Zulip Chat Archive
Stream: Is there code for X?
Topic: post composition for continuous bilinear maps
Jireh Loreaux (Feb 07 2025 at 21:41):
Suppose E
, F
, G
and H
are normed spaces over a nontrivially normed field đ
. If I have a continous bilinear map fâ : E âL[đ] F âL[đ] G
and a continuous linear map fâ : G âL[đ] H
, I would like to compose them to obtain a continuous bilinear map E âL[đ] F âL[đ] H
. Outside of normed spaces, I think this is not possible, but in a normed space, we can curry the bilinear map to a multilinear map (E Ă F) [Ă2]âL[đ] G
(I'm fudging the notation here), then compose, and then uncurry.
But presumably I shouldn't actually do this with docs#ContinuousLinearMap.uncurryLeft and the like. Where is the API for what I want to do?
Aaron Liu (Feb 07 2025 at 21:53):
Could you use docs#ContinuousLinearMap.postcomp twice?
Aaron Liu (Feb 07 2025 at 22:04):
import Mathlib
example {đ E F G H : Type*} [NormedField đ]
[NormedAddCommGroup E] [NormedSpace đ E]
[NormedAddCommGroup F] [NormedSpace đ F]
[NormedAddCommGroup G] [NormedSpace đ G]
[NormedAddCommGroup H] [NormedSpace đ H]
(fâ : E âL[đ] F âL[đ] G) (fâ : G âL[đ] H) :
E âL[đ] F âL[đ] H :=
(fâ.postcomp F).comp fâ
Jireh Loreaux (Feb 07 2025 at 22:09):
aha, thanks. It seems my claim about not working outside of normed spaces is probably bogus then. :laughing:
Eric Wieser (Feb 08 2025 at 06:28):
Should this name be standardized? I think the one for LinearMap is named something different.
Eric Wieser (Feb 08 2025 at 06:29):
This operation is compr\_2 there
Last updated: May 02 2025 at 03:31 UTC