Protein folding on a 64 qubit trapped-ion hardware via counterdiabatic quantum optimization
Alejandro Gomez Cadavid 1,2 , Pavle Nikačević 1 , Pranav Chandarana 1,2 , Sebastián V. Romero 3,4 , Enrique Solano 1 , Narendra N. Hegade 1,5, Miguel Angel Lopez-Ruiz 6 , Claudio Girotto 6 , Hanna Linn 6 , Hakan Doga 6 , Evgeny Epifanovsky 6 , Panagiotis Kl. Barkoutsos 6 , Ananth Kaushik 6 , Martin Roetteler 6 · 2026
Read Protein folding on a 64 qubit trapped-ion hardware via counterdiabatic quantum optimization by Alejandro Gomez Cadavid 1,2 , Pavle Nikačević 1 , Pranav Chandarana 1,2 , Sebastián V. Romero 3,4 , Enrique Solano 1 , Narendra N. Hegade 1,5, Miguel Angel Lopez-Ruiz 6 , Claudio Girotto 6 , Hanna Linn 6 , Hakan Doga 6 , Evgeny Epifanovsky 6 , Panagiotis Kl. Barkoutsos 6 , Ananth Kaushik 6 , Martin Roetteler 6 (2026) 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-2604-26861.txt.
Introduction
Gomez and thirteen colleagues report the largest trapped-ion demonstration of lattice protein folding to date, on a 64-qubit barium system. Six short peptide chains of 14 to 16 residues are encoded on a tetrahedral lattice, producing dense Hamiltonians with interactions among up to five bodies. A bias-field counterdiabatic method steers sampling toward low-energy shapes, and a consensus pipeline merges quantum contact data with feasible backbone geometry. The hybrid workflow reaches classical reference energies in several cases and beats random-seeded pipelines.
Catalog record
| CREATOR | Alejandro Gomez Cadavid 1,2 , Pavle Nikačević 1 , Pranav Chandarana 1,2 , Sebastián V. Romero 3,4 , Enrique Solano 1 , Narendra N. Hegade 1,5, Miguel Angel Lopez-Ruiz 6 , Claudio Girotto 6 , Hanna Linn 6 , Hakan Doga 6 , Evgeny Epifanovsky 6 , Panagiotis Kl. Barkoutsos 6 , Ananth Kaushik 6 , Martin Roetteler 6 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 26324 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.26861
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Protein folding on a 64 qubit trapped-ion hardware via counterdiabatic quantum optimization Alejandro Gomez Cadavid 1,2 , Pavle Nikačević 1 , Pranav Chandarana 1,2 , Sebastián V. Romero 3,4 , Enrique Solano 1 , Narendra N. Hegade 1,5, Miguel Angel Lopez-Ruiz 6 , Claudio Girotto 6 , Hanna Linn 6 , Hakan Doga 6 , Evgeny Epifanovsky 6 , Panagiotis Kl. Barkoutsos 6 , Ananth Kaushik 6 , Martin Roetteler 6 Abstract We report the largest trapped-ion hardware demonstration of lattice protein-folding optimization to date, using bias-field digitized counterdiabatic quantum optimization (BF-DCQO) on a fully connected 64-qubit Barium development system similar to the forthcoming IonQ Tempo line. Six peptide sequences with 14–16 amino-acid residues are encoded using a coarse-grained tetrahedral lattice model, yielding higher-order spin-glass Hamiltonians with long-range interactions involving up to five-body terms and mapped to 46–61 qubits. The resulting instances are demanding for near-term quantum hardware because low-energy configurations must satisfy backbone-geometry constraints while optimizing dense residue-contact interactions. BF-DCQO uses a non-variational bias-feedback mechanism, where low-energy samples from each round define longitudinal fields that guide subsequent quantum evolutions. Across the studied instances, BF-DCQO shifts raw sampled energy distributions toward lower energies than uniform random sampling, with the strongest improvements appearing in residue-contact variables. To preserve this signal, we introduce a consensus-based post-processing pipeline that combines quantum-learned contact information with feasible backbone geometries. The resulting hybrid workflow reaches the classical reference energy in multiple instances and improves over the corresponding random-seeded pipeline. These results show that BF-DCQO can generate structured samples for dense protein-folding Hamiltonians at previously unexplored trapped-ion scales. Keywords—Protein folding, Counterdiabatic Quantum Algorithms, Trapped-Ion Hardware I Introduction Predicting the three-dimensional structure of a protein from its amino acid sequence, known as the protein folding problem, remains one of the grand challenges in computational biology [ 14 , 1 ] . The configurational space grows exponentially with chain length, and finding the minimum-energy conformation is NP-hard even on simplified lattice models. While deep-learning approaches have achieved remarkable accuracy for known protein families, the underlying combinatorial optimization naturally maps to formulations amenable to quantum computation [ 26 , 30 ] . Lattice models reduce the continuous folding problem to a discrete optimization over turn sequences on a regular lattice, producing Ising Hamiltonians that quantum hardware can target directly. Reference [ 26 ] demonstrated quantum annealing on small peptides using D-Wave hardware, while Ref. [ 2 ] explored coarse-grained models on quantum annealers. Reference [ 30 ] introduced a tetrahedral lattice encoding with physically realistic Miyazawa-Jernigan interaction potentials, capturing some important features of protein geometry while producing gate-model-compatible Ising Hamiltonians. Subsequent research branched in several directions. Algorithmic alternatives, such as digitized-counterdiabatic methods [ 7 ] and later BF-DCQO, pushed hardware demonstrations up to 12 amino acids using 33 trapped-ion qubits [ 31 ] . Critical benchmarking and resource studies showed both the limitations of current quantum hardware on simplified peptide-folding tasks and the costs of existing encodings [ 4 , 20 ] . Finally, representational extensions were introduced, including alternative turn-based encodings that reached 20 amino acids using 114 superconducting qubits [ 24 ] ; higher-degree-of-freedom cubic and FCC lattices [ 18 ] ; as well as problem-agnostic approaches, which reached up to 26 amino acids using 45 superconducting qubits [ 21 ] . Variational algorithms such as the Quantum Approximate Optimization Algorithm (QAOA) [ 15 ] and the Variational Quantum Eigensolver (VQE) [ 27 ] have been the primary gate-model approaches for combinatorial optimization, but they require iterative classical-quantum feedback loops that are difficult to scale. Digitized Counterdiabatic Quantum Optimization (DCQO) [ 16 , 7 ] offers an alternative: it combines counterdiabatic (CD) driving [ 33 , 9 ] with digital quantum circuits, using analytically derived circuit parameters with no variational optimization. For short evolution times, the dynamics are dominated by the CD term, yielding shallow circuits. The bias-field extension (BF-DCQO) [ 5 , 32 ] adds an iterative refinement, where low-energy measurement outcomes feed back as longitudinal bias fields in subsequent rounds, progressively increasing initial state’s overlap with low-energy configurations. The bias update is a simple closed-form operation (later given by Eq. 7 ) that avoids gradient estimations. BF-DCQO has demonstrated strong performance on higher order binary optimization problems [ 32 ] . Here, we apply it to protein folding on trapped-ion hardware beyond the previously achieved scale (33 qubits [ 31 ] ), tackling instances up to 16 amino acids using 61 all-to-all connected qubits on a 64-qubit Barium development system similar to the forthcoming IonQ Tempo line. Trapped-ion platforms are particularly well-suited for this problem, as their all-to-all connectivity eliminates the SWAP overhead required by nearest-neighbor architectures to implement the dense Ising couplings arising from pairwise bead interactions [ 19 ] . Therefore, to the best of our knowledge, this work contains the largest protein-folding demonstration on trapped-ion hardware up to date, as well as the largest one leveraging the dense encoding originally proposed in Ref. [ 30 ] . Our contributions: (1) We present the first application of BF-DCQO to protein folding in the range of 46–61 qubits on trapped-ion hardware. (2) We demonstrate that BF-DCQO captures meaningful interaction structure, reaching the exact reference energy in 4 of the 6 studied sequences when paired with a consensus-based post-processing. How to choose the post-processing is a central question for any demonstration on current quantum hardware. Then, we also show that the consensus pipeline can preserve the quantum-learned contacts, while a more naive per-sample repair, which only re-optimizes contact qubits, can erase the structure learned by the quantum algorithm. (3) We analyze the structure of the quantum signal, finding that it concentrates in the interaction qubits, consistent with how the bias-field mechanism operates. II Problem Formulation II-A Tetrahedral Lattice Model We adopt the coarse-grained lattice protein model of Ref. [ 30 ] , ignoring side-chains. Each amino acid residue is represented by a main-chain bead placed on the vertices of a tetrahedral lattice. The protein backbone is encoded as a self-avoiding walk, in which each turn between consecutive beads is one of 4 tetrahedral directions, requiring 2 qubits per turn. The total Hamiltonian is the sum of two terms, H=H_{\text{back}}+H_{\text{contact}}, (1) where H_{\text{back}} ensures physical geometries by penalizing consecutive turns along the same axis (backtracking), and H_{\text{contact}} encodes pairwise contact interactions between backbone beads that are separated by five or more residues along the chain but come into spatial proximity due to folding, which induces an interaction defined by the Miyazawa–Jernigan potential [ 23 ] while penalizing overlapping conformations to enforce the self-avoiding walk. The constraint term H_{\text{back}} carries large penalty coefficients to ensure that the ground state of H corresponds to a physically valid conformation. The interaction contact terms ( H_{\text{contact}} ) encode the physical energetics of self-avoiding sequences, where lower contact energy indicates a more favorable folded structure. II-B Qubit Layout The total number of qubits is n_{\text{q}}=n_{\text{contact}}+n_{\text{geom}} : • Geometry qubits ( n_{\text{geom}} ): encode the turn sequence on the tetrahedral lattice. These must respect the no-backtracking constraints. • Contact qubits ( n_{\text{contact}} ): indicator variables for pairwise bead interactions. For a fixed backbone geometry, the contact configuration can be determined efficiently by inspecting the 3D structure. Our instances range from 46 to 61