Geometric Complexity Theory: Introduction

These are lectures notes for the introductory graduate courses on geometric complexity theory (GCT) in the computer science department, the university of Chicago. Part I consists of the lecture notes for the course given by the first author in the spring quarter, 2007. It gives introduction to the basic structure of GCT. Part II consists of the lecture notes for the course given by the second author in the spring quarter, 2003. It gives introduction to invariant theory with a view towards GCT. No background in algebraic geometry or representation theory is assumed. These lecture notes in conjunction with the article \cite{GCTflip1}, which describes in detail the basic plan of GCT based on the principle called the flip, should provide a high level picture of GCT assuming familiarity with only basic notions of algebra, such as groups, rings, fields etc.

💡 Research Summary

The document is a two‑part set of lecture notes that together form a comprehensive introduction to Geometric Complexity Theory (GCT). Part I, authored by Ketan D. Mulmuley, presents the overall architecture of GCT, while Part II, by Milind Sohoni, supplies the invariant‑theoretic background needed for the program. The notes assume only elementary algebra (groups, rings, fields) and basic computational complexity, deliberately avoiding any prerequisite in algebraic geometry or representation theory.

The core idea of GCT is to translate lower‑bound questions in complexity theory—most famously the P versus NP problem—into upper‑bound decision problems in algebraic geometry and representation theory, a meta‑logical maneuver called the “flip”. In characteristic 0 (the integer setting) the authors associate to each complexity class a family of algebraic varieties, called class varieties: χ_P for P and χ_NP for NP. These varieties carry a natural action of the general linear group G = GL_n(ℂ). The coordinate rings R_P and R_NP become G‑algebras, and their homogeneous components R_P(d) and R_NP(d) are finite‑dimensional G‑representations.

If NP were contained in P, then χ_NP would embed as a G‑subvariety of χ_P, inducing a surjection of coordinate rings and, by complete reducibility of GL_n(ℂ) representations (Weyl’s theorem), an embedding of each R_NP(d) into R_P(d) as a G‑subrepresentation. An “obstruction” of degree d is defined as an irreducible G‑representation that appears in R_NP(d) but not in R_P(d). The existence of an obstruction for every sufficiently large d would contradict the embedding, thereby proving NP ⊄ P in characteristic 0.

The “GCT flip” conjecture (GCT‑flip1) asserts that, although constructing an obstruction seems as hard as proving a lower bound, there should exist a polynomial‑time algorithm that, given n, produces an obstruction showing χ_NP(n,m) cannot embed in χ_P(n,m) (with m≈n log n). Thus the hard lower‑bound problem is reduced to an apparently easier constructive problem.

To make this program concrete, the notes develop a substantial amount of representation theory:

• Finite groups, the symmetric group S_n, and the general linear group GL_n(ℂ) are introduced from scratch.
• Basic constructions (induction, restriction, tensor products) and the complete reducibility of compact and reductive groups are proved via Weyl’s unitary trick.
• Detailed treatment of S_n representations using Young diagrams, Specht modules, and the Frobenius character formula is given, leading to the definition of Littlewood–Richardson coefficients.
• GL_n(ℂ) representations are built via highest‑weight vectors and Weyl’s character formula, with an alternative approach using Weyl’s unitary trick.

The notes then focus on the combinatorial constants that appear in the obstruction problem:

• Littlewood–Richardson coefficients, Kronecker coefficients, and plethysm coefficients are introduced as multiplicities in tensor product, inner product, and composition of representations, respectively.
• Decision problems concerning the non‑vanishing of these coefficients are examined. The “stretching function” technique and saturation theorems are discussed, showing that for certain groups (e.g., GL_n) the non‑vanishing is governed by linear inequalities.
• Positive integer programming and saturated integer programming are presented as algorithmic frameworks that could decide these non‑vanishing questions in polynomial time.

Algebraic geometry basics are then covered: varieties, orbit closures, Grassmannians, and the definition of class varieties. The notion of an obstruction is formalized in geometric terms, and the “second fundamental theorem” and Borel–Weil theorem are invoked to argue why obstructions should exist.

A substantial portion of the notes is devoted to quantum groups. Hopf algebras, the q‑analogue of the unitary group U_q, and the standard quantum group are defined. The authors explain how crystal bases and crystal operators give a combinatorial model for quantum group representations. Using these tools, they propose a “positivity hypothesis”: the structural constants (Littlewood–Richardson, Kronecker, plethysm) should admit positive formulas (no alternating signs) when expressed in the language of quantum groups. This hypothesis links the GCT program to deep number‑theoretic conjectures, notably the Riemann hypothesis over finite fields, via works of Deligne, Katz, and others.

Part II (Sohoni’s contribution) revisits invariant theory with a computational perspective: finite group actions, the symmetric group, the special linear group SL_n, and affine algebraic group actions. Concepts such as orbits, invariants, the null‑cone, destabilizing flags, and stability (in the sense of Geometric Invariant Theory) are introduced, providing the language needed to discuss class varieties and their orbit closures.

Overall, the lecture notes lay out a roadmap:

1. Model complexity classes as G‑varieties.
2. Translate lower‑bound statements into the existence of representation‑theoretic obstructions.
3. Conjecture that these obstructions can be constructed efficiently (the flip).
4. Reduce the construction to positivity statements about structural constants.
5. Approach positivity via quantum group representation theory and related number‑theoretic conjectures.

If each step can be realized, GCT would provide a non‑relativizing, non‑naturalizing proof that NP ≠ P (in characteristic 0). The notes are both a pedagogical introduction and a research agenda, outlining the mathematical machinery required and indicating where the major open problems lie.