通过具有局部相干性的降阶实现方差降低
Variance reduction via deflation with local coherence
AI总结:
针对实现平移平均的随机估计量会引入新涨落的问题,本研究探索用非精确降阶子空间提升同位旋矢量矢量关联函数精度,通过$N_\text{f}=2$的非微扰$\text{O}(a)$改进Wilson费米子实验测量了降阶子空间的相对贡献与方差。
摘要:
在足够大的体积中,夸克线连通图的平移平均可使方差与体积成反比降低。然而,实现平移平均的随机估计量会引入新的涨落来源,在某些情况下这类涨落可能相对较大。本研究中,我们探究了非精确降阶子空间能否用于提升同位旋矢量矢量关联函数的精度。我们使用$N_\text{f}=2$种非微扰$\text{O}(a)$改进Wilson费米子开展数值实验,测量了降阶子空间对中心值的相对贡献及对应的方差。
英文摘要:
In large enough volumes, translation-averaging for quark-line connected diagrams reduces the variance inversely proportional to the volume. Stochastic estimators which implement translation averaging however introduce new sources of fluctuations, which in some cases can be relatively large. In this work, we explore whether inexact deflation subspaces can be used to improve the precision of the isovector vector correlators. We perform numerical experiments with $N_\mathrm{f}=2$ non-perturbatively $\mathrm O(a)$-improved Wilson fermions and measure the relative contribution from the deflation subspace to the central value and the corresponding variance.
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。