Lovász theta function

Lovász theta function
Name	Lovász theta function
Othernames	theta number, Schmüdgen–Lovász theta
Introduced	1979
Introducedby	László Lovász
Field	Combinatorics; Graph theory; Semidefinite programming
Applications	Graph coloring; Shannon capacity; Quantum information; Coding theory

Lovász theta function The Lovász theta function is a graph parameter introduced by László Lovász in 1979 that provides an efficiently computable bound on the independence number and chromatic number of a graph while tightly relating to the Shannon capacity problem proposed by Claude Shannon. It occupies a central role connecting combinatorial optimization, semidefinite programming studied by Michel Goemans and Laurent Vandenberghe, and information-theoretic topics developed by Richard Hamming and David Slepian. The theta function is notable for yielding exact values for several classical graphs studied by Paul Erdős, Václav Chvátal, and William Tutte.

Definition and basic properties

The Lovász theta function θ(G) for a finite simple graph G is defined as the optimal value of a semidefinite program associated to an orthonormal representation of the complement graph and a unit vector, a framework that ties to work by Noga Alon and Mihalis Yannakakis. Fundamental inequalities include α(G) ≤ θ( Ḡ ) ≤ χ̄(G) where α denotes the independence number studied by Erdős–Ko–Rado collaborators and χ̄ the fractional chromatic number related to results of Hans-Jakob Zassenhaus and Rudolf Mathon. The theta function is multiplicative under the strong product, a property leveraged in analyses by Claude Shannon and Harold Kroto when exploring zero-error communication and combinatorial capacities. It is monotone under induced subgraphs, invariant under graph isomorphism considered by Frank Harary, and polynomial-time computable via interior-point methods developed by Niels Karmarkar and Karmarkar's contemporaries.

Equivalent formulations

θ(G) admits multiple equivalent formulations: as a semidefinite program, as the minimum of the maximum reciprocal squared inner products over orthonormal representations of the complement, and as the maximum of a trace objective over positive semidefinite matrices constrained by adjacency, reflecting duality principles central to Karmarkar and Yannakakis's work. These formulations connect to spectral bounds like the Hoffman bound advanced by Alan Hoffman and to matrix completion problems examined by Paul Erdős collaborators. Duality yields expressions involving weighted adjacency matrices and eigenvalue optimization reminiscent of studies by Issai Schur and David Hilbert on quadratic forms.

Computation and algorithms

Computationally, θ(G) is solvable in polynomial time by semidefinite programming solvers that implement interior-point algorithms from Niels Karmarkar, Arkadi Nemirovski, and Yurii Nesterov. Practical computation uses software libraries inspired by numerical linear algebra pioneered by Gene Golub and J. H. Wilkinson, and optimization toolboxes influenced by Michael Grant and Stephen Boyd. For very large graphs, approximation methods combine randomized rounding techniques due to Prabhakar Raghavan and Umesh Vazirani and spectral heuristics developed by Fan Chung and László Babai; these yield scalable bounds though with weaker guarantees than exact SDP solutions of J. F. Bonnans style.

Applications

The theta function bounds the Shannon capacity of graphs, informing zero-error information theory rooted in Claude Shannon and expanded by Richard Ahlswede and János Körner. It provides efficient bounds for graph coloring problems investigated by Paul Erdős and Endre Szemerédi and facilitates approximation algorithms linked to the Max-Cut problem studied by Michel Goemans and David Williamson. In quantum information, θ(G) appears in nonlocal games and entanglement-assisted capacity results developed by John Bell-inspired researchers and Charles Bennett’s colleagues. Coding theory applications trace to Richard Hamming and Elias, where theta-based bounds limit code sizes for graphs modeling symbol confusability.

Relations to other graph parameters

θ(G) interpolates between classical parameters: it upper-bounds the independence number α(G) associated with extremal work by Paul Erdős and lower-bounds variants of the chromatic number χ(G) studied by George Pólya and Richard Stanley. It relates tightly to the fractional chromatic number χ_f(G) investigated by Edward O. Thorp and to the vector chromatic number introduced in spectral graph theory by Fan Chung and Béla Bollobás. Connections to eigenvalue bounds involve the adjacency and Laplacian spectra explored by Alfredo Hoffman and Miroslav Fiedler, while ties to semidefinite relaxations place θ(G) alongside hierarchies proposed by Jean-Bernard Lasserre and Margaret Wright.

Examples and notable values

For the odd cycle C_5 studied by Tibor Gallai and Paul Erdős, θ(C_5) equals √5, a celebrated exact computation by László Lovász that resolved a special Shannon capacity case. For complementary graphs like the Kneser graphs analyzed by Martin Kneser and László Lovász himself in his earlier topological combinatorics work, theta yields tight bounds that match chromatic results of László Lovász’s theorem on Kneser chromatic number. For complete graphs K_n and edgeless graphs, θ attains trivial values n and 1 respectively, aligning with classical results by Arthur Cayley and William Tutte on graph extremal structures. Many strongly regular graphs cataloged by Andries E. Brouwer and J. H. van Lint admit closed-form theta values obtainable from their eigenvalue parameters.

Category:Graph theory