密度泛函方法精度的考虑
发布网友
发布时间：2024-10-15 01:21
共1个回答
热心网友
时间：2024-10-15 02:46
Density Functional Theory (DFT) calculations introduce an additional step in the main stages of Hartree-Fock calculations. This step involves the numerical integration of the functional (or its derivatives). Consequently, DFT calculations' accuracy is influenced not only by the numerical errors inherent in Hartree-Fock calculations (such as integration precision, SCF convergence, and CPHF convergence) but also by the number of grid points used in numerical integration. Gaussian 03 uses a 'finer' integration grid by default (Integral=FineGrid), which optimizes calculation accuracy with minimal additional computational cost. It's not advisable to use a more sparse grid in DFT calculations. It's important to note that when comparing energies (e.g., calculating energy differences, generating heats), all calculations must use the same integration grid. In scenarios requiring particularly stringent geometric optimization calculations for certain molecular systems, it might be necessary to use a finer grid. The integration grid can be selected differently throughout the calculation path using the Int(Grid=N) option (refer to the IOp, Int=Grid, Stable, TD, DenFit key terms for detailed instructions).
For applications involving energy, analytical gradients, and analytical frequencies; ADMP calculations.
Relevant keywords: IOp, Int=Grid, Stable, TD, DenFit
扩展资料
Gaussian 03 提供相当多的密度泛函理论(DFT) 75,76,448,449模型（DFT 方法和应用 的讨论参见448, 450-461）。所有的DFT 模型都能计算能量78，解析梯度和真正的解析 频率197-199。对大多数的一般模型，计算推荐使用由freqmem（见第5 章）估算的最佳 内存大小。 自洽反应场(SCRF)可与DFT 能量，优化和频率计算合用，以模拟溶液中的体系。 纯DFT 计算通常需要利用密度拟合。细节参见本章前面的基组部分。
