You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
What was the design decision behind separating evaluation and the Jacobian function:
/// Get the curve Jacobians.
/// This is the main interface for GTSAM
virtual void getJacobians(unsigned derivativeOrder, ...
I ask because in the case of GP's, my jacobians essentially double as the interpolation coefficients (similar to alpha and 1-alpha in the case of linear interpolation). I know Paul mentioned previously that it is not good to store the Jacobian's because they take too much memory, so I am doing them functionally, but at the moment I have to do it twice (once for evaluation and once for getJacobians).
The text was updated successfully, but these errors were encountered:
What was the design decision behind separating evaluation and the Jacobian function:
I ask because in the case of GP's, my jacobians essentially double as the interpolation coefficients (similar to alpha and 1-alpha in the case of linear interpolation). I know Paul mentioned previously that it is not good to store the Jacobian's because they take too much memory, so I am doing them functionally, but at the moment I have to do it twice (once for evaluation and once for getJacobians).
The text was updated successfully, but these errors were encountered: