DiscoverDCP: A Data-Driven Approach for Construction of Disciplined Convex Programs via Symbolic Regression

TL;DR

DiscoverDCP is a novel data-driven framework that combines symbolic regression with DCP rules to automatically identify convex models, ensuring global convexity and interpretability for control and optimization applications.

Contribution

It introduces a method that guarantees convexity by construction, enabling the discovery of more flexible and accurate convex models compared to traditional fixed forms.

Findings

Produces interpretable convex models for control tasks

Ensures global convexity without post-hoc verification

Finds more accurate convex surrogates than traditional methods

Abstract

We propose DiscoverDCP, a data-driven framework that integrates symbolic regression with the rule sets of Disciplined Convex Programming (DCP) to perform system identification. By enforcing that all discovered candidate model expressions adhere to DCP composition rules, we ensure that the output expressions are globally convex by construction, circumventing the computationally intractable process of post-hoc convexity verification. This approach allows for the discovery of convex surrogates that exhibit more relaxed and accurate functional forms than traditional fixed-parameter convex expressions (e.g., quadratic functions). The proposed method produces interpretable, verifiable, and flexible convex models suitable for safety-critical control and optimization tasks.