Unfolding (functions)

In mathematics, an unfolding of a function is a certain family of functions.

Let M be a smooth manifold and consider a smooth mapping f : M \to \mathbb{R}. Let us assume that for given x_0 \in M and y_0 \in \mathbb{R} we have f(x_0) = y_0 . Let N be a smooth k-dimensional manifold, and consider the family of mapping (parameterised by N) given by F : M \times N \to \mathbb{R} . We say that F is a k-parameter unfolding of f if F(x,0) = f(x) for all x. In other words the functions f : M \to \mathbb{R} and F : M \times \{0\} \to \mathbb{R} are the same: the function f is contained in, or is unfolded by, the family F.

Let f : \mathbb{R}^2 \to \mathbb{R} be given by f(x,y) = x^2 + y^5. An example of an unfolding of f would be F : \mathbb{R}^2 \times \mathbb{R}^3 \to \mathbb{R} given by

F((x,y),(a,b,c)) = x^2 + y^5 + ay + by^2 + cy^3.

As is the case with unfoldings, x and y are called variables and a, b, and c are called parameters - since they parameterise the unfolding.

In practice we require that the unfoldings have certain nice properties. In \mathbb{R} notice that f is a smooth mapping from M to \mathbb{R} and so belongs to the function space C^{\infty}(M,\mathbb{R}). As we vary the parameters of the unfolding we get different elements of the function space. Thus, the unfolding induces a function \Phi : N \to C^{\infty}(M,\mathbb{R}). The space \mbox{diff}(M) \times \mbox{diff}(\mathbb{R}), where \mbox{diff}(M) denotes the group of diffeomorphisms of M etc., acts on C^{\infty}(M,\mathbb{R}). The action is given by (\phi,\psi) \cdot f = \psi \circ f \circ \phi^{-1}. If g lies in the orbit of f under this action then there is a diffeomorphic change of coordinates in M and \mathbb{R} which takes g to f (and vice versa). One nice property that we may like to impose is that

\mbox{Im}(\Phi) \pitchfork \mbox{orb}(f)

where "\pitchfork" denotes "transverse to". This property ensures that as we vary the unfolding parameters we can predict - by knowing how the orbit foliate C^{\infty}(M,\mathbb{R}) - how the resulting functions will vary.

There is an idea of a versal unfolding. Every versal unfolding has the property that \mbox{Im}(\Phi) \pitchfork \mbox{orb}(f) , but the converse is false. Let x_1,\ldots,x_n be local coordinates on M, and let \mathcal{O}(x_1,\ldots,x_n) denote the ring of smooth functions. We define the Jacobian ideal of f, denoted by J_f as follows:

J_f := \left\langle \frac{\partial f}{\partial x_1}, \ldots, \frac{\partial f}{\partial x_n} \right\rangle.

Then a basis for a versal unfolding of f is given by quotient

\frac{\mathcal{O}(x_1,\ldots,x_n)}{J_f}

This quotient is known as the local algebra of f. The dimension of the local algebra is called the Milnor number of f. The minimum number of unfolding parameters for a versal unfolding is equal to the Milnor number; that is not to say that every unfolding with that many parameters will be versal! Consider the function f(x,y) = x^2 + y^5. A calculation shows that

\frac{\mathcal{O}(x,y)}{\langle 2x, 5y^4 \rangle} = \{y,y^2,y^3\} \ .

This means that \{y,y^2,y^3\} give a basis for a versal unfolding, and that

F((x,y),(a,b,c)) = x^2 + y^5 + ay + by^2 + cy^3

is a versal unfolding. A versal unfolding with the minimum possible number of unfolding parameters is called a miniversal unfolding.

Sometimes unfoldings are called deformations, versal unfoldings are called versal deformations, etc.

An important object associated to an unfolding is its bifurcation set. This set lives in the parameter space of the unfolding, and gives all parameter values for which the resulting function has degenerate singularities.

References