IBM Research - Tokyo

Hybrid quantum-classical algorithms offer a promising pathway to address computational challenges in domains such as chemistry, physics, and materials science. In this work, we present a hybrid embedding framework to simulate the dative interaction between a lithium ion and a pyridine molecule. This represents the first implementation of quantum hardware for evaluating the binding energy of a non-covalent dative-bonded complex. Due to the current limitations of classical simulators and noisy intermediate-scale quantum (NISQ) devices—typically restricted to circuits with a few dozen qubits—we employ an embedding strategy where a chemically active region is treated on a quantum processor, and the remainder is modeled classically. Without error mitigation, results from quantum hardware show notable deviations from exact energies. To address this, we implement two combined error mitigation strategies, CEM1 and CEM2, both of which significantly improve total energy estimates. This work highlights the practical potential of variational quantum algorithms for studying complex molecular systems relevant to catalysis, hydrogen storage, and battery technologies.

This paper addresses few-shot semantic segmentation (FSS) guided by text, where we classify unseen novel classes using image and text references as in-context examples, without the need for training. We enhance the quality and stability of the segmentation masks generated by FSS by combining the capability of open-vocabulary zero-shot semantic segmentation (ZSS) based on foundation models for image and text. We propose a training-free approach using multimodal feature matching that performs segmentation by identifying regions in a target image that match the features from both the image and text references. Experimental results demonstrate that the proposed method outperforms state-of-the-art FSS and ZSS methods.

Quantum-selected configuration interaction (QSCI) is an approach for quantum chemical calculations using current quantum computers. In conventional QSCI, Slater determinants used for the wave function expansion are sampled by iteratively performing approximate wave function preparation and subsequent measurement in the computational basis, and then the subspace Hamiltonian matrix is diagonalized on a classical computer. In this approach, preparation of a high-quality approximate wave function is necessary to accurately compute total energies. Here we propose a Hamiltonian simulation-based QSCI (HSB-QSCI) to avoid this difficulty, by sampling the Slater determinants from quantum states generated by the real-time evolution of approximate wave functions. We provide numerical simulations for the lowest spin-singlet and triplet states of oligoacenes (benzene, naphthalene, and anthracene), phenylene-1,4-dinitrene, and hexa-1,2,3,4,5-pentaene. We found that the HSB-QSCI is applicable not only to molecules where the Hartree-Fock provides a good approximation of the ground state, but also to strongly correlated systems where preparing a high-quality approximate wave function is hard. Hardware demonstrations of the HSB-QSCI are also reported for carbyne molecules expressed by up to 36 qubits, using an IBM quantum processor. The HSB-QSCI captures more than 99.18% of the correlation energies in the active space by considering about 1% of all the Slater determinants in 36 qubit systems, illustrating the ability of the proposed method to efficiently consider important electronic configurations.

Efficiently solving combinatorial optimization problems (COPs) such as Max-Cut is challenging because the resources required increase exponentially with the problem size. This study proposes a hardware-friendly method for solving the Max-Cut problem by implementing a spiking neural network (SNN)-based Boltzmann machine (BM) in neuromorphic hardware systems. To implement the hardware-oriented version of the spiking Boltzmann machine (sBM), the stochastic dynamics of leaky integrate-and-fire (LIF) neurons with random walk noise are analyzed, and an innovative algorithm based on overlapping time windows is proposed. The simulation results demonstrate the effective convergence and high accuracy of the proposed method for large-scale Max-Cut problems. The proposed method is validated through successful hardware implementation on a 6-transistor/2-resistor (6T2R) neuromorphic chip with phase change memory (PCM) synapses. In addition, as an expansion of the algorithm, several annealing techniques and bias split methods are proposed to improve convergence, along with circuit design ideas for efficient evaluation of sampling convergence using cell arrays and spiking systems. Overall, the results of the proposed methods demonstrate the potential of energy-efficient and hardware-implementable approaches using SNNs to solve COPs. To the best of the author's knowledge, this is the first study to solve the Max-Cut problem using an SNN neuromorphic hardware chip.

Recent advancements in quantum computing have opened new avenues for tackling long-standing complex combinatorial optimization problems that are intractable for classical computers. Predicting secondary structure of mRNA is one such notoriously difficult problem that can benefit from the everincreasing maturity of quantum computing technology. Accurate prediction of mRNA secondary structure is critical in designing RNA-based therapeutics as it dictates various steps of an mRNA life cycle, including transcription, translation, and decay. The current generation of quantum computers have reached utilityscale, allowing us to explore relatively large problem sizes. In this paper, we examine the feasibility of solving mRNA secondary structures on a quantum computer with sequence length up to 60 nucleotides representing problems in the qubit range of 10 to 80. We use Conditional Value at Risk (CVaR)-based VQE algorithm to solve the optimization problems, originating from the mRNA structure prediction problem, on the IBM Eagle and Heron quantum processors. To our encouragement, even with "minimal" error mitigation and fixed-depth circuits, our hardware runs yield accurate predictions of minimum free energy (MFE) structures that match the results of the classical solver CPLEX. Our results provide sufficient evidence for the viability of solving mRNA structure prediction problems on a quantum computer and motivate continued research in this direction.

We propose an exact method of synthesizing Nearest Neighbor Architecture (NNA) compliant networks of CNOT gates on 2-D architecture quantum computers using SMT solvers. To minimize noisy CNOT gates in the circuit, previous methods have employed SMT solvers and partitioning to obtain a lower CNOT gate count than contemporary methods. We observe that there is no need to fix the positions of T-gates relative to the source circuit; we can freely move them relative to the original, in turn enabling the gates that implement the logic feeding into them to be freely placed. In this paper, we introduce a set of constraints based around relaxing the constraints on the logic that acts on the T-gates in the circuit. Using these, we improve both the runtime and the quantum cost of the generated circuits. We compare our method to the method by Ding et al., and find a reduction 29.22% in the runtime and 58.11% in the quantum cost. We also find an advantage vs. IBM's SabreSwap.

Microreactor technologies have emerged as versatile platforms with the potential to revolutionize chemistry and materials research, offering sustainable solutions to global challenges in environmental and health domains. This survey paper provides an in-depth review of recent advancements in microreactor technologies, focusing on their role in facilitating accelerated discoveries in chemistry and materials. Specifically, we examine the convergence of microfluidics with machine intelligence and automation, enabling the exploitation of the cyber-physical environment as a highly integrated experimentation platform for rapid scientific discovery and process development. We investigate the applicability and limitations of microreactor-enabled discovery accelerators in various chemistry and materials contexts. Despite their tremendous potential, the integration of machine intelligence and automation into microreactor-based experiments presents challenges in establishing fully integrated, automated, and intelligent systems. These challenges can hinder the broader adoption of microreactor technologies within the research community. To address this, we review emerging technologies that can help lower barriers and facilitate the implementation of microreactor-enabled discovery accelerators. Lastly, we provide our perspective on future research directions for democratizing microreactor technologies, with the aim of accelerating scientific discoveries and promoting widespread adoption of these transformative platforms.

As the global population ages, assessment of cognitive and gait impairments is crucial for early identification and intervention for age-related disorders such as dementia and frailty; yet this quantitative, multi-faceted measurement requires trained professionals and specialized equipment. Drawing tests are widely used in clinical practice or as a self-administered tool for screening cognitive and motor dysfunctions mainly in particular domains such as visuospatial cognition and hand dexterity. We developed models for extending the applications of drawing tests to predict a wide range of cognitive and gait measures by analyzing the digitally captured drawing process. Specifically, we collected drawing data with a digital tablet from 189 older adults, along with neuropsychological examination-based cognitive measures relevant to memory, executive function, and global cognition and optical motion capture-based gait measures including gait speed, step length, and step time. We then evaluated regression models to predict multiple cognitive and gait measures solely from drawing data by combining global statistical drawing features with time-series embeddings obtained using self-supervised representation learning with deep neural networks. As a result, our models predicted both cognitive and gait measures with standardized mean absolute errors of 0.54 to 0.75. Our findings showed, for the first time, the feasibility of the use of drawing analysis for predicting multidimensional cognitive and gait measures, suggesting the potential of digital-drawing measures as a proxy marker for common aging-relevant clinical outcomes under multiple neurological and geriatric conditions.