[Paper Review] Flag Algebras: A First Glance

Marcel K. de Carli Silva, Fernando Mário de Oliveira Filho+1 more|arXiv (Cornell University)|Jul 16, 2016

Limits and Structures in Graph Theory6 references3 citations

TL;DR

This paper introduces flag algebras as a computational framework for solving extremal combinatorics problems, such as Mantel’s theorem, by formulating them as semidefinite programs. It enables automated, computer-assisted derivation of tight upper bounds on extremal graph parameters through conic relaxations of limit functionals.

ABSTRACT

The theory of flag algebras, introduced by Razborov in 2007, has opened the way to a systematic approach to the development of computer-assisted proofs in extremal combinatorics. It makes it possible to derive bounds for parameters in extremal combinatorics with the help of a computer, in a semi-automated manner. This article describes the main points of the theory in a complete way, using Mantel's theorem as a guiding example.

Motivation & Objective

• To present a systematic, computationally tractable approach to extremal combinatorics problems using flag algebras.
• To demonstrate how flag algebras can derive tight upper bounds on graph parameters, such as the maximum edge density in triangle-free graphs.
• To show that the theory allows semi-automated proofs of extremal results by reducing them to optimization over conic relaxations.
• To illustrate the method using Mantel’s theorem as a guiding example, showing how the bound of 1/2 for triangle-free graphs is recovered computationally.
• To establish a bridge between abstract extremal combinatorics and practical computation via semidefinite programming.

Proposed method

• Formalizes extremal problems as optimization over limit functionals that capture asymptotic subgraph densities.
• Relaxes the complex set of limit functionals Φ into a computationally tractable conic relaxation using non-negative combinations of flag densities.
• Expresses the relaxation as a semidefinite program by representing constraints via positive semidefinite matrices and linear inequalities.
• Uses flag densities (e.g., of 3-vertex subgraphs) to construct linear constraints on the objective function and matrix variables.
• Applies the duality and feasibility conditions of semidefinite programming to verify that λ = 1/2 is an upper bound for triangle density.
• Employs conic combinations of flags and positive semidefinite matrices to generate valid inequalities that tighten the relaxation.

Experimental results

Research questions

• RQ1 Can extremal graph theory problems be systematically reduced to computable optimization problems?
• RQ2 How can one obtain tight upper bounds on subgraph densities in H-free graphs using computational methods?
• RQ3 What is the role of flag algebras in transforming abstract extremal problems into semidefinite programs?
• RQ4 To what extent can the flag algebra method reproduce known extremal results like Mantel’s theorem automatically?
• RQ5 How can the trade-off between computational tractability and bound quality be managed in flag algebra relaxations?

Key findings

• The flag algebra method successfully recovers Mantel’s theorem, proving that the maximum edge density in a triangle-free graph is at most 1/2.
• A feasible solution with λ = 1/2 and Q = (1/2) * [[1, -1], [-1, 1]] satisfies all semidefinite constraints, confirming the bound.
• The method transforms the extremal problem into a semidefinite program with linear constraints derived from 3-vertex flags.
• The approach is generalizable beyond graphs, extending to colored graphs, permutations, and other combinatorial structures.
• The framework enables automated, computer-assisted proofs of extremal results, as demonstrated by applications to triangle and pentagon counts.
• Complementary slackness and other optimization techniques in the flag algebra framework can reveal structural properties of extremal sequences.