The problem of computing spectra of operators is arguably one of the most investigated areas of computational mathematics. Recent progress and the current paper reveal that, unlike the finite-dimensional case, infinite-dimensional problems yield a highly intricate infinite classification theory determining which spectral problems can be solved and with which type of algorithms. Classifying spectral problems and providing optimal algorithms is uncharted territory in the foundations of computational mathematics. This paper is the first of a two-part series establishing the foundations of computational spectral theory through the Solvability Complexity Index (SCI) hierarchy and has three purposes. First, we establish answers to many longstanding open questions on the existence of algorithms. We show that for large classes of partial differential operators on unbounded domains, spectra can be computed with error control from point sampling operator coefficients. Further results include computing spectra of operators on graphs with error control, the spectral gap problem, spectral classifications, and discrete spectra, multiplicities and eigenspaces. Second, these classifications determine which types of problems can be used in computer-assisted proofs. The theory for this is virtually non-existent, and we provide some of the first results in this infinite classification theory. Third, our proofs are constructive, yielding a library of new algorithms and techniques that handle problems that before were out of reach. We show several examples on contemporary problems in the physical sciences. Our approach is closely related to Smales program on the foundations of computational mathematics initiated in the 1980s, as many spectral problems can only be computed via several limits, a phenomenon shared with the foundations of polynomial root finding with rational maps, as proved by McMullen.
Download