* Seuraavan viikon tapahtumat merkitty tähdellä
Tuomas Kelomäki (Aalto University)
Fast and smooth? Khovanov homology and computational complexity
* Tuesday 09 December 2025, 10:15, M3 (M234)
At the turn of the century, Khovanov upgraded the Jones polynomial into a homology theory of knots, which is sensitive to smooth structures in 4D. The Jones polynomial can be recovered from Khovanov homology, so Khovanov homology is at least as hard to compute as the Jones polynomial, and it is an open question how much harder it is. In this talk, we will try to explain why mathematicians should care about fast computations of Khovanov homology. We will also explore polynomial and non-polynomial time algorithms for both Jones polynomial and Khovanov homology of braids. Joint work with Dirk Schütz.
Meeri Palokangas (Aalto University)
A quasi-likelihood-based gradient boosting machine for humanitarian demand prediction (MSc presentation)
* Tuesday 09 December 2025, 11:15, M203
Predicting food aid demand requires balancing theoretical soundness with computational practicality while handling the messiness of zero-inflated humanitarian data. This thesis develops a natural gradient boosting approach for probabilistic demand forecasting by implementing the Extended Quasi-Likelihood (EQL) function within NGBoost, enabling native support for Tweedie-parametrized Compound Poisson-Gamma distributions without the computational barriers of exact likelihood estimation. The method generates full predictive distributions, which allows supply planners to quantify uncertainty and make risk-aware decisions in prepositioning and predictive procurement. The approach is evaluated on handover data from the World Food Programme (WFP) from April 2024 to March 2025, along with backtesting across six years. Results show conservative calibration errors of for central quantiles and relatively strong performance at extreme values where humanitarian logistics decisions are most consequential. While point prediction accuracy doesn't outperform the benchmark LGBM approach, the superior distributional calibration directly addresses WFP's stated need for quantified uncertainty in supply chain planning and makes the trade-off justified. The work bridges quasi-likelihood statistical theory with practical machine learning, allowing feasibility of learning on standard hardware, as opposed to the full likelihood approximation methods of these kinds of distributions, which are too heavy for repeated computations. However, the theoretical regularity (in the Riemannian sense) of the quasi-likelihood statistical manifolds, which is proposed in this work, remains unproven, leaving important theoretical questions about why the approach works empirically.
Aalto Stochastics & Statistics Seminar / Leskelä
Prof. Tuomas Hytönen (Aalto University)
Commutators, finite rank approximation, classical and quantum derivatives
* Tuesday 09 December 2025, 15:15, M1 (M232)
Further information
Quantifying the failure of the commutative law ab = ba for objects beyond numbers is key to diverse topics in pure and applied mathematics. This is achieved by studying the properties of the related commutator. Anything that one might wish to feed to a computer is necessarily finite, and hence it is useful to know how well such commutators can be estimated by finite rank approximations. It turns out that the rate of approximation has a sharp threshold that cannot be beaten, while the best possible approximation rate has connections to both classical differentiability and a certain quantum analogue. A general framework for these questions is contained in my recent work with Riikka Korte.
Dr. Nikolay Barashkov (Max Planck Institute for Mathematics in the Sciences, Leipzig)
TBA
Wednesday 07 January 2026, 10:15, Y346
Adria Marin Salvador (University of Oxford)
TBA
Tuesday 13 January 2026, 10:15, M3 (M234)
Prof. Steven Gabriel (University of Maryland, NTNU, and Aalto University)
Solving Supply Chain Equilibrium Problems in Energy and Other Infrastructure Areas using a Difference-of-Convex Functions Algorithm
Tuesday 13 January 2026, 15:15, M1 (M232)
Further information
We describe a novel application of the difference-of-convex function algorithm (DCA) to a variety of equilibrium problems using the mixed complementarity problem (MCP) format. These problems involve bilinear constraints, i.e., complementarity and can be approximated iteratively via convex subproblems using DCA. We develop the necessary theory to make this possible and showcase how it works on several MCPs in energy and other infrastructure areas such as: the Brazilian natural gas market and water markets.
Prof. Lalitha Vadlamani (IIIT Hyderabad)
TBA
Monday 19 January 2026, 15:15, M3 (M234)
ANTA Seminar / Hollanti et al.
Kai Hippi (Aalto University)
TBA
Tuesday 20 January 2026, 10:15, M3 (M234)
Joonas Vättö (Aalto University)
TBA
Tuesday 27 January 2026, 10:15, M3 (M234)
Prof. Andrea Pinamonti (Università di Trento)
TBA
Wednesday 04 February 2026, 10:15, M3 (M234)
Seminar on analysis and geometry
Prof. Anders Hansen (University of Cambridge)
TBA
Tuesday 10 February 2026, 15:15, M1 (M232)
Dr. John Urschel (MIT)
TBA
Tuesday 10 March 2026, 15:15, U5 (U147)
Show the events of the past year
Sivusta vastaa: webmaster-math [at] list [dot] aalto [dot] fi