ABACUS原子轨道基组数据库第二版

该数据集包含26,648个文件，总未压缩数据大小约为1.2 GB。其重点是构建适用于第一性原理计算的高效率、高精度数值原子轨道基组数据，适合模拟大规模系统（数千原子或更多）中的原子和分子结构。在构建过程中采用了两种不同的优化算法，并以平面波基组计算的波函数数据作为参考态，生成了适用于68种元素的三套范数守恒赝势的原子轨道数据。为了满足不同计算速度和精度的需求，我们采用了分层优化的方法，从低精度轨道开始逐步提升至更高精度，从而实现了批量生成不同基组尺寸的数据。
此外，还进行了系统的测试以确保数据质量，使用状态方程、键长-能量关系曲线等指标验证生成的原子轨道基组数据与传统平面波基组数据之间的精度差异。验证结果表明，使用构建的原子轨道基组得到的结果与标准平面波基组结果之间的平均误差低至0.66 meV/atom。该数据集已在ABACUS程序中得到广泛应用，展现了很高的可复用性。它在凝聚态物理、材料科学和化学模拟等领域发挥了重要作用，极大提升了复杂系统模拟的效率与精度。

This dataset contains 26,648 files, with a total uncompressed size of approximately 1.2 GB. It focuses on constructing high-efficiency, high-precision numerical atomic orbital basis set data for first-principles calculations, which is suitable for simulating atomic and molecular structures in large-scale systems (thousands of atoms or more). During the construction process, two distinct optimization algorithms were employed, using wavefunction data calculated via plane-wave basis sets as reference states, to generate atomic orbital data for three sets of norm-conserving pseudopotentials applicable to 68 elements. To meet the requirements of varying computational speeds and precision demands, a hierarchical optimization approach was adopted, starting from low-precision orbitals and gradually upgrading to higher precision levels, thereby enabling batch generation of data with different basis set sizes.

In addition, systematic tests were carried out to ensure data quality, with metrics including the equation of state (EOS) and bond length-energy relationship curves used to verify the precision discrepancy between the generated atomic orbital basis set data and traditional plane-wave basis set data. The verification results indicate that the average error between the results obtained using the constructed atomic orbital basis sets and the standard plane-wave basis set results is as low as 0.66 meV/atom. This dataset has been widely utilized in the ABACUS program, exhibiting excellent reusability. It has played a crucial role in fields such as condensed matter physics, materials science, and chemical simulation, greatly enhancing the efficiency and accuracy of complex system simulations.