Closed favonia closed 6 years ago
Thanks for the Thought, I agree. A couple comments:
I think I know how to mostly implement (0), by replacing recursive calls to run
with one of those explicit re-indexings M_\Xi
or whatever.
I am not sure how to achieve optimization (1). Can you unleash some thoughts about that?
2. I am not sure how to achieve optimization (1). Can you unleash some thoughts about that?
When working with faces, if the cofibration ri = ri'
is already true (which could happen for boundaries of neutral values), drop phi
(and do whatever neccesary).
Implemented some optimization 1 in #270 by 8eff62b.
@favonia Why is it justified to drop phi
in that case? Do we somehow know that if ri = ri'
, then the restriction is already stronger than phi
? I don't quite understand.
@favonia Why is it justified to drop
phi
in that case? Do we somehow know that ifri = ri'
, then the restriction is already stronger thanphi
? I don't quite understand.
You're right. I'll fix that.
Done it the current PR #270.
@jonsterling There are two optimizations of act/run I regard as essential in the new domain:
run phi2
should be completely "dropped" inrun phi1 @@ run phi2 m
.m
is already aphi
-value,run phi m
should be a no-op as much as possible.I know how to do optimization 1 when working with faces, and I wonder if optimization 0 was already implemented in #270?