I'm watching Leonard Susskind's lectures on relativity (http://www.youtube.com/watch?v=s8UrYIZhm60&feature=youtu.be&t=31m08s), and he just introduced Gauss's law on gravity, that is:

$$\vec\nabla \cdot \vec A = -4\pi G \rho$$

To my understanding, this says that divergence at point $(x,y,z)$ is non-zero only if there is mass at that point, i.e. $\rho(x,y,z) \ne 0$.

However, this seems in conflict with the idea of divergence that I got from the definition. If $\vec A(x,y,z)$ is non-zero, and the vector field around $(x,y,z)$ is not constant, then the sum of the partial derivatives will be non-zero, regardless of whether there is any mass at $(x,y,z)$.

Is something wrong in my reasoning or am I not getting Gauss's law right?

  • $\begingroup$ If the derivatives are non-zero, and there is a non-zero divergence, then there is a density at that point. The geometric understanding of divergence is related to density via Gauss' law. $\endgroup$
    – JamalS
    Dec 9, 2016 at 14:15
  • $\begingroup$ Why? If I have a single mass in space at $(0,0,0)$, the acceleration vectors at $(x,y,z)$ and $(x,y,z+dz)$ will point in slightly different directions, thus the derivatives will be non-zero, but there's no density there. $\endgroup$
    – rand
    Dec 9, 2016 at 14:31

1 Answer 1


The flaw with your reasoning is that the vector field being not constant around $(x,y,z)$ does not necessarily imply that the divergence is not $0$ at that point, although what you said does hold the other way around. (Think back on the definition of divergence.) For example, $(y^2,z^2,x^2)$ is not a constant vector field, but it's divergence is indeed $0$. The equation $∇⃗ ⋅A =−4πGρ$ actually arises from the integral identity $$\int_V\nabla\cdot\vec{F} dV=\oint_S\vec{F}\cdot d\vec{a}$$ which is the integral form of Gauss law in mathematics. Putting in the gravity field $\vec{A}$ due to mass distribution $\rho$ into the formula, and using the fact that it should hold anywhere, you get $∇⃗ ⋅A =−4πGρ$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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