scholarly journals Basis Set Extrapolations for Density Functional Theory

Author(s):  
Peter Kraus

Improving results of correlated wavefunction theory calculations by extrapolating from successive basis sets is nowadays a common practice. However, such approaches are uncommon in density functional theory, especially due its faster convergence towards the basis set limit. In this work I present approaches for basis set extrapolation in density functional theory that enable users to obtain results of 4-zeta quality from 3- and 2-zeta calculations, i.e. at a fraction of the computational cost. The extrapolation techniques work well with modern density functionals and common basis sets.<br>

2020 ◽  
Author(s):  
Peter Kraus

Improving results of correlated wavefunction theory calculations by extrapolating from successive basis sets is nowadays a common practice. However, such approaches are uncommon in density functional theory, especially due its faster convergence towards the basis set limit. In this work I present approaches for basis set extrapolation in density functional theory that enable users to obtain results of 4-zeta quality from 3- and 2-zeta calculations, i.e. at a fraction of the computational cost. The extrapolation techniques work well with modern density functionals and common basis sets.<br>


2007 ◽  
Vol 06 (04) ◽  
pp. 675-685 ◽  
Author(s):  
XIAO-HONG LI ◽  
RUI-ZHOU ZHANG ◽  
XIAN-ZHOU ZHANG ◽  
XIN-LU CHENG ◽  
XIANG-DONG YANG

The heats of formation (HOFs) for 15 thiol compounds are calculated by employing the hybrid density functional theory (B3LYP, B3PW91, B3P86) methods with 6-311G** basis set and the complete basis set (CBS-Q) method. It is demonstrated that the B3P86 and CBS-Q methods are accurate to compute the reliable HOFs for thiol compounds. In order to test whether the B3P86 method has a low basis set sensitivity, the HOFs for six thiol compounds are also calculated by using the B3P86 method with 6-31+G*, 6-31+G**, and 6-311+G** basis sets for comparison. We also extend our study by employing the nonlocal BLYP method together with 6-31+G* basis set to calculate the HOFs for thiol compounds. The obtained results are compared with the experimental results. It is noted that the B3P86 method is not sensitive to the basis sets. Considering the inevitably computational cost of CBS-Q method and the reliability of the B3P86 calculation, the B3P86 method with a moderate or a larger basis set such as 6-311G** and 6-311+G** may be more suitable to calculate the HOFs of thiol compounds. In addition, we believe that the maximum error associated with the calculated HOFs is less than 6 kcal/mol for the B3P86/6-311G** method and it is expected that the error bar is more likely 1–5 kcal/mol for the HOFs of thiol compounds.


2019 ◽  
Author(s):  
Kamal Batra ◽  
Stefan Zahn ◽  
Thomas Heine

<p>We thoroughly benchmark time-dependent density- functional theory for the predictive calculation of UV/Vis spectra of porphyrin derivatives. With the aim to provide an approach that is computationally feasible for large-scale applications such as biological systems or molecular framework materials, albeit performing with high accuracy for the Q-bands, we compare the results given by various computational protocols, including basis sets, density-functionals (including gradient corrected local functionals, hybrids, double hybrids and range-separated functionals), and various variants of time-dependent density-functional theory, including the simplified Tamm-Dancoff approximation. An excellent choice for these calculations is the range-separated functional CAM-B3LYP in combination with the simplified Tamm-Dancoff approximation and a basis set of double-ζ quality def2-SVP (mean absolute error [MAE] of ~0.05 eV). This is not surpassed by more expensive approaches, not even by double hybrid functionals, and solely systematic excitation energy scaling slightly improves the results (MAE ~0.04 eV). </p>


2008 ◽  
Vol 07 (05) ◽  
pp. 943-951 ◽  
Author(s):  
XIAO-HONG LI ◽  
ZHENG-XIN TANG ◽  
ABRAHAM F. JALBOUT ◽  
XIAN-ZHOU ZHANG ◽  
XIN-LU CHENG

Quantum chemical calculations are used to estimate the bond dissociation energies (BDEs) for 15 thiol compounds. These compounds are studied by employing the hybrid density functional theory (B3LYP, B3PW91, B3P86, PBE0) methods and the complete basis set (CBS-Q) method together with 6-311G** basis set. It is demonstrated that B3P86 and CBS-Q methods are accurate for computing the reliable BDEs for thiol compounds. In order to test whether the non-local BLYP method suggested by Fu et al.19 is general for our study and whether B3P86 method has a low basis set sensitivity, the BDEs for seven thiol compounds are also calculated using BLYP/6-31+G* and B3P86 method with 6-31+G*, 6-31+G**, and 6-311+G** basis sets for comparison. The obtained results are compared with the available experimental results. It is noted that B3P86 method is not sensitive to the basis set. Considering the inevitable computational cost of CBS-Q method and the reliability of the B3P86 calculations, B3P86 method with a moderate or a larger basis set may be more suitable to calculate the BDEs of the C–SH bond for thiol compounds.


2011 ◽  
Vol 89 (6) ◽  
pp. 657-662 ◽  
Author(s):  
Nicholas J. Russ ◽  
Chun-min Chang ◽  
Jing Kong

We present an efficient algorithm for evaluating the exchange-correlation contribution to the nuclear gradients of density-functional theory calculation within the local spin-density approximation. The algorithm is an extension of the multiresolution exchange-correlation (mrXC) method, which treats smooth and compact parts of the electron density separately. The nuclear gradient of the smooth density is calculated on the even-spaced grid while the compact part of the density is handled on the normal atom-centered grid (ACG). The overall formulism is still formally based on the ACG, and thus does not change the results of the existing ACG-based algorithms for all-electron density-functional theory (DFT) calculations. The variation of the positions and weights of ACG owing to the nuclear perturbation is also handled rigorously. Benchmark calculations with different basis sets and sizes of ACG show that mrXC reduces the computational cost by several times without loss of accuracy. It also lessens the impact on the CPU time when the size of the ACG is increased.


2019 ◽  
Author(s):  
Kamal Batra ◽  
Stefan Zahn ◽  
Thomas Heine

<p>We thoroughly benchmark time-dependent density- functional theory for the predictive calculation of UV/Vis spectra of porphyrin derivatives. With the aim to provide an approach that is computationally feasible for large-scale applications such as biological systems or molecular framework materials, albeit performing with high accuracy for the Q-bands, we compare the results given by various computational protocols, including basis sets, density-functionals (including gradient corrected local functionals, hybrids, double hybrids and range-separated functionals), and various variants of time-dependent density-functional theory, including the simplified Tamm-Dancoff approximation. An excellent choice for these calculations is the range-separated functional CAM-B3LYP in combination with the simplified Tamm-Dancoff approximation and a basis set of double-ζ quality def2-SVP (mean absolute error [MAE] of ~0.05 eV). This is not surpassed by more expensive approaches, not even by double hybrid functionals, and solely systematic excitation energy scaling slightly improves the results (MAE ~0.04 eV). </p>


Sign in / Sign up

Export Citation Format

Share Document