Vectorized the integrators for batched calls . 2-4 x performance improvement when integrating multiples states and their derivatives. Also implemented a new central shooting defect that can take advantage of vectorized calls. Results in 2-4x runtime improvement for the CentralShooting transcription.
Vectorized the integrators for batched calls . 2-4 x performance improvement when integrating multiples states and their derivatives. Also implemented a new central shooting defect that can take advantage of vectorized calls. Results in 2-4x runtime improvement for the CentralShooting transcription.