arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2304.05262quant-ph
收藏

序列惩罚方法用于VQE研究激发态

Sequence of penalties method to study excited states using VQE

Rodolfo Carobene, Stefano Barison, Andrea Giachero

AI总结:

提出一种基于递增惩罚序列的VQE扩展方法,无需修改电路即可精确估计能量并研究激发态,在经典模拟和量子设备上均优于原始VQE及其他扩展方法。

摘要:

我们提出了一种变分量子本征求解器(VQE)的扩展方法,该方法能够实现更精确的能量估计,并可用于研究激发态。该方法基于在代价函数中引入一系列递增的惩罚项。这种方法不需要修改电路,因此可以在不增加额外深度成本的情况下应用。通过数值模拟,我们展示了能够产生具有所需物理性质(如总自旋和电荷)的变分态。我们在经典模拟器和当前可用的量子设备上评估了其性能,计算了不同物理构型下小分子体系的势能曲线。最后,我们将我们的方法与原始VQE及另一种扩展方法进行了比较,在能量和目标物理量方面均与精确模拟取得了更好的一致性。

英文摘要:

We propose an extension of the Variational Quantum Eigensolver (VQE) that leads to more accurate energy estimations and can be used to study excited states. The method is based on the introduction of a sequence of increasing penalties in the cost function. This approach does not require circuit modifications and thus can be applied with no additional depth cost. Through numerical simulations, we show that we are able to produce variational states with desired physical properties, such as total spin and charge. We assess its performance both on classical simulators and on currently available quantum devices, calculating the potential energy curves of small molecular systems in different physical configurations. Finally, we compare our method to the original VQE and to another extension, obtaining a better agreement with exact simulations for both energy and targeted physical quantities.

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。

↑