Let
and choose . Write for the positively oriented boundary of the square with . The Cauchy integral formula gives, for every ,
Because and are compact and separated by a positive distance, Riemann sums for this contour integral converge uniformly in . Thus, for any , there are points and constants such that
It remains to approximate each resolvent by polynomials on . Let be the set of for which is polynomially uniformly approximable on . If , the geometric series
shows that . Moreover, the resolvent propagation across a complementary component argument shows that is both open and closed relative to . The complement of a square is connected, and is nonempty, so .
In particular, every lies in . Approximate the finitely many resolvents accurately enough that their weighted sum differs by less than . Combining this with the Riemann-sum estimate produces a polynomial satisfying
This proves the claimed elementary polynomial approximation on a square.
For , write . If and
then
uniformly on . Uniformly polynomially approximable functions form an algebra closed under uniform limits, so . Hence is open.
The set is also closed relative to . Indeed, if and , then
so is again uniformly approximable by polynomials. Since is closed, every point of has a neighbourhood in ; relative closedness of then gives a smaller neighbourhood disjoint from . Thus both and are open.
It is not always true that is nonempty. Take . Any polynomial uniformly close on to a bounded resolvent , with , must itself be bounded on , and hence must be constant. Uniform limits of constants are constant, whereas is not. Therefore
Boundedness of does not force to be empty either. Take and . If polynomials converged uniformly to on , then uniform convergence and contour integration would give
a contradiction. Hence

Articles by others on the same topic (0)

There are currently no matching articles.