Draw2Think: Harnessing Geometry Reasoning through Constraint Engine Interaction

TL;DR

Draw2Think introduces a novel framework that externalizes geometric reasoning onto an interactive constraint engine, enabling verifiable, accurate, and auditable geometric problem solving.

Contribution

It recasts geometric reasoning into an agentic interaction with GeoGebra, enabling exact constraint verification and improved accuracy over prior latent methods.

Findings

Achieves 95.9% predicate-level construction correctness on GeoGoal.

Improves outcome accuracy by up to 16.4% on benchmarks.

Attains 68.2% strict rendering scores on GenExam-math.

Abstract

Vision-language models solve geometry problems with rising accuracy, yet their intermediate states remain latent and unverifiable: a relation expressed in textual reasoning or drawing code carries no guarantee that a constraint-satisfying configuration realizes it. We observe that existing externalization methods based on rendered pixels or one-shot scripts fail to provide exact, per-action geometric guarantees. Enforcing geometric relations by algebraic definition closes this gap: the workspace becomes a constraint-checked evolving canvas. We present Draw2Think, a framework that recasts geometric reasoning from latent spatial inference into agentic interaction with the GeoGebra constraint engine. In a Propose-Draw-Verify loop, Draw2Think externalizes hypotheses onto an executable canvas, measures exact geometric quantities, and feeds structured observations back to the model, so subsequent reasoning proceeds from checked canvas state grounded by the shared workspace. This externalization makes two properties separately auditable: model-level Construction Fidelity (whether the canvas realizes the intended configuration) and engine-level Measurement Faithfulness (exact values and relations from canvas constraints). Across construction, outcome, and rendering evaluations, Draw2Think builds canvases that pass 95.9% predicate-level and 84.0% strict problem-level construction checks on GeoGoal, improves outcome accuracy by up to 4.1%/16.4% on planar/solid benchmarks, and attains 68.2%/90.5% strict/relaxed rendering scores on GenExam-math. Project page is available at https://draw2think.github.io/