Ohyun Kwon | Searcharxiv

arXiv2025

Towards Quantum Advantage in Chemistry

Molecular simulations are widely regarded as leading candidates to demonstrate quantum advantage--defined as the point at which quantum methods surpass classical approaches in either accuracy or scale. Yet the qubit counts and error rates required to realize such an advantage remain uncertain; resource estimates for ground-state electronic structure span orders of magnitude, and no quantum-native method has been validated at a commercially relevant scale. Here we address this uncertainty by executing the iterative qubit coupled-cluster (iQCC) algorithm, designed for fault-tolerant quantum hardware, at unprecedented scale using a quantum solver on classical processors, enabling simulations of transition organo-metallic complexes requiring hundreds of logical qubits and millions of entangling gates. Using this approach, we compute the lowest triplet excited state (T$_1$) energies of Ir(III) and Pt(II) phosphorescent organometallic compounds and show that iQCC achieves the lowest mean absolute error (0.05 eV) and highest R$^2$ (0.94) relative to experiment, outperforming leading classical methods. We find these systems remain classically tractable up to $\sim$200 logical qubits, establishing the threshold at which quantum advantage in computational chemistry may emerge and clarifying resource requirements for future quantum computers.

physics.chem-ph↗

arXiv2022

Image-to-Graph Transformers for Chemical Structure Recognition

For several decades, chemical knowledge has been published in written text, and there have been many attempts to make it accessible, for example, by transforming such natural language text to a structured format. Although the discovered chemical itself commonly represented in an image is the most important part, the correct recognition of the molecular structure from the image in literature still remains a hard problem since they are often abbreviated to reduce the complexity and drawn in many different styles. In this paper, we present a deep learning model to extract molecular structures from images. The proposed model is designed to transform the molecular image directly into the corresponding graph, which makes it capable of handling non-atomic symbols for abbreviations. Also, by end-to-end learning approach it can fully utilize many open image-molecule pair data from various sources, and hence it is more robust to image style variation than other tools. The experimental results show that the proposed model outperforms the existing models with 17.1 % and 12.8 % relative improvement for well-known benchmark datasets and large molecular images that we collected from literature, respectively.

cs.CV↗