Protein folding with an all-to-all trapped-ion quantum computer

Sebastián V. Romero ​ 1 , 2 {}^{\lx@orcidlink{0000-0002-4675-4452}{\orcidlogo}1,2}, Alejandro Gomez Cadavid ​ 1 , 2 {}^{\lx@orcidlink{0000-0003-3271-4684}{\orcidlogo}1,2}, Pavle Nikačević ​ 1 {}^{\lx@orcidlink{0000-0002-8832-5541}{\orcidlogo}1}, Enrique Solano ​ 1 {}^{\lx@orcidlink{0000-0002-8602-11 · 2025

Read Protein folding with an all-to-all trapped-ion quantum computer by Sebastián V. Romero ​ 1 , 2 {}^{\lx@orcidlink{0000-0002-4675-4452}{\orcidlogo}1,2}, Alejandro Gomez Cadavid ​ 1 , 2 {}^{\lx@orcidlink{0000-0003-3271-4684}{\orcidlogo}1,2}, Pavle Nikačević ​ 1 {}^{\lx@orcidlink{0000-0002-8832-5541}{\orcidlogo}1}, Enrique Solano ​ 1 {}^{\lx@orcidlink{0000-0002-8602-11 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-07866.txt.

Introduction

Sebastián Romero and co-authors folded small proteins on a quantum computer in a 2025 paper. The team used trapped-ion machines, whose qubits all connect to each other directly. Their algorithm is a digital shortcut method that steers the machine toward low-energy answers without the slow tuning loops other quantum methods need.

They folded chains of up to twelve amino acids on a tetrahedral lattice, the largest protein folding run on quantum hardware reported so far. The same method cracked hard logic puzzles at their tipping point and spin-glass problems using all thirty-six available qubits. In every case the machine found the best answer. The authors argue that the pairing of their shortcut method with fully connected ions is a strong path toward useful quantum computing.

Catalog record

CREATORSebastián V. Romero ​ 1 , 2 {}^{\lx@orcidlink{0000-0002-4675-4452}{\orcidlogo}1,2}, Alejandro Gomez Cadavid ​ 1 , 2 {}^{\lx@orcidlink{0000-0003-3271-4684}{\orcidlogo}1,2}, Pavle Nikačević ​ 1 {}^{\lx@orcidlink{0000-0002-8832-5541}{\orcidlogo}1}, Enrique Solano ​ 1 {}^{\lx@orcidlink{0000-0002-8602-11
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-07866

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Protein folding with an all-to-all trapped-ion quantum computer Sebastián V. Romero ​ 1 , 2 {}^{\lx@orcidlink{0000-0002-4675-4452}{\orcidlogo}1,2}, Alejandro Gomez Cadavid ​ 1 , 2 {}^{\lx@orcidlink{0000-0003-3271-4684}{\orcidlogo}1,2}, Pavle Nikačević ​ 1 {}^{\lx@orcidlink{0000-0002-8832-5541}{\orcidlogo}1}, Enrique Solano ​ 1 {}^{\lx@orcidlink{0000-0002-8602-1181}{\orcidlogo}1}, Narendra N. Hegade ​ 1 {}^{\lx@orcidlink{0000-0002-9673-2833}{\orcidlogo}1}, Miguel Angel Lopez-Ruiz, Claudio Girotto, Masako Yamada, Panagiotis Kl. Barkoutsos, Ananth Kaushik, Martin Roetteler Abstract We experimentally demonstrate that the bias-field digitized counterdiabatic quantum optimization (BF-DCQO) algorithm, implemented on IonQ’s fully connected trapped-ion quantum processors, offers an efficient approach to solving dense higher-order unconstrained binary optimization (HUBO) problems. Specifically, we tackle protein folding on a tetrahedral lattice for up to 12 amino acids, representing the largest quantum hardware implementations of protein folding problems reported to date. Additionally, we address MAX 4-SAT instances at the computational phase transition and fully connected spin-glass problems using all 36 available qubits. Across all considered cases, our method consistently achieves optimal solutions, highlighting the powerful synergy between non-variational quantum optimization approaches and the intrinsic all-to-all connectivity of trapped-ion architectures. Given the expected scalability of trapped-ion quantum systems, BF-DCQO represents a promising pathway toward practical quantum advantage for dense HUBO problems with significant industrial and scientific relevance. I Introduction Protein folding, a fundamental biological process, poses a grand challenge in scientific and industrial applications due to its immense medical relevance, particularly in drug discovery and disease understanding. From a computational perspective, protein folding is often framed as a search for the lowest-energy configuration on a lattice, which inherently maps to finding the ground state of a general p -spin-glass model. This formulation places it squarely within the class of Higher-Order Unconstrained Binary Optimization (HUBO) problems. Combinatorial optimization problems, including protein folding, appear frequently across numerous scientific and industrial domains, ranging from logistics and transportation to chemistry and biology. These problems typically require finding optimal or near-optimal solutions within vast configuration spaces. Many, like HUBOs encountered in protein folding, are NP-hard, meaning that, in the worst case, the computational time required to solve them scales exponentially with the system size. While various classical solvers exist, including problem-specific or general-purpose algorithms such as simulated annealing Kirkpatrick et al. 1983 , tabu search Glover et al. 1993 , CPLEX Cplex 2009 , and Gurobi Gurobi Optimization, LLC 2023 , they often struggle with the non-convex energy landscapes characteristic of these complex problems, particularly in higher-order binary optimization Bertsekas et al. 2003 ; Floudas et al. 2005 ; Burer and Letchford 2012 ; Arjevani et al. 2020 ; Danilova et al. 2022 ; Xu and Liberti 2024 . These optimization problems can be reformulated as energy minimization tasks over spin-glass models, whose ground states encode the optimal solutions Lucas 2014 . Quadratic unconstrained binary optimization (QUBO) problems naturally map to Ising models, while their higher-order (HUBO) counterparts can be mapped into general p -spin-glasses Gardner 1985 , also known as higher-order Ising models in quantum optimization Pelofske et al. 2023a ; Pelofske et al. 2024 . As problem complexity increases and classical resources become limiting, quantum computing emerges as a promising alternative. Quantum algorithms such as adiabatic quantum optimization (AQO) Albash and Lidar 2018 and the quantum approximate optimization algorithm (QAOA) Farhi et al. 2014 , a digitized form of adiabatic quantum computing Barends et al. 2016 , offer a new approach to tackle computationally intractable optimization problems Abbas et al. 2024 ; Boulebnane and Montanaro 2024 ; Kotil et al. 2025 ; Koch et al. 2025 . Breakthrough developments in quantum technologies are making quantum computers increasingly capable of addressing real-world optimization problems. In parallel, quantum algorithms have been proposed with the potential for theoretical speedups over classical solvers Durr and Hoyer 1999 ; Somma et al. 2008 ; Wocjan and Abeyesinghe 2008 ; Hastings 2018 ; Montanaro 2018 ; Montanaro 2020 ; Chakrabarti et al. 2022 ; Dalzell et al. 2023 . QAOA, in particular, has garnered attention as a versatile technique for large-scale optimization problems, showing compelling evidence of quantum advantage in combinatorial optimization. For example, benchmark experiments on 8-SAT instances up to 20 variables Boulebnane and Montanaro 2024 and the low autocorrelation binary sequence (LABS) problem Boehmer 1967 ; Schroeder 1970 ; Shaydulin et al. 2024 have demonstrated that QAOA can outperform classical solvers in terms of time to solution, highlighting the practical relevance of quantum algorithms in optimization. However, HUBO problems are considerably harder to solve than QUBO problems and frequently appear in applications like the LABS problem Shaydulin et al. 2024 ; Koch et al. 2025 , factorization Hegade et al. 2021a , and satisfiability problems Battiti 2009 . Despite recent experimental implementations of HUBO problems on quantum hardware Pelofske et al. 2023a ; Pelofske et al. 2024 ; Barron et al. 2024 , their implementation is particularly challenging due to current hardware limitations, such as qubit connectivity, decoherence, and gate fidelities. Recent advances, specifically targeting the protein folding problem, include quantum annealing Perdomo et al. 2008 ; Perdomo-Ortiz et al. 2012 ; Babbush et al. 2014 ; Babej et al. 2018 and variational quantum algorithms Robert et al. 2021 ; Chandarana et al. 2023 ; Pamidimukkala et al. 2024 ; Wang and Zhou 2024 . The limitations of these models include the approximate treatment of chemical interactions and little flexibility in the possible geometric structure since the amino acid positions are constrained to the imposed lattice. Nevertheless, as long as the obtained optimal local structure of a studied protein domain (or an oligopeptide) is within the “funnel” leading to the global minimum, the accurate folded protein structure can be easily reached using the standard atomistic geometry optimization techniques starting from the obtained structure as the initial geometry. Additionally, the high density of the associated p -spin-glass makes it challenging to implement time-evolution-based quantum algorithms on limited connectivity architectures due to swap overheads. Counterdiabatic (CD) protocols Demirplak and Rice 2003 ; Berry 2009 ; Chen et al. 2010 ; del Campo 2013 ; Sels and Polkovnikov 2017 ; Claeys et al. 2019 ; Takahashi and del Campo 2024 , designed to suppress undesired diabatic transitions, have shown superior performance in optimization tasks. In addition to digitized counterdiababtic QAOA Chandarana et al. 2022 and digitized counterdiababtic quantum optimization algorithm (DCQO) Hegade et al. 2022 , as well as the recent branch-and-bound DCQO Simen et al. 2025 , the bias-field DCQO (BF-DCQO) algorithm Cadavid et al. 2025 ; Romero et al. 2024 ; IBM Quantum 2025 combines concepts from quantum annealing with dynamically updated bias fields Graß 2019 ; Grass 2022 . BF-DCQO evolves the system towards optimal solutions without relying on variational methods that can suffer from barren plateaus Cerezo et al. 2024 ; Larocca et al. 2025 , resulting in a purely quantum and scalable algorithm for HUBO problems Romero et al. 2024 . Recently, it has been demonstrated that BF-DCQO exhibits a runtime quantum advantage compared to classical solvers in achieving approximate solutions for specific HUBO problems Chandarana et al. 2025 . In this work, we combine the BF-DCQO algorithm’s resource efficiency and fast convergence with IonQ’s trapped-ion hardware, which offers all-to-all connectivity and high gate fidelities IonQ 2025 . Using this setup, we tackle the protein folding problem for sequences of up to 12 amino acids mapped onto 33 qubits. To further

Read the complete text