nLab
infinitesimal singular simplicial complex

Context

Synthetic differential geometry

differential geometry

synthetic differential geometry

Axiomatics

Models

Concepts

Theorems

Applications

Contents

Idea

The infinitesimal singular simplicial complex of a space XX in a smooth topos (𝒯,R)(\mathcal{T},R) is the infinitesimal analogue of the singular simplicial complex X Δ R X^{\Delta_R^\bullet} (see interval object) that in degree kk is the space of kk-simplices Δ R kX\Delta^k_R \to X in XX: the infinitesimal singular simplicial complex has in degree kk the infinitesimal kk-simplices in XX.

There are several ways to make the notion of “infinitesimal kk-simplex in XX” precise. Here we describe a notion promoted by Anders Kock, where an “infinitesimal kk-simplex” in XX for XX a suitably locally linear space , is a (k+1)(k+1)-tuple (x 0,,x k)X × k+1(x_0,\cdots, x_k) \in X^{\times^{k+1}} of points in XX that are pairwise infinitesimal neighbours in XX.

One central application of the singular simplicial complex is in the definition of differential forms in synthetic differential geometry.

Definition

The basic definition applies to spaces of the form R nR^n and is generalized from there to spaces that “locally look like” R nR^n in one way or other.

Let here and in the following (𝒯,R)(\mathcal{T},R) be a smooth topos.

in R nR^n

Write, as usual

D(n):={(ϵ i)n|ϵ iϵ j=0}R n D(n) := \{(\epsilon_i) \in n | \epsilon_i \epsilon_j = 0\} \hookrightarrow R^n

for the infinitesimal space of first order infinitesimal neighbours of the origin of R nR^n, with its canonical inclusion into R nR^n.

Two elements x,yR nx , y \in R^n are called first order infinitesimal neighbours, denoted x 1yx \sim_1 y, if their difference is in the image of this inclusion.

(x 1y)(ϵD(n):(xy)=ϵ). (x \sim_1 y) \Leftrightarrow (\exists \epsilon \in D(n) : (x-y) = \epsilon) \,.

Write

(R n) Δ inf k:={(x 0R n,,x kR n)|x i 1x j}. (R^n)^{\Delta^k_{inf}} := \{ (x_0 \in R^n, \cdots, x_k \in R^n) | x_i \sim_1 x_j \} \,.

This naturally forms a simplicial object X Δ inf bullet:Δ op𝒯X^{\Delta_{inf}^{bullet}} : \Delta^{op} \to \mathcal{T}. This is the infinitesimal simplicial singular complex of R nR^n.

A more detailed discussion of this is in the entry infinitesimal object in the section Spaces of infinitesimal simplices.

In smooth loci

warning this section is as such not drawn from the literature, it seems

Recall that a smooth locus in 𝒯\mathcal{T} is an object A\ell A that is the joint limit over some

R n0fJR R^n \stackrel{\stackrel{f \in J}{\to}}{\stackrel{0}{\to}} R

here JHom 𝒯(R n,R)J \subset Hom_{\mathcal{T}}(R^n,R) is an ideal.

Declare that two generalized elements x,yAx,y \in \ell A are infinitesimal neighbours if their image under the injection

AR n \ell A \hookrightarrow R^n

is a pair of infinitesimal neighbour in R nR^n. Then let

(A) Δ inf (R n) Δ R (\ell A)^{\Delta^\bullet_{inf}} \hookrightarrow (R^n)^{\Delta_R^\bullet}

be the sub-simplicial object of infinitesimal neighbours in R nR^n that are points in A\ell A.

Observation

(linearity of space of infinitesimal neighbours)

If p,qAp, q \in \ell A are infinitesimal neighbours in the smooth locus AR n\ell A \subset R^n, then for all tRt \in R also the element p+t(qp)p + t(q-p) formed by linear combination in R nR^n is in A\ell A and hence is an infinitesimal neighbour of pp there.

Proof

Because by the Kock-Lawvere axiom valid in the smooth topos (𝒯,R)(\mathcal{T},R) we have for all f:R nRf : R^n \to R

0=f(q)=f(p+(qp))=f(p)+ i(qp) i( if)(q)= i(qp) i( if)(q). 0 = f(q) = f(p + (q-p)) = f(p) + \sum_i (q-p)_i (\partial_i f)(q) = \sum_i (q-p)_i (\partial_i f)(q) \,.

Therefore also

f(p+t(qq))=t i(qp) i( if)(q)=0. f(p + t(q-q)) = t \sum_i (q-p)_i (\partial_i f)(q) = 0 \,.

Examples

Consider the circle S 1S^1 regarded as the smooth locus S 1={(x,y)R 2|x 2+y 2=1}S^1 = \{(x, y) \in R^2 | x^2 + y^2 = 1\}.

For aS 1R 2a \in S^1 \subset R^2 an infinitesimal neighbour (a+ϵ)(a + \epsilon) in R 2R^2 is again a point on the circle, and hence an infinitesimal neighbour of aa in S 1S^1, if

a x 2+2a xϵ x+a y 2+2a yϵ y=1 a_x^2 + 2 a_x \epsilon_x + a_y^2 + 2 a_y \epsilon_y = 1

which, due to a x 2+a y 2=1a_x^2 + a_y^2 = 1 is equivalent to

2a xϵ x+2a yϵ y=0. 2 a_x \epsilon_x + 2 a_y \epsilon_y = 0 \,.

This is solved by ϵ\epsilon of the form

ϵ=δ(a y a x) \epsilon = \delta \left( \array{ a_y \\ - a_x } \right)

for some fixed δD\delta \in D.

In formal manifolds

use that each manifold is locally isomorphic to an R nR^n and that the neighbourhood relation only needs an infinitesimal neighbourhood. Proceed locally as above and then patch. See references below.

Inclusion into the finite singular simplicial complex

The lined topos (𝒯,R)(\mathcal{T}, R) also comes canonically for every object X𝒯X \in \mathcal{T} with the finite singular simplicial complex Π(X):[n]X R n\Pi(X) : [n] \mapsto X^{R^n} induced from regarding

(0 *,1 *):**R (0_*, 1_*) : {*} \coprod {*} \to R

as an interval object (see there for details).

Definition

(inclusion of infinitesimal into finite simplices)

For A=:XR n\ell A =: X \hookrightarrow R^n a smooth locus define for all nn \in \mathbb{N} a morphism

ι n:X Δ inf kX Δ R k=X R k \iota_n : X^{\Delta^k_{inf}} \to X^{\Delta_R^k} = X^{R^k}

by defining it on generalized elements as

(x 0,,x k)(tx 0+ i=1 kt i(x ix i1)). (x^0, \cdots, x^{k}) \mapsto (\vec t \mapsto x^0 + \sum_{i=1}^{k} t_i (x^i - x^{i-1}) ) \,.
Proposition

The morphisms ι n\iota_n constitute a morphism of simplicial objects

ι:X Δ inf X Δ R k \iota : X^{\Delta^\bullet_{inf}} \hookrightarrow X^{\Delta_R^k}

in that they respects the face and degenracy maps on each side.

Proof

Straightforward checking:

For instance

  • The inner face maps d id_i on X Δ inf kX^{\Delta_{inf}^{k}} omit the iith point in the (k+1)(k+1)-tuple of points, while on X Δ kX^{\Delta^{k}} they act by pullback along (t 1,,t k)(t 1,,t i1,t i,t i,t i+1,cdots,t k)(t_1, \cdots, t_k) \mapsto (t_1, \cdots, t_{i-1}, t_{i}, t_i, t_{i+1}, cdots, t_k). That means that in the sum above t it_i appears twice to yield

    +t i(x ix i1)+t i(x i+1x i)+=+t i(x i+1x i1)+ \cdots + t_i(x^i - x^{i-1}) + t_i(x^{i+1} - x^i) + \cdots = \cdots + t_i(x^{i+1} - x^{i-1} ) + \cdots

    which indeed corresponds to omission of the iith point x ix^i.

The collection of first order infinitesimal neighbours of a space XX arranges itself into the infinitesimal path ∞-groupoid? Π inf(X)\Pi^{inf}(X). Various concepts derive from this one:

of differential forms may be understood in terms of functions on Π(x) inf\Pi(x)^{inf}. This is described at

A deRham space is the colimit over a Π inf(X)\Pi^{inf}(X).

References

In the language of synthetic differential geometry the infinitesimal singular complex for “formal manifolds” (internally defined manifolds with an infinitesimal thickening to all orderes) is described (with the simplicial structure not made explicit) in

section I.18 of

and in section 2.8 of

  • Anders Kock, Synthetic geometry of manifolds (pdf)

Discussion of this that does make the simplicial structure explicit and relates it to the Dold-Kan correspondence is in

The details of what X Δ inf kX^{\Delta^k_{inf}} is like concretely on representables in the smooth topos PSh(kAlg op)PSh(k-Alg^{op}) of algebraic geometry, i.e. on affine schemes is worked out in detail in

The formulas given there should more or less directly carry over to smooth toposes with smooth loci by replacing ordinary rings with smooth algebras.

to be discussed

As the title suggests, the infinitesimal singular simplicial complex is tightly related to differential forms in synthetic differential geometry: the deRham complex is the normalized Moore cochain complex of the cosimplicial algebra C (X Δ inf )C^\infty(X^{\Delta^\bullet_{inf}}) of functions on the spaces of infinitesimal simplices.

There is also

  • Dubuc, Kock, On 1-form classifiers , Communications in Algebra 12 (1984)

  • Dubuc, C C^\infty-schemes, Amer. J. of Math. 103 (1981)

  • Kumpera, Spencer, Lie Equations , Annals of Math. Studies 73 (1973)

There is also a version of the infinitesimal singular simplicial context in the context of nonstandard analysis. See

  • Zivaljevic, On a cohomology theory based on hyperfinite sums of microcomplexes .

Revised on October 29, 2013 14:32:03 by Urs Schreiber (82.169.114.243)