CNRS Researcher in Mathematics (Chargé de recherche)
Part-time Professor at École polytechnique
Email (work): siarhei.finski(at)polytechnique.edu
Email (personal): finski.siarhei(at)gmail.com
Office: 06.1006, CMLS, École polytechnique
My research lies at the intersection of complex and algebraic geometry, pluripotential theory and global analysis on manifolds, and draws on functional analysis. I have also contributed to index theory and probability theory.
Broadly, my work is guided by two themes. The first is approximation: how infinite-dimensional objects of geometric analysis, such as spaces of functions, metrics and operators, can be approximated by finite-dimensional ones. The second is localization: objects of interest in geometry are often global in nature, yet as a parameter in their definition becomes large, their behavior might become local.
A guiding example is a theorem of Tian (1990), which can be seen as a holomorphic analogue of the Nash embedding theorem. Nash showed that every Riemannian manifold can be isometrically embedded into a Euclidean space. Holomorphic isometric embeddings into projective spaces are much more rigid, but Tian showed that one can construct holomorphic embeddings that are arbitrarily close to isometric, by letting the dimension of the projective space grow. The proof relies on the fact that the Bergman kernel, a global object, becomes local in this limit.
In my work, I study complex manifolds through the spaces of holomorphic sections of high tensor powers of a line bundle, the same spaces that give Tian's embeddings. These finite-dimensional spaces are classical objects of algebraic geometry, but their behavior as the tensor power grows can be studied with tools from analysis. Localization phenomena, which have no obvious algebraic counterpart, make analytic methods especially natural here.
The interplay between algebra and analysis works in both directions: I use analytic methods to compute algebraic invariants, such as the Kodaira-Iitaka dimension, and conversely use algebraic notions, such as filtrations or stability conditions, to study problems in geometric analysis, such as the Wess-Zumino-Witten equation.