
A Probabilistic Perspective on the Hamiltonian Diffeomorphism Group
Adrian Dawid of the University of Cambridge presents an Institute for Advanced Study symplectic geometry seminar on what it means for a Hamiltonian diffeomorphism to be random. He constructs a family of probability measures on the Hamiltonian diffeomorphism group Ham(M) of a closed symplectic manifold, showing the family has full support and that invariants such as the Hofer norm, spectral invariants, stable commutator length, and homogeneous quasimorphisms behave as well-defined random variables under it. In the second half, Dawid uses these measures to build random walks probing the large-scale geometry of Ham(M) under the Hofer metric, showing that on complex projective space the expected Hofer norm after k steps grows at least like the square root of k, while a walk confined to a commutative subgroup grows no faster than that rate. He closes by proving the rate is exactly order square root of k for a torus-invariant walk on complex projective space.