We saw in Equation (2.16) that
![$\displaystyle \frac{{\bf r}- {\bf r}' }
{\vert{\bf r} - {\bf r}'\vert^3} = -\nabla\!\left(\frac{1}{\vert{\bf r} - {\bf r'}\vert}\right).$](img1614.png) |
(2.249) |
This equation can be combined with the generalized Biot-Savart law, (2.248), to give
![$\displaystyle {\bf B}({\bf r}) = \frac{\mu_0}{4\pi}\int\nabla\!\left(\frac{1}{\vert{\bf r} - {\bf r'}\vert}\right)\times {\bf j}({\bf r}')\,dV'.$](img1615.png) |
(2.250) |
It follows that
![$\displaystyle {\bf B} = \nabla\times {\bf A},$](img1616.png) |
(2.251) |
where
![$\displaystyle {\bf A}({\bf r})= \frac{\mu_0}{4\pi}\int\frac{{\bf j}({\bf r}')}{\vert{\bf r}-{\bf r}'\vert}\,dV'.$](img1617.png) |
(2.252) |
(See Section A.24.)
Here, the vector field
is known as the magnetic vector potential.
It is possible to prove that the
magnetic vector potential defined in the previous equation is a divergence-free field.
Note that
![$\displaystyle \frac{\partial}{\partial x}\! \left(\frac{1}{\vert{\bf r} - {\bf ...
...c{\partial}{\partial x'} \!\left(\frac{1}{\vert{\bf r} - {\bf r}'\vert}\right),$](img1619.png) |
(2.253) |
which implies that
![$\displaystyle \nabla\left( \frac{1}{\vert{\bf r} - {\bf r}'\vert}\right) = - \nabla'
\left( \frac{1}{\vert{\bf r} - {\bf r}'\vert}\right),$](img1620.png) |
(2.254) |
where
is the operator
. (See Section A.19.) Taking the divergence of Equation (2.252), and making use of
the previous relation, we obtain
![$\displaystyle \nabla\cdot{\bf A} = \frac{\mu_0}{4\pi} \int {\bf j}({\bf r'})\cd...
...\bf r'})\cdot
\nabla'\left( \frac{1}{\vert{\bf r} - {\bf r}'\vert}\right)\,dV'.$](img1623.png) |
(2.255) |
Now,
![$\displaystyle \int_{-\infty}^{\infty} g \,\frac{\partial f}{\partial x}\,dx = \...
...infty}^{\infty} - \int_{-\infty}^{\infty}
f\,\frac{\partial g}{\partial x}\,dx.$](img1624.png) |
(2.256) |
However, if
as
then we can neglect the
first term on the right-hand side of the previous equation, and write
![$\displaystyle \int_{-\infty}^{\infty} g\, \frac{\partial f}{\partial x}\,dx = -
\int_{-\infty}^{\infty} f\, \frac{\partial g}{\partial x}\,dx.$](img1627.png) |
(2.257) |
A simple generalization of this result yields
![$\displaystyle \int {\bf g}\cdot\nabla f\,dV = - \int f\,\nabla\cdot {\bf g}\,dV,$](img1628.png) |
(2.258) |
provided that
as
, etc cetera.
Thus, Equation (2.255)
yields
![$\displaystyle \nabla\cdot{\bf A} = \frac{\mu_0}{4\pi}\int \frac{\nabla' \cdot {\bf j}({\bf r}') }
{\vert{\bf r} - {\bf r}'\vert} \,dV',$](img1630.png) |
(2.259) |
provided that
is bounded as
. Now, the flux of electric charge out of a surface
, enclosing a volume
, is
![$\displaystyle \oint_S {\bf j}\cdot d{\bf S} = \int_V \nabla\cdot{\bf j}\,dV,$](img1632.png) |
(2.260) |
where use has been made of the divergence theorem. (See Section A.20.)
However, for a steady current distribution, this flux must be zero, otherwise positive or negative electric
charge would build up inside
. Moreover, the flux must be zero for all possible volumes,
, which implies that
![$\displaystyle \nabla\cdot{\bf j} = 0$](img1633.png) |
(2.261) |
for a steady current distribution. Hence, we deduce from Equation (2.259) that
![$\displaystyle \nabla\cdot{\bf A} = 0$](img1634.png) |
(2.262) |
for a steady current distribution.