research36 min studyLesson 833 in Mathematics

Lipschitz maps and the area formula

Calculate geometric Jacobians and retain the multiplicity of a folded parametrization.

Before you begin

On this page

The question for today

What you’ll understand

State the area formula for a Lipschitz map with source dimension no larger than target dimension.
Compute an m-dimensional Jacobian from the derivative Gram matrix.
Distinguish unsigned multiplicity-weighted area from image measure and oriented cancellation.

A parametrization can stretch a surface and can also cover the same point more than once. Integrating its Jacobian measures both effects, so it need not equal the unweighted measure of the image. The area formula makes the missing multiplicity explicit. A folded interval gives a complete calculation, while a planar patch mapped into three dimensions shows why the correct Jacobian is a Gram determinant rather than an arbitrarily chosen square minor.

Differentiability almost everywhere supplies a Jacobian

Let f map Euclidean m-space to Euclidean n-space with m no larger than n, and assume it is Lipschitz. Rademacher's theorem supplies a derivative at Lebesgue-almost every source point. Define the m-dimensional Jacobian as the square root of the determinant of the derivative's Gram matrix. It vanishes when the derivative has rank below m. Values assigned to the derivative on its null exceptional set do not affect the source integral.

When source and target dimensions coincide, the formula becomes the absolute value of the ordinary determinant. For a curve, it becomes speed. For a graph or immersed patch, it measures the volume spanned by its derivative columns in the higher-dimensional target. A single coordinate projection can underestimate that volume, and orientation signs do not belong in this unsigned Jacobian. These distinctions remain necessary even for linear maps.

J_m f(x)=\sqrt{\det(Df(x)^{\mathsf T}Df(x))},\qquad m\le n.

The m derivative columns determine the Gram matrix; its determinant measures squared m-volume.↔ Scroll wide equations horizontally.

The image integral counts every preimage

For a Lebesgue-measurable source set A, the area formula equates the integral of this Jacobian with the target Hausdorff integral of the number of preimages lying in A. The target uses normalized m-dimensional Hausdorff measure, even when it sits in a larger ambient space. Fibers may be finite or infinite, and their counting function is interpreted as a nonnegative extended quantity. The equality remains meaningful when either side is infinite.

A weighted version sums a nonnegative measurable source weight over the preimages. Signed weights require absolute integrability before subtracting positive and negative parts. If f is injective on the source set, multiplicity is one on the image and the formula simplifies to geometric image area. Without injectivity, replacing the counting function by one discards overlapped sheets and generally changes the answer.

\int_AJ_mf\,d\mathcal L^m=\int_{\mathbb R^n}\#(A\cap f^{-1}(y))\,d\mathcal H^m(y).

The right side is multiplicity-weighted image area, not merely the measure of the image set.↔ Scroll wide equations horizontally.

Orientation introduces a different kind of bookkeeping

The area formula counts unsigned coverage. If one branch travels forward and another backward over the same target segment, their areas add. An oriented current can instead cancel those contributions when it is pushed forward. Keeping these two operations separate is essential later: a folded curve can have positive parametrized length but represent the zero oriented current after the two sheets cancel.

For an injective affine map, there is no folding to count, and a constant Jacobian factors out of the integral. Shear can preserve planar area while changing edge lengths and shape, showing why visual distortion alone does not determine measure change. Conversely, a higher-dimensional immersion may have a Jacobian involving several components at once. Computing its Gram matrix gives a coordinate-independent check and provides an exact benchmark for the general theorem.

\int_Aw(x)J_mf(x)\,dx=\int_{\mathbb R^n}\sum_{x\in A\cap f^{-1}(y)}w(x)\,d\mathcal H^m(y),\qquad w\ge0.

Weights and multiplicities add in the area formula; oriented cancellation is a separate current operation.↔ Scroll wide equations horizontally.

Work it through

A folded interval has twice its image length

Take $f(t)=t^2$ on $[-2,2]$ . Check its Lipschitz property on this domain and calculate the Jacobian integral, the measure of the image, and the target multiplicity.

The derivative is 2t and its magnitude is at most four on the domain, so f is Lipschitz there. Its one-dimensional Jacobian is $2|t|$ , whose integral from minus two to two is eight.
The image is $[0,4]$ with normalized one-dimensional measure four. Every target value strictly between zero and four has two preimages, one on each side of zero. The single exceptional zero value has no effect on the image integral.
Integrating multiplicity two over an interval of length four gives eight, matching the Jacobian integral. If one instead records the oriented pushforward of the interval current, the two branches have opposite signs; that cancellation does not alter the unsigned area calculation.

\int_{-2}^{2}2|t|\,dt=8=\int_0^4 2\,dy,\qquad \mathcal H^1(f([-2,2]))=4.

Counting each target point only once would produce the incorrect Jacobian integral.↔ Scroll wide equations horizontally.

The result

The parametrized area is eight, while the unweighted image length is four. Multiplicity is exactly the missing factor.

The living notebook

Explore the model

Interactive lab

Interactive lab · matrix

Shear changes shape while its determinant fixes area

Pause and Reset, then set diagonal scale a to 2 and shear b to 1. Predict unit-square image area two, matching the determinant readout. For a fresh comparison Reset and set a to 2 and b to 0; area is still two although the shape changes. Both maps are injective, so no multiplicity correction is needed. The instrument covers one triangular two-dimensional matrix family, not arbitrary Lipschitz maps or folds. Parameter changes retain elapsed time; Reset and both settings are needed for each comparison. For each run, take eight Steps to reach t=4 seconds, when the drawn geometry equals the selected matrix; its readouts describe that selected matrix throughout.

Interpolated geometryOriginal unit squareSelected final transform

Area scale · det A: 1.4
Singular values · σmax, σmin: 1.597, 0.876
Condition number · κ₂(A): 1.823

Diagonal scale a1.4Shear b0.6

t = 0 s

The selected matrix is A = [[a,b],[0,1]]. Playback interpolates geometry between the identity and A; readouts always describe A. Its determinant is a and its singular values come from AᵀA. Positive a makes every supported matrix invertible. This two-dimensional family is not a general eigensolver or a learned network. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forThe Jacobian measures volume distortion, while multiplicity requires additional information about global coverage.

Pause & reason

Check your understanding

Interactive checks need JavaScript. You can still read the explanation.

Read the checkpoint explanation

The target uses m-dimensional Hausdorff measure weighted by preimage count. Orientation and ambient n-volume describe different quantities, and dividing by multiplicity reverses the theorem.

Your turn

Try it yourself

Map the rectangle $[0,2]\times[0,3]$ into three-space by $F(s,t)=(2s,t,s)$ . Compute its two-dimensional Jacobian and image area. Explain why no multiplicity factor larger than one appears.

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

Use the two derivative columns and their inner products. Recovering s and t from the image checks injectivity.

Compare with the solution

The derivative columns are $(2,0,1)$ and $(0,1,0)$ , with Gram matrix diagonal entries five and one and zero off-diagonal entries. Thus $J_2F=\sqrt5$ . The map is injective, since its second and third coordinates recover t and s. The domain has area six, so the normalized image area is $6\sqrt5$ , with multiplicity one almost everywhere on the image.

Keep the idea

Take it with you

The area Jacobian uses the Gram matrix when the target has higher dimension.
Unsigned preimage counts distinguish parametrized area from image measure.
Oriented current cancellation must not be inserted into the unsigned area formula.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.