Statement

Let M,P,NM,P,N be finite-dimensional, second-countable without boundary, let SNS\subseteq N be an , and let F:M×PNF:M\times P\to N be a . Write Fp(x)=F(x,p)F_p(x)=F(x,p). The parametric transversality theorem states that if FF is , then FpF_p is transverse to SS for almost every pPp\in P. Equivalently, the exceptional parameters form a measure-zero subset in every coordinate chart of PP. In particular, every nonempty open subset of PP contains a parameter whose slice is transverse.

Reduction to Sard's theorem

Transversality makes Z=F1(S)Z=F^{-1}(S) an embedded submanifold of the M×PM\times P. For the restricted projection π:ZP\pi:Z\to P, a parameter pp is a regular value exactly when FpSF_p\pitchfork S. The conclusion therefore follows by applying to π\pi. This reduction is presented in Guillemin–Pollack, Chapter 2.

Consequences and examples

If S={y}S=\{y\}, the theorem says that almost every member of a family is transverse to the point yy, hence has yy as a . Translation families in give a standard application: once the total evaluation map is a submersion, almost every translate meets a fixed submanifold transversely.

The theorem is a principal device for turning an adjustable finite-dimensional parameter into a generic geometric position.

Conventions and scope
References
  1. Victor Guillemin and Alan Pollack, Differential Topology, AMS Chelsea Publishing, 2010 reprint. DOI record. Relevant: Chapter 2, the transversality theorem with parameters.
  2. Morris W. Hirsch, Differential Topology, Graduate Texts in Mathematics 33, Springer, 1976. DOI record. Relevant: Chapter 3, parametric transversality and Sard's theorem.