Splitting lemma (functions)

In mathematics, especially in singularity theory, the splitting lemma is a useful result due to René Thom which provides a way of simplifying the local expression of a function usually applied in a neighbourhood of a degenerate critical point.

Let ${\displaystyle f:(\mathbb {R} ^{n},0)\to (\mathbb {R} ,0)}$ be a smooth function germ, with a critical point at 0 (so ${\displaystyle (\partial f/\partial x_{i})(0)=0}$ for ${\displaystyle i=1,\dots ,n}$). Let V be a subspace of ${\displaystyle \mathbb {R} ^{n}}$ such that the restriction f |_V is non-degenerate, and write B for the Hessian matrix of this restriction. Let W be any complementary subspace to V. Then there is a change of coordinates ${\displaystyle \Phi (x,y)}$ of the form ${\displaystyle \Phi (x,y)=(\phi (x,y),y)}$ with ${\displaystyle x\in V,y\in W}$, and a smooth function h on W such that

${\displaystyle f\circ \Phi (x,y)={\frac {1}{2}}x^{T}Bx+h(y).}$

This result is often referred to as the parametrized Morse lemma, which can be seen by viewing y as the parameter. It is the gradient version of the implicit function theorem.

There are extensions to infinite dimensions, to complex analytic functions, to functions invariant under the action of a compact group, ...