Yuli Levtov

I'm not sure if I'm misunderstanding how jit.rota works, or if this is a bug (probably the former), but could someone please explain why the second matrix here is not filled with 0 1 2 3 0 1 2 3 as I would expect?

jit-rota-boundmode-misunderstanding

Not a bug. Check what happens at the output of jit.rota: with scale 0.5, you loose half the data (0 2 remain, 1 3 are dropped). Then, you upsample with your jit.matrix @adapt 0.

Ah OK - understood. Got it now by putting an intermediary matrix in the middle which is equal to the target size. Attached for posterity.

jit.rota boundmode misunderstanding?