Quemix and Idemitsu Kosan propose a molecular dynamics method combining orbital-free DFT and a machine-learned interatomic potential to efficiently simulate electric-field-driven ionic transport and changes in interfacial charge states.

By utilizing silicon vacancies (VSi) in SiC as quantum sensors, the study enables direct, high-spatial-resolution measurement of electric fields inside devices. High electric fields up to approximately 2.3 MV/cm and their spatial mapping are demonstrated.

Amplitude encoding of classical data is one of the practical bottlenecks in quantum computing workflows. This work introduces a new approximate amplitude encoding method using the Adaptive Interpolating Quantum Transform (AIQT). By learning a data-adaptive transform basis, the approach concentrates information into fewer coefficients and achieves up to about a 50% reduction in reconstruction error compared with Fourier-based methods.

Quemix researchers propose AIQT, a quantum-native framework for transform learning that interpolates between quantum transformations with minimal parameters. The approach extends the core idea of General Transform (GT) to quantum circuits.

Gekko et al. from Quemix have published a paper proposing a novel transformation method, the General Transform (GT), designed to enhance the representational power of machine learning models. The paper introduces a framework that replaces traditional fixed transforms—such as the discrete Fourier transform—with a learnable combination of transforms that adapts to both the dataset and the task. The weights of the transforms are optimized jointly with the model, enabling genera

https://arxiv.org/abs/2501.07211 Kosugi, Huang, Nishi, and Matsushita present an innovative approach that dramatically improves the efficiency of encoding molecular information on a quantum computer. Their paper proposes a novel method for encoding molecular orbitals in the first-quantized formalism—a fundamental way to represent molecules—on a quantum computer. This method leverages a concrete procedure, executed on a classical computer, to fine-tune the accuracy and success

Quemix researchers have developed a new quantum algorithm for solving the linear advection-diffusion equation.

FMQAを適用し、アニーリングマシンで任意のポテンシャルを使った結晶構造探索、最適な原子配列を効率的に発見する可能性。

https://arxiv.org/abs/2407.16345 Huang et al. at Quemix have proposed an approximate real-time evolution operator for potentials using only one ancillary qubit, demonstrating its application in first-quantized Hamiltonian simulations. This technique has the potential to significantly improve the efficiency of Hamiltonian simulations on near-term quantum computers and is expected to find practical applications in fields such as quantum physics and materials science.

Quemixの研究者たちは、確率的虚時間発展法(PITE)を用いた無軌道密度汎関数理論(OF-DFT)による基底状態電子密度の計算手法を提案しました。量子位相推定(QPE)と組み合わせることで、最低エネルギー固有値の取得を高速化し、自己無撞着場(SCF)計算を部分的に加速すること

https://www.nature.com/articles/s41428-024-00915-8 The study offers insights into the polymerization behavior and potential applications of the novel monomers.