Skip to main content
Aller à la page d’accueil de la Commission européenne (s’ouvre dans une nouvelle fenêtre)
français fr
CORDIS - Résultats de la recherche de l’UE
CORDIS

(Non)-Metric Embeddings

Objectif

"Many problems we face are naturally expressed in terms of distances. Rather than solving each one on an ad hoc basis, researchers have built a rich toolbox. Metric embeddings play a central role: points of a complex metric space are embedded into a simpler one (e.g. a tree metric or Euclidean space) while approximately preserving structure. This lets us solve the problem in the simple space—where fast, well-understood algorithms exist—and then transfer solutions back. Metric embeddings have enabled breakthroughs for many different problems. However, many real networks are not metric. Problems with non-metric distances are considerably harder, and our current toolkit is quite limited. The overarching goal of this proposal is to develop a toolkit for non-metric distance problems and apply it to advance multiple core tasks.

The guiding idea of embeddings is to impose more ""structure"" on the input space. Completely general distances are notoriously intractable, so we target non-metrics that have some structure:

(1) Quasimetrics — shortest-path distances in directed graphs (lacking symmetry). We will develop tools such as stochastic embeddings into DAGs, locality-sensitive partial orderings, and directed low-diameter decompositions, with special focus on realistic instances that have additional structure (e.g. small highway dimension). Expected impact includes faster directed shortest paths, improved sparsest-cut approximations, compact routing, asymmetric TSP, and beyond.

(2) Hop-constrained (h.c.) distances — the u–v distance is the minimum-weight path using at most h hops. H.c. distances are important for increased reliability and lower latency. We will study stochastic embeddings into trees, embeddings into spanning trees, and embeddings of topologically structured graphs into low-treewidth graphs. Expected impact includes better (non-h.c.) oblivious routing, distributed universal optimality, h.c. compact routing, and more."

Champ scientifique (EuroSciVoc)

Ce projet n'a pas encore été classé par EuroSciVoc.
Proposez les domaines scientifiques qui vous semblent les plus pertinents et aidez-nous à améliorer notre service de classification.

Mots‑clés

Programme(s)

Thème(s)

Régime de financement

HORIZON-ERC - HORIZON ERC Grants

Voir tous les projets financés dans le cadre de ce programme de financement

Appel à propositions

(s’ouvre dans une nouvelle fenêtre) ERC-2026-STG

Voir tous les projets financés au titre de cet appel

Institution d’accueil

BAR ILAN UNIVERSITY
Contribution nette de l'UE
€ 1 499 015,00
Coût total
€ 1 499 015,00

Bénéficiaires (1)