I come to you, yet again, with another question regarding the proof of the homotopy-type invariance of simplicial homology groups. As an intermediate step, Armstrong's "Basic Topology" proposes the following problem:

Let $f,g:|K|\to|L|$ be continuous maps, such that the sets $f^{-1}($star$(v,L))\cap g^{-1}($star$(v,L))$ with $v$ a vertex of $L$ form an open covering of $|K|$. Find an integer $m$ and a simplicial map $s:|K^m|\to|L|$ which simplicially approximates both $f$ and $g$.

Here are the definitions involved:

  • By $|K|$ we denote the underlying space of the simplicial complex $K$, i.e., the set of points which belong to some simplex of $K$.
  • The open star of a vertex $v$ in $K$, represented by star$(v,K)$, is the set of points which are interior to some simplex of $K$ that has $v$ as a vertex.
  • A simplicial map is a map which takes simplexes linearly into simplexes, i.e., given a simplex $(v_0,\dots, v_n)$, we have:
    1. $(s(v_0),\dots,s(v_n))$ is also a simplex (not necessarily of the same dimension)
    2. if $x=\sum_{i=0}^n\lambda_iv_i$ with $\lambda_i\geqslant 0$ for each $i$ and $\sum_{i=0}^n\lambda_i=1$ then $s(x)=\sum_{i=0}^n\lambda_is(v_i)$
  • A simplicial approximation of a given continuous map $f:|K|\to|L|$ is a simplicial map $s:|K|\to|L|$ such that $s(x)$ belongs to the only simplex which contains $f(x)$ in its interior.
  • The barycentric subdivision of a simplicial complex is explained in more detailed in one of my other questions.
  • The Simplicial Approximation Theorem guarantees the existence of a simplicial approximation for any continuous map if we take a big enough barycentric subdivision of the simplicial complex of the domain.

I thought maybe the idea was to associate to each vertex of $v_0\in K$ the vertex $v_1\in L$ such that $v_0\in f^{-1}($star$(v_1,L))\cap g^{-1}($star$(v_1,L))$. However, I don't know which barycentric subdivision to take in order for this to be a simplicial map or a simplicial approximation.

  • $\begingroup$ This is a nice problem, so it would be good to give definitions of these terms for posterity in this post. $\endgroup$ – John Samples Jun 14 '18 at 23:31
  • $\begingroup$ Thanks, @JohnSamples ! I've tried to add the definitions of the main concepts involved. If you have any other suggestion or feel I have omitted anything, please let me know. I'll also try to add more details to my answer shortly. $\endgroup$ – Hercule Poirot Jun 15 '18 at 10:50

I think I got it figured now. Here's my proof, in case it's helpful to someone else:

Take the Lebesgue number $\delta$ of the open covering. We know the mesh of the barycentric subdivisions (the diameter of the biggest simplex) goes to zero as $n$ grows: $\mu(K^n)\overset{n\to\infty}{\longrightarrow}0$. Therefore, we can take $m\in\mathbb{N}$ such that $\mu(K^m)<\frac{\delta}{2}$.

That way, if we take a vertex $u\in K^m$, we have diam$($star$(u,K^m))<2\frac{\delta}{2}=\delta$ and there is a vertex $v_1\in L$ such that star$(u,K^m)\subseteq f^{-1}($star$(v_1,L))\cap g^{-1}($star$(v_1,L))$.

That way, the inclusions $f($star$(u,K^m))\subseteq $star$(v_1,L)$ and $g($star$(u,K^m))\subseteq $star$(v_1,L)$ are true. We take $s(u)=v_1$ and the proof of the Simplicial Appoximation Theorem lets us extend $s:|K^m|\to |L|$ such that it is a simplicial map and a simplicial approximation of both $f$ and $g$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.