Michi’s blog » Blog Archive » Coordinatization with hom complexes

Coordinatization with hom complexes

December 7th, 2009

These are notes from a talk given at the Stanford applied topology seminar by Gunnar Carlsson from 9 Oct 2009. The main function of this blog post is to get me an easily accessible point of access for the ideas in that talk.

Coordinatization

First off, a few words on what we mean by coordinatization: as in algebraic geometry, we say that a coordinate function is some $X\to\mathbb R$ or possibly some $X\to\mathbb C$ , with all the niceness properties we’d expect to see in the context we’re working.

A particularly good example is Principal Component Analysis which yields a split linear automorphism on the ambient space that maximizes spread of the data points in the initial coordinates.

Topological coordinatization

The core question we’re working with right now is this:
Given a space (point cloud) X, and a (persistent) view of H_*(X) , can we use some map $H_*(X)\to H_*(Y)$ to generate a map $X\to Y$ inducing that map?

In topology there are ways around to compute spaces of maps directly from available homological information. The key phrase here is the Eilenberg-Moore spectral sequence, by which we compute $H^*(\Omega X)$ using H^*(X) . We note that $\Omega X = Top_*(S^1,X)$ with the compact open topology.

We fix a field k of coefficients. We can find a spectral sequence that computes
$Tor_{H^*(X)}(k,k)’ \Rightarrow H^*(\Omega X)$

Variations of this spectral sequence allow us to compute H^*(Map(X,Y)) . A homotopy class of maps thus corresponds to a cohomology element from H^*(F(X,Y)) .

Monad approximation

Define
$Sp^2(X) = X \times_{\mathbb Z/2} X$ (unordered pairs)
and similarily unordered tuples
$Sp^n(X) = X^n/\Sigma_n$
we can get a map, for pointed spaces (X,*)
$Sp^n(X) \to Sp^{n+1}(X)$
$(x_1,\dots,x_n)\mapsto(x_1,\dots,x_n,*)$

In the limit, we get the union $Sp^\infty(X)$ .

The Dold-Thom theorem tells us:
$\pi_i(Sp^\infty(X)) = H_i(X)$

Suppose we are really looking for maps $Z\to Sp^\infty(X,*) \supseteq X = Sp^1(X)$ .

We pick chain complexes C_*, D_* . There is a hom chain complex Hom(C_*,D_*) consisting of graded maps $C_*\to D_*$ with the homotopy/induced boundary
$\Delta f = \partial_Df\pm f\partial_C$

For this complex, we get
H_0(Hom(C_*,D_*)) is the family of chain homotopy classes of chain maps $C_*\to D_*$ .

Recall: f,g are chain homotopic if there is a family $h_i: C_i\to D_{i+1}$ such that $f-g = \partial_D h_i + h_{i-1}\partial_C$ .

Homotopic space maps yield homotopic chain complex maps. And thus
$Top_*(Z,Sp^\infty(X)) = H_*(Hom(C_*,D_*),\Delta)$

How, though, do we get from $[Z,Sp^\infty(X)]$ to [Z,X] ?

Cosimplicial spaces

A simplicial space is a contravariant $\Delta\to X$ . It has a total space or realization given by
$\coprod X_n\times\Delta[n]/\sim$

In the dual setting of cosimplicial spaces, the total space is a subspace of $X^0\times(X^1)^{\Delta[1]}\times(X^2)^{\Delta[2]}\times(X^3)^{\Delta[3]}\times\dots$ .

There is a cosimplicial space
$Sp^\infty(X) \leftrightarrow Sp^\infty Sp^\infty(X) \leftrightarrow \dots$
whose total space is .

Furthermore, there is a spectral sequence starting at $H_*(Sp^\infty(X))$ (well understood space!)
going through
$\pi_*(Top_*(Z,Sp^\infty Sp^\infty Sp^\infty X)) \leftrightarrow \pi_*(Top_*(Z,Sp^\infty Sp^\infty X)) \leftrightarrow \pi_*(Top_*(Z,Sp^\infty X))$
which recovers $\pi_*(Top_*(Z,X))$ for us.

The mapping space Top_*(Z,X) maps to $Top_*(Z,Sp^\infty X)$ , which sends H_*(Hom(C_*Z,C_*X)) to chain maps. Differentials give us conditions on chain maps to come from actual maps on the topological level $Z\to X$ .

Suppose now that is a simplicial complex, and H_*X is known, say isomorphic to H_*S^1 .

We pick S^1 as a model space, and work in Hom(C_*X,C_*S^1) . We pick out conditions form the spectral sequence above to pick out chain maps that might come from topological maps.

Suppose we had some chain map $C_*X\overset{\varphi}{\to}C_*S^1$ that sends $\sigma\mapsto\sum_{i=0}^n\eta_i$ .

We look for a chain homotopic map $\hat\varphi$ such that $\hat\varphi(\sigma)$ is a simplex (this will be WAY too optimistic, usually) or at least $\hat\varphi(\sigma) = \sum_{i=0}^n\eta_i$ for small and $\eta_i$ close together in graph distances.

Yi Ding: there are likely many simplicial maps $X\to S^1$ inducing any given $H_*X\to H_*S^1$ . This makes optimizations on quality of maps (as measured in smallness properties of preimages of given simplices) troublesome, since we have a large search space and are likely to get stuck in local minima.

The topological approach creates very many maps, but we just want ONE – albeit with quality assurances. The big question is how to get there:

Shrinking the space? Homotopy collapses to reduce the number of generators?

Ask for more side conditions? If so, which?

Minimize the size of preimages and kernels – smoothing the resulting map?

Run with one condition until we reach a minimum, then change conditions to get out of that particular sink, repeat until it all looks stable?

Thus we need to search for strategies to improve convergence to actually good maps.

The questions we pose are similar to:

Jeff Ericksson, et al.: Given X, $x\in H_n(X)$ , how do we find a good representative $\xi\in x$ ?

Finding good maps is really about finding good representatives in Hom(C_*X,C_*Y) . Given simplicial bases $\{\sigma_i\}\subseteq X, \{\eta_i\}\subseteq Y$ we can pick out a first basis
$\varphi_{\sigma_i\eta_j}(\sigma_k)=\delta_{ik}\eta_j$ (Kronecker $\delta$ ) for .

Finally: many problems in data analysis call for analyzing contractible spaces. So if we pick out the notion of a boundary, doing all of this relative to a boundary would give us increased power to use the resulting methods in data analysis.

2 People had this to say...

Andres Angel Said...

Dold-Thom theorem?… interesting stuff…

December 7th, 2009 at 23:57

Michi Said...

Oh look. You’re right. *edits*

December 8th, 2009 at 0:10

Want your say?

post navigation

about: Michi is a recent PhD working in homological algebra and applied algebraic topology. This blog is his outlet for texts with some manner of thought put into them. Over at his LiveJournal intimate details and streams of consciousness might be found.
Not all here is mathematics. All here, though, are my personal thoughts and opinions. Please read the about page (linked above) for more details.
This blog uses statcounter.com for logging and traffic analysis. In order to identify return visitors, this site will issue a cookie on viewing the blog.

Travel plans

Recent Comments

Tags: MATH198

Categories

Blogroll

Family

Mathematician blogs

Archives