Optimal Transport, Wasserstein Gradient Flows, and the JKO sc.. (Lecture 1) by Filippo Santambrogio

Optimal Transport, Wasserstein Gradient Flows, and the JKO sc.. (Lecture 1) by Filippo Santambrogio

Title history
  • Optimal Transport, Wasserstein Gradient Flows, and the JKO sc.. (Lecture 1) by Filippo Santambrogio — observed on 20261104-0000 UTC (YouTube)

by Filippo Santambrogio · View the channel on PSI

Formal & Physical Sciences Mathematics PBKCalculus and mathematical analysisPBKFFunctional analysis and transforms
🎙 Filippo Santambrogio 👥 75K 📅 November 3, 2026 ⏱ 92 min 👁 653 📄 tutorial 🧭 2026-11-04
Available in: English (current) Français

Keywords

Wasserstein distanceMonge problemKantorovich dualityJKO schemeGradient flows

Summary

Filippo Santambrogio opens a lecture series on optimal transport, Wasserstein gradient flows, and the JKO scheme. He begins by recalling the Monge formulation of optimal transport, which seeks a mapping T that pushes one probability measure to another while minimizing a cost. He illustrates the multiplicity of transport maps and the difficulty of proving existence, due to the non-linear constraint on the Jacobian determinant. He then introduces Kantorovich’s relaxation, which considers joint distributions on the product space, leading to a linear programming problem. Under compactness assumptions, existence of a Kantorovich solution is established via weak-* compactness. The lecture proceeds to discuss convex duality as a tool for deriving optimality conditions, highlighting how duality uncovers potentials and suggests that an optimal solution can be of Monge type. The talk sets the stage for subsequent lectures on the JKO scheme, which uses the Wasserstein distance as a metric in minimizing movements to approximate gradient flows for PDEs.

155 words

Critical Evaluation

Value of the Information & Strength of the Argument

The lecture provides an accessible yet rigorous introduction to the foundational concepts of optimal transport and its connection to gradient flows. The argumentation is clear and progressive: it starts from the Monge problem, identifies its shortcomings, and motivates Kantorovich’s reformulation. The existence proof is presented with sufficient details, and the introduction of duality is motivated by the linear structure of the problem. The value lies in its pedagogical clarity, making it an excellent resource for graduate students and researchers entering the field.

Scientific Rigor, Source Quality, Title Accuracy

Scientific rigor is high; the lecturer builds on classical results (Monge, Kantorovich) and uses standard functional analysis arguments (Banach-Alaoglu, Prokhorov). The video is part of an ICTS school, ensuring institutional credibility. The title matches content precisely, covering the announced topics. No external sources are cited in the description besides the program link, but the content aligns with well-established literature, including Santambrogio’s own textbooks.

160 words

Title / Content Match

The title accurately reflects the content: the lecture introduces optimal transport, Wasserstein gradient flows, and the JKO scheme. The truncated abbreviation in the title does not hinder understanding.

Quality & Reliability

9/10

Lecture by a leading expert in optimal transport and calculus of variations. The mathematical content is rigorously presented, consistent with established literature, and delivered in a didactic manner.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Concurring Sources

Contribution & Novelties

This lecture serves as a self-contained introduction to the key mathematical tools needed for understanding JKO schemes and Wasserstein gradient flows. Its main contribution is to bridge the gap between classical optimal transport theory and modern applications in PDEs and machine learning. The exposition is particularly valuable for beginners, clarifying the transition from Monge to Kantorovich and the role of convex duality.

Pour aller plus loin :

  • Optimal transport (Wikipedia) — Foundational overview of the topic, covering both Monge and Kantorovich formulations.
  • Wasserstein metric (Wikipedia) — Definition and properties of the metric used in gradient flows.
  • Gradient flow (Wikipedia) — General concept of gradient descent in metric spaces, relevant to the JKO scheme.
  • Jordan, Kinderlehrer, Otto (1998) — Variational formulation of the Fokker–Planck equation — The original paper introducing the JKO scheme.

132 words

Radar Profile

The scores are uniformly high (9/10) across all dimensions, reflecting a technically deep, information-dense, and reliable lecture. The moderate view count and likes do not detract from the scientific quality, which is aimed at a specialized audience.

Reliability 9/10