Say we have an $N$-dimensional hypervolume, which is bounded by a parametric $N-1$-dimensional hypersurface. We can state the dimensions as $x_1\dots x_n$ and the parameters as $t_1\dots t_{n-1}$.

Would the formula:

$\frac{1}{N}\idotsint\begin{vmatrix}x_1 & \dots & x_N \\\frac{\partial x_1}{\partial t_1} & \dots & \frac{\partial x_N}{\partial t_1} \\\vdots & \ddots & \vdots \\\frac{\partial x_1}{\partial t_{N-1}} & \dots & \frac{\partial x_N}{\partial t_{N-1}} \\\end{vmatrix}dt_1 \dots dt_{N-1}$

allow me to find its hypervolume?


No, that is not quite right. The actual formula is hard to express in closed form.

Let's see where the original formula comes from $\int y(t)x'(t)dt$. This comes from taking $\int y(t)dx$ and substituting in $x'(t)dt$ for $dx$.

First of all, the variable that is used in the integral is not arbitrary. Doing this in n dimensions requires choosing one variable. This is the variable that the area "under" is measured. I will call this variable $y$, and all the other variables $x_1$ through $x_n$. In the case with two dimensions, the formula becomes $\int y(t_1)x_1'(t_1)dt_1$.

We can create an equivalent formula for more dimensions, an example with 3:

$$\iint y(t_1, t_2)dx_1dx_2$$

Now we apply a change of variables with multiple integrals. An explanation of how to do this can be found here: https://en.wikipedia.org/wiki/Integration_by_substitution#Substitution_for_multiple_variables. Basically, substitute $dx_1...dx_n$ with $$\left|d\begin{bmatrix} \frac{dx_1}{dt_1} & \dots & \frac{dx_1}{dt_n}\\ \vdots & \ddots & \vdots\\ \frac{dx_n}{dt_1} & \dots & \frac{dx_n}{dt_n} \end{bmatrix}\right|$$ Where $d(m)$ is the discriminant of matrix $m$. This gives the formula: $$ \int\dots\int y \left|d\begin{bmatrix} \frac{dx_1}{dt_1} & \dots & \frac{dx_1}{dt_n}\\ \vdots & \ddots & \vdots\\ \frac{dx_n}{dt_1} & \dots & \frac{dx_n}{dt_n} \end{bmatrix}\right|dt_1\dots dt_n$$

Note that this function doesn't always work as nicely as the integral works in 2 dimensions. It may count the volume of certain areas multiple times. Strictly, it calculates the total volume between the function and the $y=0$ plane, counting volume below the plane as negative. Removing the absolute value bars from the Jacobian discriminant may make the integral work like the 2-dimensional version, but I can't confirm that.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.