Novel implementations based on dense tensor storage are presented for thesinglet-reference perfect quadruples (PQ) [Parkhill, Lawler, and Head-Gordon,J. Chem. Phys. 130, 084101 (2009)] and perfect hextuples (PH) [Parkhill andHead-Gordon, J. Chem. Phys. 133, 024103 (2010)] models. The methods areobtained as block decompositions of conventional coupled-cluster theory thatare exact for four electrons in four orbitals (PQ) and six electrons in sixorbitals (PH), but that can also be applied to much larger systems. PQ and PHhave storage requirements that scale as the square, and as the cube of thenumber of active electrons, respectively, and exhibit quartic scaling of thecomputational effort for large systems. Applications of the new implementationsare presented for full-valence calculations on linear polyenes (C n H n+2 ),which highlight the excellent computational scaling of the presentimplementations that can routinely handle active spaces of hundreds ofelectrons. The accuracy of the models is studied in the {\pi} space of thepolyenes, in hydrogen chains (H 50 ), and in the {\pi} space of polyacenemolecules. In all cases, the results compare favorably to density matrixrenormalization group values. With the novel implementation of PQ, activespaces of 140 electrons in 140 orbitals can be solved in a matter of minutes ona single core workstation, and the relatively low polynomial scaling means thatvery large systems are also accessible using parallel computing.
展开▼
机译：提出了基于密集张量存储的新颖实现，用于单参参考完美四倍体（PQ）[Parkhill，Lawler，and Head-Gordon，J。化学物理130，084101（2009）]和完美的六胞胎（PH）[Parkhill andHead-Gordon，J. Chem。物理133，024103（2010）]模型。这些方法是作为常规耦合簇理论的块分解而获得的，这些分解对于四个轨道（PQ）中的四个电子和六个轨道（PH）中的六个电子是精确的，但也可以应用于更大的系统。 PQ和PH的存储需求分别按方形和活性电子数量的立方进行缩放，并且对于大型系统而言，其计算工作量呈四次缩放。提出了新实现的应用，用于线性多烯（C n H n + 2）的全价计算，突出了该实现的出色计算比例，可以常规处理数百个电子的有效空间。在多烯的{\ pi}空间，氢链（H 50）和多并苯分子的{\ pi}空间中研究了模型的准确性。在所有情况下，结果均优于密度矩阵重新归一化组值。通过PQ的新颖实现，可以在单个核心工作站上在几分钟内解决140个轨道中140个电子的有效空间，并且多项式缩放比例相对较低，这意味着使用并行计算也可以访问很多大型系统。
展开▼
