By a case-free approach we give a precise description of the closure of a Steinberg fiber within a twisted wonderful compactification of a simple linear algebraic group. In the nontwisted case this description was earlier obtained by the first author.

Consider a perturbed KdV equation: ut +uxxx −6uux = εf(u(·)), x ∈ T = R/Z, where the nonlinear perturbation defines analytic operators u(·) 􏰀→ f(u(·)) in sufficiently smooth Sobolev spaces. Assume that the equation has an ε-quasi- invariant measure μ and satisfies some additional mild assumptions. Let uε(t) be a solution. Then on time intervals of order ε−1, as ε → 0, its actions I(uε(t,·)) can be approximated by solutions of a certain well-posed averaged equation, provided that the initial datum is μ-typical.

A well-known result on pseudodifferential operators states that the noncommutative residue (Wodzicki residue) of a zero-order pseudodifferential projection on a closed manifold vanishes. This statement is non-local and implies the regularity of the eta invariant at zero of Dirac type operators. We prove that in the odd dimensional case an analogous statement holds for the algebra of projective pseudodifferential operators, i.e., the noncommutative residue of a projective pseudodifferential projection vanishes. Our strategy of proof also simplifies the argument in the classical setting. We show that the noncommutative residue factors to a map from the twisted K-theory of the co-sphere bundle and then use arguments from twisted K-theory to reduce this question to a known model case. It can then be verified by local computations that this map vanishes in odd dimensions.

Motivated by multi-task machine learning with Banach spaces, we propose the notion of vector-valued reproducing kernel Banach spaces (RKBSs). Basic properties of the spaces and the associated reproducing kernels are investigated. We also present feature map constructions and several concrete examples of vector-valued RKBSs. The theory is then applied to multi-task machine learning. Especially, the representer theorem and characterization equations for the minimizer of regularized learning schemes in vector-valued RKBSs are established.

No abstract uploaded!