Good Practices in Database Generation for Benchmarking Density Functional Theory

Wiley Interdisciplinary Reviews Computational Molecular Science, Journal Year: 2025, Volume and Issue: 15(1)

Published: Jan. 1, 2025

ABSTRACT The hundreds of density functional theory (DFT) methods developed over the past three decades are often referred to as “zoo” DFT approximations. In line with this terminology, numerous benchmark studies might be considered “safari” evaluation efforts, reflecting their abundance, diversity, and wide range application methodological aspects. These benchmarks have played a critical role in establishing dominant approach quantum chemical applications remain essential for selecting an appropriate method specific properties (e.g., reaction energy, barrier height, or noncovalent interaction energy) systems organic, inorganic, organometallic). vital tool both users selection developers design parameterization. This review provides best‐practice guidance on key aspects benchmarking, such quality reference values, dataset size, geometries, basis sets, statistical analysis, electronic availability data. Additionally, we present flowchart assist systematically choosing these aspects, thereby enhancing reliability reproducibility benchmarking studies.

Language: Английский

---

Improving the Reliability of, and Confidence in, DFT Functional Benchmarking through Active Learning

Journal of Chemical Theory and Computation, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 2, 2025

Validating the performance of exchange-correlation functionals is vital to ensure reliability density functional theory (DFT) calculations. Typically, these validations involve benchmarking data sets. Currently, such sets are usually assembled in an unprincipled manner, suffering from uncontrolled chemical bias, and limiting transferability results a broader space. In this work, data-efficient solution based on active learning explored address issue. Focusing─as proof principle─on pericyclic reactions, we start BH9 set design reaction space around initial by combinatorially combining templates substituents. Next, surrogate model trained predict standard deviation activation energies computed across selection 20 distinct DFT functionals. With model, designed explored, enabling identification challenging regions, i.e., regions with large divergence, for which representative reactions subsequently acquired as additional training points. Remarkably, it turns out that function mapping molecular structure divergence readily learnable; convergence reached upon acquisition fewer than 100 reactions. our final updated more challenging─and arguably representative─pericyclic curated, demonstrate has changed significantly compared original subset.

Language: Английский

---

Meta-GGA that describes weak interactions in addition to bond energies and band gaps

Physical review. B./Physical review. B, Journal Year: 2025, Volume and Issue: 111(15)

Published: April 18, 2025

We show that the recently proposed Lebeda-Aschebrock-Kümmel (LAK) meta-generalized gradient approximation, accuracy of which was previously established for atomization energies, bond lengths, and band gaps, also captures weak interactions near equilibrium without a dispersion correction. discuss how this is achieved. Furthermore, we among semilocal cost pure functionals, LAK one reaches highest large GMTKN55 database general thermochemistry kinetics. Next, explain design strategy enhancement factor engineering. Its key idea to complement exact constraints with construction principles. Finally, areas research in use may offer advantages over existing functionals. Published by American Physical Society 2025

Language: Английский

---

Data Quality in the Fitting of Approximate Models: A Computational Chemistry Perspective

Journal of Chemical Theory and Computation, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 18, 2024

Empirical parametrization underpins many scientific methodologies including certain quantum-chemistry protocols [e.g., density functional theory (DFT), machine-learning (ML) models]. In some cases, the fitting requires a large amount of data, necessitating use data obtained using low-cost, and thus low-quality, means. Here we examine effect low-quality on resulting method in context DFT methods. We multiple G2/97 sets different qualities to fit DFT-type Encouragingly, this can tolerate relatively proportion which may be attributed physical foundations models modest number parameters. Further examination "ML-quality" shows that adding small high-quality ones not offer tangible benefits. On other hand, when is limited scope, diversification by improves performance. Quantitatively, for parametrizing (and perhaps also ML models), caution should taken more than 50% set contains questionable average error full 20 kJ mol–1. One follow recently proposed transferability principles ensure diversity set.

Language: Английский

---

High-quality reference quantities for the LiTC set of small Li clusters

Chemical Physics Letters, Journal Year: 2024, Volume and Issue: unknown, P. 141795 - 141795

Published: Dec. 1, 2024

Language: Английский

---

The Best DFT Functional Is the Ensemble of Functionals

Advanced Science, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 25, 2024

Abstract The development of better density functional theory (DFT) methods is one the most active research areas, given importance DFT for ubiquitous molecular and materials simulations. However, this primarily focuses on improving a specific exchange‐correlation Kohn–Sham functional. Here, robust procedure proposed constructing transferable ensembles functionals that perform superior to any constituent individual It shown such built only with predating GMTKN55 benchmark 2017 can reach record‐low weighted error 1.62 kcal mol −1 compared 3.08 best DENS24 are also introduced as practical consistently accurate performance various simulations at affordable cost. open‐source be used online. Additionally, it integrated into SCF by creating mixed functionals, which have same accuracy but faster than independent functionals.

Language: Английский

---