse_data_from_modelfunction was causing kalman filter tests to fail because of how I wrote the numba wrappers. I also got it to cause sigfaults, which is obviously bad.The entire method of numbafying the model for estimation needs to be revisited. Preferably implement a function to compile a jitclass model object with relevant numerical methods (fit, impulse_response, check_bk, etc).