quantumNoise

数据集包含在不同IBM量子机器上相同量子电路的许多运行中。我们使用了9台不同的机器，对于每台机器，我们运行电路的2000执行。该电路沿其具有9个不同的测量步骤。为了获得9个结果分布，对于每次执行，电路的各个部分被附加9次 (在对IBM API的同一调用中，因此，在尽可能短的时间内) 测量每次的新步骤。对IBM API的调用遵循两种不同的策略。一种是为了最大限度地增加对接口的调用次数，将代码与尽可能多的运行并行化，甚至每次运行8000次，但考虑从内存中1000 8次来获得概率。另一种策略较慢，没有并行化，随后执行之间的等待时间最短。采用后者是为了及时获得更均匀的分配执行。This dataset comprises numerous runs of the same quantum circuit across distinct IBM quantum computers. We employed 9 different machines, and for each machine, we executed the circuit 2000 times. This circuit incorporates 9 separate measurement steps throughout its execution flow. To obtain the 9 result distributions, for each execution run, segments of the quantum circuit were appended 9 times within a single call to the IBM API—thereby minimizing the total time—to measure the new step following each addition. The calls to the IBM API follow two distinct strategies. The first strategy maximizes the number of API invocations by parallelizing the code across as many runs as possible, even executing 8000 times per run; it uses 1000 shots across 8 batches sourced from memory to calculate probabilities. The second strategy is slower, with no parallelization, and maintains the shortest possible waiting time between consecutive executions. The latter strategy was adopted to achieve a more evenly distributed execution schedule in a timely fashion.

搜集汇总

背景与挑战

背景概述

该数据集收集了在9台不同IBM量子机器上运行的相同量子电路数据，每台机器执行2000次，并采用并行化和非并行化两种策略来获取电路测量步骤的结果分布。其核心目的是用于学习量子设备的噪声指纹，以支持相关研究和分析。