Einstein's General Relativity

Derivations, Applications and Considerations – by Albert Prins

Appendix 6 — Derivation of Gauss's Theorem

We start with a cube of infinitesimally small dimensions.

vector_6_1_1
A flux \(\vec{F}\) passes through this infinitesimal cube. This flux is not uniform everywhere and therefore is a function of \(x,y,z,t\).

The flux is a vector, because it has both magnitude and direction:

\[ \vec{F}_{\text{flux}} = \vec{F}(x,y,z,t) \tag{1} \]

Flux through a surface

Consider now the right side of the cube, a plane parallel to the \(y\)-\(z\) plane. The flux through this surface is determined by the component of \(\vec{F}\) perpendicular to that plane.

If \(\xi\) is the angle between \(\vec{F}\) and the surface, then:

\[ \vec{F}_{\text{right}} = \vec{F} \,\sin\xi \, dy\,dz \tag{2} \]

We represent the surface as a vector \(d\vec{A}\) perpendicular to the plane:

\[ d\vec{A} = \vec{dy} \times \vec{dz}, \qquad |dA| = \sin\xi \, dy\,dz \tag{3} \]

The flux through the right side is then:

\[ \vec{F}_{\text{right}} = \vec{F} \sin\xi \, dy\,dz = \vec{F} \cos\!\left(\tfrac{\pi}{2}-\xi\right) dA = \vec{F} \cos\varphi \, dA = \vec{F}\cdot d\vec{A} \tag{4} \]

Here \(d\vec{A}\) is perpendicular to the surface, and \(\varphi\) is the complementary angle of \(\xi\). We recognize the dot product:

\[ Flux_{\text{right}}=\vec{F}d\vec{A}\,\cos\phi=\vec{F}\cdot d\vec{A} \tag{5} \]

Flux through the total surface of the cube

For a finite cube, the total flux is the sum of the contributions from all six faces:

\[ F_{\text{cube}} = \sum_{\text{all faces}} \vec{F}\cdot d\vec{A} \tag{6} \]

We can write this as a single integral over the closed surface:

\[ F_{\text{cube}} = \oint_{\partial V} \vec{F}\cdot d\vec{A} \tag{7} \]

Alternative approach: flux as a limit

In the \(x\)-direction, the incoming flux is:

\[ F_{\text{left}} = F_x \, dy\,dz \tag{8} \]

The flux leaving the right side is:

\[ F_{\text{right}} = (F_x + dF_x)\, dy\,dz \tag{9} \]

The net flux in the \(x\)-direction:

\[ F_x^{\text{net}} = F_{\text{right}} - F_{\text{left}} = dF_x\, dy\,dz \tag{10} \]

Similarly:

\[ F_y^{\text{net}} = dF_y\, dx\,dz, \qquad F_z^{\text{net}} = dF_z\, dx\,dy \tag{11–12} \]

The total flux through the cube:

\[ F_{\text{cube}} = dF_x\,dy\,dz + dF_y\,dx\,dz + dF_z\,dx\,dy \tag{13} \]

Rewritten using partial derivatives:

\[ F_{\text{cube}} = \left( \frac{\partial F_x}{\partial x} + \frac{\partial F_y}{\partial y} + \frac{\partial F_z}{\partial z} \right) dx\,dy\,dz \]

\[ F_{\text{cube}} = (\vec{\nabla}\cdot\vec{F})\, dV \tag{15} \]

The operator ∇

\[ \vec{\nabla} = \frac{\partial}{\partial x}\,\hat{e}_x + \frac{\partial}{\partial y}\,\hat{e}_y + \frac{\partial}{\partial z}\,\hat{e}_z = \left( \frac{\partial}{\partial x}, \frac{\partial}{\partial y}, \frac{\partial}{\partial z} \right) \tag{14} \]

Then equation (13) becomes:

\[ F_{\text{cube}} = (\vec{\nabla}\cdot\vec{F})\, dV \tag{15} \]

Net flux through the cube

By integrating over the entire volume of the cube we find:

\[ F_{\text{cube}} = \iiint_{\text{cube}} (\nabla \cdot \vec{F})\, dV \tag{16} \]

Gauss's Theorem

Equation (7) gave the flux through the closed surface:

\[ F_{\text{cube}} = \oint_{\partial V} \vec{F}\cdot d\vec{A} \]

Equation (16) gave the same flux as a volume term:

\[ F_{\text{cube}} = \iiint_V (\vec{\nabla}\cdot\vec{F})\, dV \]

Since both expressions describe the same flux, it follows:

\[ \oint_{\partial V} \vec{F}\cdot d\vec{A} = \iiint_V (\vec{\nabla}\cdot\vec{F})\, dV \tag{17} \]

Since the volume was arbitrary (not necessarily a cube), this holds for any closed volume:

\[ \oint_{\partial V} \vec{F}\cdot d\vec{A} = \iiint_V (\vec{\nabla}\cdot\vec{F})\, dV \tag{18} \]

This is Gauss's Theorem (also called the Divergence Theorem).

Special case: zero flux

If the net flux through the closed surface is zero:

\[ \oint_{\partial V} \vec{F}\cdot d\vec{A} = 0 \]

then from Gauss's theorem:

\[ \iiint_V (\vec{\nabla}\cdot\vec{F})\, dV = 0 \tag{19} \]

Since the volume is arbitrary:

\[ \vec{\nabla}\cdot\vec{F} = 0 \tag{20} \]

Written in components:

\[ \frac{\partial F_x}{\partial x} + \frac{\partial F_y}{\partial y} + \frac{\partial F_z}{\partial z} = 0 \tag{21} \]

In Einstein notation (summing over repeated index \(\alpha\)):

\[ \frac{\partial F_\alpha}{\partial x_\alpha} = 0 \tag{22} \]