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
It may be possible to get a speed up from deterministic projections in big landscapes by using sparse matrix classes
For a 3-stage, 2000 patch dispersal model, switching from matrix to Matrix (and adding import & dealing with classes) here doesn't slow down pop:::as.matrix.dynamic() much (7.3s vs 3.9s) and results in much faster multiplications (1.2s vs 5.3s).
This only skips true 0s, so is as accurate as the dense matrix version. If the matrix construction end can be sped up, this may be worthwhile in all dispersal cases.
The text was updated successfully, but these errors were encountered:
@skiptonium definitely worth thinking about sparse matrices for dlmpr. should be easy enough to switch over to SpMat class once the whole thing is prototyped
It may be possible to get a speed up from deterministic projections in big landscapes by using sparse matrix classes
For a 3-stage, 2000 patch dispersal model, switching from
matrix
toMatrix
(and adding import & dealing with classes) here doesn't slow downpop:::as.matrix.dynamic()
much (7.3s vs 3.9s) and results in much faster multiplications (1.2s vs 5.3s).This only skips true 0s, so is as accurate as the dense matrix version. If the matrix construction end can be sped up, this may be worthwhile in all dispersal cases.
The text was updated successfully, but these errors were encountered: