We present real-space quantum Monte Carlo (QMC) calculations of the scandate ${\mathrm{LaScO}}_3$ that proved to be challenging for traditional electronic structure approaches due to strong correlation effects resulting in inaccurate band gaps from DFT and $GW$ methods when compared with existing experimental data. Besides calculating an accurate QMC band gap corrected for supercell size biases and in agreement with numerous experiments, we also predict the cohesive energy of the crystal using the standard fixed-node QMC without any empirical or nonvariational parameters. We show that promotion (optical) gap and fundamental gap agree with each other illustrating a clear absence of significant excitonic effects in the ideal crystal. We obtained these results in perfect consistency in two independent tracks that employ different basis sets (plane wave versus localized Gaussians), different codes for generating orbitals (quantum espresso versus crystal), different QMC codes (qmcpack versus qwalk) and different high-accuracy pseudopotentials (ccECPs versus Troullier-Martins) presenting the maturity and consistency of QMC methodology and tools for studies of strongly correlated problems.

中文翻译：

---

LaScO3的多体电子结构的实空间量子蒙特卡罗

我们提出扫描日期的实空间量子蒙特卡罗（QMC）计算 ${\mathrm{拉斯考}}_3$ 由于强大的相关效应，导致DFT和DFT的带隙不准确，这对传统的电子结构方法具有挑战性。 $G\mathrm{w\; ^}$方法与现有实验数据进行比较。除了计算针对超级晶胞尺寸偏差校正的准确QMC带隙并与大量实验一致外，我们还使用标准的固定节点QMC预测晶体的内聚能，而没有任何经验或不变参数。我们显示促进（光学）间隙和基本间隙彼此一致，说明在理想晶体中显然没有明显的激子效应。我们在两个独立的轨道上获得了完美的一致性，这些轨道采用了不同的基础集（平面波与局部高斯分布），用于生成轨道的不同代码（量子咖啡与晶体），不同的QMC代码（qmcpack与qwalk））和不同的高精度伪势（ccECP与Troullier-Martins）展示了QMC方法论和研究强相关问题的工具的成熟度和一致性。