Zulip Chat Archive

Stream: Is there code for X?

Topic: extending a finitely support function over a sigma?


Kim Morrison (Nov 21 2025 at 23:19):

Do we already have

import Mathlib

def Finsupp.embSigma {k : κ} (f : ι k →₀ R) : (Σ k, ι k) →₀ R :=
  Finsupp.embDomain (Function.Embedding.sigmaMk k) f

somewhere, and hopefully its basic API?

Eric Wieser (Nov 22 2025 at 01:33):

I'd probably claim we don't need it, since the spelling is already pretty short:

import Mathlib

-- your example isn't a mwe without these
variable {κ} {ι : κ  Type*} [Zero R]

noncomputable def Finsupp.embSigma {k : κ} (f : ι k →₀ R) : (Σ k, ι k) →₀ R :=
  f.embDomain <| .sigmaMk k

Eric Wieser (Nov 22 2025 at 01:33):

The argument that it's not needed is even more compelling if it's argued that your function should be called embDomainSigmaMk for clarity

Aaron Liu (Nov 22 2025 at 01:35):

Why do we need this? Does this combination come up a lot?

Kim Morrison (Nov 22 2025 at 02:25):

I'm using this in a proposed formalisation of abstract convex spaces, where it seems to work nicely.

Kim Morrison (Nov 22 2025 at 02:25):

I've made #31923 in the meantime.

Kim Morrison (Nov 22 2025 at 02:26):

I can try inlining the definition again there and see if it adds any pain.


Last updated: Dec 20 2025 at 21:32 UTC