Throughout K(x, s) is assumed to be real valued and continuous for A x A (A = [a, b] is a finite closed interval of the real line). For definiteness we fix the domain 9(T) of the integral operator (1) as the Hilbert space L2(A); it should be emphasized at this point that K(x, s) is not necessarily a symmetric kernel. It will be clear from the subsequent analysis that the nature of the spectral set of T is unaltered for any of the alternative specifications g(T)= Lp(A), 1 < p < 00. It is well known that T is completely continuous. Therefore, the eigenvalues A = {)'o2, ...} form a discrete set which may be infinite, finite or empty. Each eigenvalue is of finite algebraic and geometric multiplicity and 0 is the only limit point of {,A} if A is not finite. Finally, the spectrum of the transformation T, apart from point spectrum A, can contain only the origin. Let r(T) denote the spectral radius of T, i.e.,
No takes yet. Share an insight, caveat, or question.
Samuel Karlin (1964) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: