arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2504.06348quant-phcond-mat.mes-hallcond-mat.mtrl-sci
收藏

在容错量子计算机上模拟实时化学动力学的综合框架

A comprehensive framework to simulate real-time chemical dynamics on a fault-tolerant quantum computer

Felipe H. da Jornada, Matteo Lostaglio, Sam Pallister, Burak Şahinoğlu, Karthik I. Seetharam

AI总结:

本文提出一个在容错量子计算机上模拟化学实时动力学的端到端框架,通过赝离子平面波算法、量子拒绝采样优化电路和机器学习分子指纹,实现电子与核自由度的联合模拟,并给出了具体资源成本分析。

摘要:

我们提出了一个全面的端到端框架,用于在容错量子计算机上模拟化学系统的实时动力学,同时纳入电子和原子核的量子自由度。全粒子模拟在量子计算机上名义上是高效的,但在实践中不可行。因此,我们方法的核心是构建一个利用赝离子的一重量子化平面波算法。赝离子将化学上不活跃的电子和原子核合并为一个有效的动力学离子实体,将量子化学中成熟的赝势概念扩展到二体相互作用。我们明确描述了用于所有自由度初始态制备的高效量子电路,以及用于对描述相互作用的赝离子和化学活跃电子的哈密顿量进行块编码的量子电路,并利用量子拒绝采样的最新进展来优化实现。为了提取有用的化学信息,我们首先通过结合密度泛函计算与机器学习技术设计分子指纹,随后通过替代经典分子动力学模拟对其进行验证。然后,这些指纹被相干地编码到量子计算机上,以便通过振幅估计进行高效的分子识别。我们对在容错量子计算机上运行该算法的成本进行了广泛分析,涉及多个具有化学意义的系统。作为示例,模拟 $\mathrm{NH_3}$ 与 $\mathrm{BF_3}$(一个40粒子系统)之间的相互作用需要808个逻辑量子比特来编码该问题,并且每飞秒时间演化需要约 $10^{11}$ 个Toffoli门。我们的结果为面向化学和材料动力学的进一步量子算法开发奠定了基础。

英文摘要:

We present a comprehensive end-to-end framework for simulating the real-time dynamics of chemical systems on a fault-tolerant quantum computer, incorporating both electronic and nuclear quantum degrees of freedom. An all-particle simulation is nominally efficient on a quantum computer, but practically infeasible. Hence, central to our approach is the construction of a first-quantized plane-wave algorithm making use of pseudoions. The latter consolidate chemically inactive electrons and the nucleus into a single effective dynamical ionic entity, extending the well-established concept of pseudopotentials in quantum chemistry to a two-body interaction. We explicitly describe efficient quantum circuits for initial state preparation across all degrees of freedom, as well as for block-encoding the Hamiltonian describing interacting pseudoions and chemically active electrons, by leveraging recent advances in quantum rejection sampling to optimize the implementations. To extract useful chemical information, we first design molecular fingerprints by combining density-functional calculations with machine learning techniques, and subsequently validate them through surrogate classical molecular dynamics simulations. These fingerprints are then coherently encoded on a quantum computer for efficient molecular identification via amplitude estimation. We provide an extensive analysis of the cost of running the algorithm on a fault-tolerant quantum computer for several chemically interesting systems. As an illustration, simulating the interaction between $\mathrm{NH_3}$ and $\mathrm{BF_3}$ (a 40-particle system) requires 808 logical qubits to encode the problem, and approximately $10^{11}$ Toffoli gates per femtosecond of time evolution. Our results establish a foundation for further quantum algorithm development targeting chemical and material dynamics.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑