The Jacobian Determinant

The Jacobian determinant and the determinant of a transformation matrix are very closely related concepts. The key difference lies in whether the transformation is linear or nonlinear.

A Jacobian determinant is the determinant of a transformation matrix, specifically the one that provides the best local linear approximation of a nonlinear function at a specific point.

The Key Difference: Global vs. Local Scaling

Determinant of a Transformation Matrix (Linear Functions)

Your understanding is perfectly correct here. For a linear transformation represented by a matrix $A$, such that $y=Ax$:

• The transformation is the same everywhere in the space.
• The determinant, $\det(A)$, is a single, constant number that gives the global scaling factor for volume. A unit cube in the input space is transformed into a parallelepiped with a volume of $|\det(A)|$ no matter where that cube is located.

Jacobian Determinant (Nonlinear Functions)

For a nonlinear transformation $y=f(x)$, we can’t use a single matrix to describe the entire transformation.

• The Jacobian matrix, $J$, is the matrix of partial derivatives that provides the best linear approximation of $f$ at a specific input point $x_0$.
• The Jacobian determinant, $\det(J(x_{0}))$, gives the scaling factor for volume only at that specific point $x_0$.
• Crucially, the value of the Jacobian determinant changes depending on where you are in the space. A nonlinear function might stretch space in one region ($|\det(J)| > 1$) and compress it in another ($|\det(J)| < 1$).

Connections

• Forward Links: Transformations of Probability Densities
• Backward Links: Determinant, Gradients of Vector-Valued Functions