Definition

Let MM and NN be without boundary, and let f0,f1:MNf_0,f_1:M\to N be . A smooth homotopy from f0f_0 to f1f_1 is a

F:M×[0,1]NF:M\times[0,1]\longrightarrow N

such that F(x,0)=f0(x)F(x,0)=f_0(x) and F(x,1)=f1(x)F(x,1)=f_1(x) for every xMx\in M. Here the M×[0,1]M\times[0,1] has its standard structure, and smoothness includes smooth extension in the interval direction at the endpoints. Smooth homotopy is an on smooth maps and is the differential-category refinement of continuous homotopy.

Homotopy formula for differential forms

Write Ft(x)=F(x,t)F_t(x)=F(x,t). Integration in the interval direction defines an operator K:Ωk(N)Ωk1(M)K:\Omega^k(N)\to\Omega^{k-1}(M) satisfying

F1F0=dK+Kd.F_1^*-F_0^*=dK+Kd.

Thus the two pullbacks of are cochain-homotopic and induce the same map on . This is the smooth homotopy-invariance mechanism used throughout de Rham theory Tu, chapter on de Rham theory.

Examples and distinctions

If f:MNf:M\to N is constant along a smooth path in NN, that path gives a smooth homotopy between the corresponding constant maps. A is stronger than a smooth homotopy: it requires each time slice to remain in a specified class, commonly diffeomorphisms or embeddings. A general smooth homotopy imposes no such condition on the slices.

Conventions and scope

Some authors require a homotopy to be constant for tt near 00 and 11, especially when homotopies will be concatenated smoothly. That is an additional collar convention, not part of the basic definition. If MM itself has boundary, M×[0,1]M\times[0,1] naturally has corners, so the ambient smooth category must be enlarged or the definition phrased using smooth extension on a neighborhood.

References
  1. Loring W. Tu, An Introduction to Manifolds, 2nd ed., Springer, 2011. DOI record. Relevant: smooth homotopy and the homotopy formula in de Rham theory.
  2. Raoul Bott and Loring W. Tu, Differential Forms in Algebraic Topology, Springer, 1982. DOI record. Relevant: Chapter I, homotopy operators and de Rham theory.