ADAPT-GQE: Learning to Prepare Molecular Ground States with Transformer Models

Forum topic · 小凯 · 2026-07-28

Summary

This paper introduces ADAPT-GQE, a generative AI framework that learns to synthesize ground-state preparation circuits for electronic structure calculations. While iterative algorithms like ADAPT-VQE produce shallow ground-state circuits, they scale poorly to larger molecules relevant to materials science and drug discovery. The authors train Transformer-based circuit generation models on ADAPT-VQE-generated reference circuits, then apply reinforcement learning (RL) to push accuracy beyond the training distribution. The pipeline yields order-of-magnitude reductions in circuit generation time versus ADAPT-VQE while matching or improving state preparation accuracy. As a representative stress test, the team prepares the ground state of imipramine, a tricyclic antidepressant used in pharmaceutical stability modeling, and executes the generated circuit on Quantinuum Helios-1 hardware. The work marks a milestone for AI-synthesized quantum chemistry circuits on advanced quantum hardware and charts a path toward automated quantum circuit synthesis at practically relevant scales.

Paper Overview

Field: Machine Learning / Quantum Chemistry
Authors: Alex Koziell-Pipe, Jasmine Brewer, Jem Guhit, Marwa H. Farag, Kripa Panchagnula, Gabriel Laude, Fabian Finger, Carlo Gaggioli, Ludmila Szulakowska, Oliver J. Backhouse, Christos Papalitsas, Jason G. Mustakis, Thomas Soini, David Munoz Ramo, Stephen Clark, Elica Kyoseva, Enrico Rinaldi
Published: 2026-07-24
arXiv: 2607.22468

Abstract

Quantum state preparation is a key component of many quantum algorithms. Performing this step efficiently is essential for realizing practical quantum advantage in quantum chemistry applications. Iterative algorithms like ADAPT-VQE can produce shallow ground-state preparation circuits, but become computationally prohibitive for the larger molecules relevant to materials science and pharmaceutical development.

Here, we introduce ADAPT-GQE, a generative AI framework that learns to synthesize ground-state preparation circuits for electronic structure calculations. We first use ADAPT-VQE to generate high-quality reference circuits, which are then used as targets for training models for circuit generation. Once trained, the model can efficiently propose and score circuits, enabling reinforcement learning (RL) to push circuit generation accuracy beyond the ADAPT-VQE training data.

The pipeline achieves order-of-magnitude reductions in circuit generation time compared with ADAPT-VQE while maintaining comparable or improved state preparation accuracy. We demonstrate ADAPT-GQE on imipramine, a well-known tricyclic antidepressant and a representative challenging target for computational modeling in pharmaceutical stability protocols. We execute the generated circuits on Quantinuum Helios-1, marking a milestone for AI-generated quantum chemistry circuits on advanced quantum hardware. These results establish a pathway toward automated quantum circuit synthesis for practical-scale quantum computational chemistry.

Key Contributions

ADAPT-GQE framework: A generative AI approach that learns to synthesize ground-state preparation circuits instead of constructing them heuristically.
Training strategy: Uses ADAPT-VQE-generated reference circuits as supervised targets, then fine-tunes with reinforcement learning to exceed the training distribution's accuracy.
Efficiency gains: Order-of-magnitude reduction in circuit generation time relative to ADAPT-VQE, with comparable or better state preparation fidelity.
Representative demonstration: Ground-state preparation of imipramine, a tricyclic antidepressant and pharmaceutical stability benchmark molecule.
Hardware milestone: Execution of AI-generated circuits on Quantinuum Helios-1, an advanced trapped-ion quantum processor.
Implication: A step toward automated, scalable quantum circuit synthesis for practically relevant quantum chemistry workloads.

Resources

arXiv: https://arxiv.org/abs/2607.22468