Separation Results for Constant-Depth and Multilinear Ideal Proof Systems
Abstract
In this work, we establish separation theorems for several subsystems of the Ideal Proof System (IPS), an algebraic proof system introduced by Grochow and Pitassi (J. ACM, 2018). Separation theorems are well-studied in the context of classical complexity theory, Boolean circuit complexity, and algebraic complexity. In an important work of Forbes, Shpilka, Tzameret, and Wigderson (ToC, 2021), two proof techniques were introduced to prove lower bounds for subsystems of the IPS, namely the functional method and the multiples method. We use these techniques and obtain the following results. Hierarchy theorem for constant-depth IPS: Recently, Limaye, Srinivasan, and Tavenas (J. ACM 2025) proved a hierarchy theorem for constant-depth algebraic circuits. We adapt the result and prove a hierarchy theorem for constant-depth IPS. We show that there is an unsatisfiable multilinear instance refutable by a depth- IPS such that any depth-(/10) IPS refutation for it must have superpolynomial size. This result is proved by building on the multiples method. Separation theorems for multilinear IPS: In an influential work, Raz (ToC, 2006) unconditionally separated two algebraic complexity classes, namely multilinear NC1 from multilinear NC2. In this work, we prove a similar result for a well-studied fragment of multilinear-IPS. Specifically, we present an unsatisfiable instance such that its functional refutation, i.e., the unique multilinear polynomial agreeing with the inverse of the polynomial over the Boolean cube, has a small multilinear-NC2 circuit. However, any multilinear-NC1 IPS refutation (IPSLIN) for it must have superpolynomial size. This result is proved by building on the functional method.
Turn this paper into a full lesson
ArcXiv compiles a staged curriculum from this paper: 8-12 lessons across beginner → advanced, synthesised section guides, visuals, flashcards, a quiz, exercises, and on-demand deep dives per section. Grounded in the abstract, never invented.