Difference between revisions of "Robin problem"
m (label) |
|||
Line 6: | Line 6: | ||
In the simplest classical case when is a closed domain in \mathbf R^n homeomorphic to the sphere, bounded by a smooth simple surface or (when n=2) by a curve S of class C^{1,\alpha}, 0<\alpha<1, 0\in K, the solution of Robin's problem is reduced to finding a non-trivial solution \nu(x), x\in S, of the homogeneous Fredholm-type integral equation of the second kind | In the simplest classical case when K is a closed domain in \mathbf R^n homeomorphic to the sphere, bounded by a smooth simple surface or (when n=2) by a curve S of class C^{1,\alpha}, 0<\alpha<1, 0\in K, the solution of Robin's problem is reduced to finding a non-trivial solution \nu(x), x\in S, of the homogeneous Fredholm-type integral equation of the second kind | ||
− | \frac12\nu(x)+\frac{1}{k_n}\int\limits_S\nu(y)\frac{\partial}{\partial n_x}E_n(x,y)dS(y)=0,\quad x\in S,\tag{1} | + | $$\frac12\nu(x)+\frac{1}{k_n}\int\limits_S\nu(y)\frac{\partial}{\partial n_x}E_n(x,y)dS(y)=0,\quad x\in S,\label{1}\tag{1}$$ |
under the normalization condition | under the normalization condition | ||
− | \lambda(S)=\int\limits_S\nu(y)dS(y)=1.\tag{2} | + | $$\lambda(S)=\int\limits_S\nu(y)dS(y)=1.\label{2}\tag{2}$$ |
Here | Here | ||
Line 20: | Line 20: | ||
k_2=2\pi,\quad k_n=\frac{2(n-2)\pi^{n/2}}{\Gamma(n/2)} | k_2=2\pi,\quad k_n=\frac{2(n-2)\pi^{n/2}}{\Gamma(n/2)} | ||
− | for n\geq3, and dS(y) is the area element of the surface S. Equation \ | + | for n\geq3, and dS(y) is the area element of the surface S. Equation \eqref{1} is obtained when one considers the interior [[Neumann problem]] for the domain bounded by S under vanishing boundary conditions, since the simple-layer potential |
u(x)=u(x,K)=\int\limits_S\nu(y)E_n(x,y)dS(y), | u(x)=u(x,K)=\int\limits_S\nu(y)E_n(x,y)dS(y), | ||
− | called the Robin potential, equilibrium potential or [[Capacity potential|capacity potential]], should, according to the condition of Robin's problem, have a constant value on K (see [[Potential theory|Potential theory]], and also [[#References|[2]]]). The solution \nu(x) for the problem \ | + | called the Robin potential, equilibrium potential or [[Capacity potential|capacity potential]], should, according to the condition of Robin's problem, have a constant value on K (see [[Potential theory|Potential theory]], and also [[#References|[2]]]). The solution \nu(x) for the problem \eqref{1}, \eqref{2} under the indicated conditions always exists in the class of continuous functions C(S). The measure |
\lambda(E)=\int\limits_E\nu(y)dS(y),\quad E\subset S, | \lambda(E)=\int\limits_E\nu(y)dS(y),\quad E\subset S, |
Latest revision as of 15:34, 14 February 2020
equilibrium problem, electrostatic problem
A problem on the distribution of a positive Borel measure \lambda on the boundary S of a compact set K in the n-dimensional Euclidean space \mathbf R^n, n\geq2, which generates a constant Newton potential for n\geq3, or constant logarithmic potential for n=2, on any connected component of the interior of K, i.e. the problem on the equilibrium distribution of an electric charge \lambda(K) on the surface S of a conductor K.
In the simplest classical case when K is a closed domain in \mathbf R^n homeomorphic to the sphere, bounded by a smooth simple surface or (when n=2) by a curve S of class C^{1,\alpha}, 0<\alpha<1, 0\in K, the solution of Robin's problem is reduced to finding a non-trivial solution \nu(x), x\in S, of the homogeneous Fredholm-type integral equation of the second kind
\frac12\nu(x)+\frac{1}{k_n}\int\limits_S\nu(y)\frac{\partial}{\partial n_x}E_n(x,y)dS(y)=0,\quad x\in S,\label{1}\tag{1}
under the normalization condition
\lambda(S)=\int\limits_S\nu(y)dS(y)=1.\label{2}\tag{2}
Here
E_2(x,y)=\ln\frac{1}{|x-y|},\quad E_n=\frac{1}{|x-y|^{n-2}}
for n\geq3, |x-y| is the distance between two points x,y\in\mathbf R^n, n_x is the direction of the exterior normal to S at the point x\in S, \nu(x) is the derivative, or density, of the absolutely-continuous measure \lambda with respect to the Lebesgue measure on S,
k_2=2\pi,\quad k_n=\frac{2(n-2)\pi^{n/2}}{\Gamma(n/2)}
for n\geq3, and dS(y) is the area element of the surface S. Equation \eqref{1} is obtained when one considers the interior Neumann problem for the domain bounded by S under vanishing boundary conditions, since the simple-layer potential
u(x)=u(x,K)=\int\limits_S\nu(y)E_n(x,y)dS(y),
called the Robin potential, equilibrium potential or capacity potential, should, according to the condition of Robin's problem, have a constant value on K (see Potential theory, and also [2]). The solution \nu(x) for the problem \eqref{1}, \eqref{2} under the indicated conditions always exists in the class of continuous functions C(S). The measure
\lambda(E)=\int\limits_E\nu(y)dS(y),\quad E\subset S,
which provides a solution of the Robin problem, is called the equilibrium measure. In a more complicated case, when the boundary of the compact set K consists of a finite number of non-intersecting simple closed surfaces or (when n=2) curves of class C^{1,\alpha}, 0<\alpha<1 (see [2]), the Robin problem is solved in a similar way. Moreover, on bounded connected components of the open set G=CK=\mathbf R^n\setminus K the Robin potential u(x) also preserves its constant value, i.e. on the boundaries of these components the density \nu(x)=0.
Let the compact set K be connected. The constant value of the Robin potential u(x) on K,
\gamma=\int\limits_S\nu(y)E_n(x,y)dS(y),\quad x\in K,
is called the Robin constant of the compact set K. For n\geq3 it is related to the harmonic, or Newton, capacity C(K) of K by the simple relation C(K)=1/\gamma; moreover, 0<\gamma<+\infty, 0<C(K)<+\infty. For n=2, the Robin constant can assume all values -\infty<\gamma<+\infty; the harmonic capacity is then expressed by the formula C(K)=e^{-\gamma}.
In another way, the equilibrium measure \lambda is defined as the measure which yields the minimum of the energy integral
\iint\limits_{K\times K}E_n(x,y)d\mu(x)d\mu(y)
in the class of all measures \mu concentrated on K and such that \mu\geq0, \mu(K)=1. Such a measure \lambda in the case of a compact set K with a smooth boundary coincides with the one found above, but it exists also in the general case of an arbitrary compact set K\subset\mathbf R^n, n\geq2, if only C(K)>0. The corresponding equilibrium potential
u(x)=u(x;K)=\int E_n(x,y)d\lambda(y),
which is a generalization of the Robin potential, preserves the constant value \gamma=1/C(K) for n\geq3, or \gamma=-\ln C(K) for n=2, everywhere on K except perhaps at the points of some set of capacity zero.
The name "Robin problem" is connected with studies of G. Robin (see [1]).
References
[1] | G. Robin, "Sur la distribution de l'électricité à la surface des conducteurs fermés et des conducteurs ouverts" Ann. Sci. Ecole Norm. Sup. , 3 (1886) pp. 31–358 |
[2] | N.M. Günter, "Potential theory and its applications to basic problems of mathematical physics" , F. Ungar (1967) (Translated from Russian) |
[3] | N.S. Landkof, "Foundations of modern potential theory" , Springer (1972) (Translated from Russian) |
[4] | W.K. Hayman, P.B. Kennedy, "Subharmonic functions" , 1 , Acad. Press (1976) |
Comments
In [1] Robin reconsiders and generalizes a problem formulated by S. Poisson (1811).
References
[a1] | M. Tsuji, "Potential theory in modern function theory" , Chelsea, reprint (1975) |
Robin problem. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Robin_problem&oldid=33909