Ulrike Tillman: From Topology via Complex Data to Representation Theory (October 16, 2026)

Ulrike Tillman: From Topology via Complex Data to Representation Theory (October 16, 2026)

Title history
  • Ulrike Tillman: From Topology via Complex Data to Representation Theory (October 16, 2026) — observed on 20261117-1758 UTC (YouTube)
Formal & Physical Sciences Mathematics PBMathematicsPBPTopology
🎙 Ulrike Tillmann 👥 62K 📅 November 17, 2026 ⏱ 44 min 👁 8 📄 original study 🧭 2026-11-17
Available in: English (current) Français

Keywords

topological data analysispersistent homologybarcodesmultiparameter persistenceHarder-Narasimhan filtration

Summary

Ulrike Tillmann’s lecture bridges pure mathematics and applied data analysis. She first introduces the Newton Institute and its role in mathematical sciences. The core of the talk begins with topological data analysis (TDA): how to assign topology to point clouds via simplicial complexes (Čech and Vietoris-Rips), and how persistent homology captures shape features across scales. She explains barcodes and persistence landscapes as vectorisations enabling statistics. Extending to multiparameter persistence, she highlights that single-parameter homology is not noise-resistant and demonstrates how introducing a density parameter improves robustness. Applications to cancer imaging (tumor immune infiltration, kidney cells) show the practical value of multiparameter persistence in discriminating cell types. Tillmann then transitions to the theoretical side: multiparameter persistence modules as quiver representations. She introduces the Harder-Narasimhan filtration, an invariant from geometric invariant theory, and explains how it provides a finer, more complete descriptor than the rank invariant. The talk exemplifies two-way traffic between theory and applications, as data-driven questions motivate new invariants in representation theory.

162 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk is extremely valuable for its clear exposition of state-of-the-art TDA techniques and their concrete applications in biomedical imaging. Tillmann’s argumentation is logically sound: she motivates each step, from single- to multiparameter persistence, with explicit limitations (e.g., noise sensitivity) and then shows how theory addresses them. The argument that multiparameter persistence outperforms standard persistent homology in noisy data is supported with simulations and real data examples. The leap to quiver representations is well motivated: the Harder-Narasimhan filtration offers a richer invariant than the rank invariant, and its geometric origin is explained intuitively. While proofs are not given, the presentation is coherent and convincing for a mathematically literate audience.

Scientific Rigor, Source Quality, Title Accuracy

The speaker demonstrates scientific rigor by citing specific prior work (e.g., Carlson and Zamarogen on persistence algorithms and vectorisations) and referencing joint work with students and colleagues in Oxford. The link provided in the description points to the Simons Foundation event page, which is a legitimate institutional source. The title accurately reflects the thematic progression from topology through complex data to representation theory, though the applied biology sections are not fully captured. No discordant sources or controversies are addressed, and the talk stays within the speaker’s expertise, making the information reliable.

215 words

Title / Content Match

The title accurately reflects the narrative arc from topological data analysis to representation theory, though the applied biology parts are broader than the title suggests.

Quality & Reliability

8/10

The talk is given by a leading expert in algebraic topology and topological data analysis, presenting well-established methods alongside novel research. Arguments are clearly structured with mathematical precision, but as a conference presentation it lacks peer review and detailed proofs.

Key Moments

Cited Sources

Contribution & Novelties

The talk’s novel contribution is the introduction of the Harder-Narasimhan filtration as an invariant for multiparameter persistence modules, stemming from geometric invariant theory. It provides a new, finer descriptor that may lead to a complete invariant and better practical performance. The presentation also emphasises the two-way interplay between theory and applications, illustrating how biological data motivated theoretical advances.

Pour aller plus loin :

102 words

Radar Profile

The radar profile shows strong scores across quantity, quality, technical depth, and reliability, indicating a balanced and rigorous presentation. The slight dip in 'quality' relative to quantity may reflect the lack of formal references, but overall it is a highly informative and trustworthy talk.

Reliability 8/10