Default	Effects
m	5% of the number of cells, with a mini- mum of 50 control points	Too small: the approximation of the vector field in RKHS is too sparse (underfitting); Too large: the optimization of the loss function is memory- and time-consuming.
λ	3	Too small: overfitting; Too large: underfitting.
w	determined by the distribution of the data (see below)	Too small: large bandwidth means all control points have approximately equal contributions to all surrounding states in the vector field, and the vector field becomes linear; Too large: small bandwidth means control points have insufficient influence over distant states, result- ing in zero velocities evaluated for distant cells.
Nmax	500	Too small: the algorithm is terminated before reasonable convergence (underfitting); Too large: when convergence is hard to achieve, the algorithm takes too long with negligible improve- ments in minimizing the loss function.