Resolution Width Lifts to Near-Quadratic-Depth Res($\oplus$) Size by Dmitry Itsykson

Resolution Width Lifts to Near-Quadratic-Depth Res($\oplus$) Size by Dmitry Itsykson

🎙 Dmitry Itsykson 👥 74K 📅 September 28, 2026 ⏱ 41 min 👁 3 📄 original study 🧭 2026-09-28
Available in: English (current) Français

Keywords

resolutionwidthdepthliftingparity

Summary

In this lecture, Dmitry Itsykson presents joint work with Vladimir Podolskii and Alexander Sheratov on the relationship between resolution width and the depth of resolution over parity (Res(⊕)). The talk begins by defining Res(⊕) and its parameters: size, width, and depth. It then states two main results: first, a pure quadratic size lower bound for Res(⊕) refutations of lifted formulas, derived from the resolution width of the original unsatisfiable CNF. Second, and more central, a lifting theorem showing that any Res(⊕) refutation of a lifted formula with size S must have depth at least W²/log S, where W is the resolution width of the base formula. This implies depth lower bounds of n^(2−ε) for formulas with near-exponential size refutations. The proof uses a combinatorial game characterisation of resolution width and a novel notion of ‘H-good’ affine restrictions. A subsequent result shows that depth n^(2−ε) Res(⊕) does not p-simulate resolution, and that any proof system that p-simulates regular resolution and is p-simulated by such bounded-depth Res(⊕) is NP-hard to automate. The argument relies on constructing a distribution of paths in the refutation and analysing forced vs free moves to control semantic implications. The talk concludes with open questions on the possibility of improving the bounds and the relationship between bounded-depth and full Res(⊕).

212 words

Critical Evaluation

Value of the Information & Strength of the Argument

The results are of significant value in proof complexity, providing concrete lower bounds that connect resolution width to depth in the setting of parity resolution. The first result offers a clean quadratic size bound, while the main lifting theorem is technically deep, introducing a robust framework based on structured restrictions and closure properties. The proof strategy is well-argued: it progressively builds a distribution over paths, using game-theoretic arguments to ensure that the final node is ’easy to kill’ while any particular semantic consequence is unlikely. The argumentation is rigorous, with careful attention to the definitions of safe matrices, closures, and H-good restrictions. The presentation is concise but self-contained for an expert audience, and the logical flow from the key lemma to the final theorems is clear.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is high: the talk relies on established results (e.g., Ben-Sasson’s width-size relation, Atserias-Dalmau games) without extensive citation within the talk, but the underlying work is presumably peer-reviewed. The only explicit source provided is the program page of the ICTS workshop, which serves as context rather than a direct citation of the paper. The title exactly matches the content: the talk is entirely devoted to proving that resolution width ’lifts’ to near-quadratic depth for Res(⊕). No irrelevant topics are introduced. The lack of external references in the description is a minor limitation, but the technical depth and coherence of the presentation compensate.

245 words

Title / Content Match

The title accurately describes the content: the talk demonstrates that resolution width lifts to near-quadratic depth lower bounds for resolution over parity (Res(⊕)).

Quality & Reliability

8/10

The talk presents rigorous theoretical results with proof sketches, based on a joint paper likely peer-reviewed. The presentation is formal and technically precise, although no external verification is provided.

Key Moments

Cited Sources

Contribution & Novelties

The lecture presents original results: a quadratic size lower bound for Res(⊕) via a lifting argument, and a stronger lifting of resolution width to depth lower bounds. The novel technical tools include ‘H-good’ restrictions and a path distribution argument that control semantic implications. The result that bounded-depth Res(⊕) does not simulate resolution, and the connection to automation hardness, are significant contributions to proof complexity.

Pour aller plus loin :

  • Resolution (logic) — Foundational system in proof complexity; essential background for the talk.
  • Proof complexity — Overview of the field and its main questions, including the Cook-Reckhow program.
  • Communication complexity — The talk mentions lifting from communication complexity; this concept underpins the technical machinery.

113 words

Radar Profile

The profile shows a very high technical level (9) and good quality (8) with moderate information quantity (7). This reflects a dense, specialized research talk aimed at an expert audience, where depth of content exceeds breadth.

Reliability 8/10