In Silico Chemical Experiments in the Age of AI: From Quantum Chemistry to Machine Learning and Back

Advanced Materials, Год журнала: 2024, Номер 36(30)

Опубликована: Май 25, 2024

Abstract Computational chemistry is an indispensable tool for understanding molecules and predicting chemical properties. However, traditional computational methods face significant challenges due to the difficulty of solving Schrödinger equations increasing cost with size molecular system. In response, there has been a surge interest in leveraging artificial intelligence (AI) machine learning (ML) techniques silico experiments. Integrating AI ML into increases scalability speed exploration space. remain, particularly regarding reproducibility transferability models. This review highlights evolution from, complementing, or replacing energy property predictions. Starting from models trained entirely on numerical data, journey set forth toward ideal model incorporating physical laws quantum mechanics. paper also reviews existing their intertwining, outlines roadmap future research, identifies areas improvement innovation. Ultimately, goal develop architectures capable accurate transferable solutions equation, thereby revolutionizing experiments within materials science.

Язык: Английский

---

Reference CC3 Excitation Energies for Organic Chromophores: Benchmarking TD-DFT, BSE/GW, and Wave Function Methods

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер unknown

Опубликована: Сен. 5, 2024

To expand the QUEST database of highly accurate vertical transition energies, we consider a series large organic chromogens ubiquitous in dye chemistry, such as anthraquinone, azobenzene, BODIPY, and naphthalimide. We compute, at CC3 level theory, singlet triplet energies associated with low-lying excited states. This leads to collection more than 120 new excitation energies. For several transitions, have been able determine CCSDT compact basis set, finding minimal deviations from values for most Subsequently, employ these reference benchmark lower-order wave function approaches, including popular ADC(2) CC2 schemes, well time-dependent density-functional theory (TD-DFT), both without applying Tamm-Dancoff approximation (TDA). At TD-DFT level, evaluate panel global, range-separated, local, double hybrid functionals. Additionally, assess performance Bethe-Salpeter equation (BSE) formalism relying on G0W0 evGW quasiparticle evaluated various starting points. It turns out that ADC(2.5) are models among those respective O(N5) O(N6) scalings system size. In contrast, CCSD does not outperform CC2. The best performing exchange-correlation functionals include BMK, M06-2X, M06-SX, CAM-B3LYP, ωB97X-D, LH20t, average approximately 0.20 eV or slightly below. Errors can be further reduced by considering hybrids. Both SOS-ωB88PP86 SOS-ωPBEPP86 exhibit particularly attractive performances overall quality par CC2, whereas PBE0-DH PBE-QIDH only less efficient. BSE/evGW calculations based Kohn-Sham points found effective but much their counterparts.

Язык: Английский

---

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

Journal of Chemical Theory and Computation, Год журнала: 2025, Номер unknown

Опубликована: Фев. 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.

Язык: Английский

---

WFOT: A Wave Function Overlap Tool between Single- and Multi-Reference Electronic Structure Methods for Spectroscopy Simulation

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(11), С. 4804 - 4819

Опубликована: Июнь 3, 2024

We report the development of a novel diagnostic tool, named wave function overlap tool (WFOT), designed to evaluate between functions computed at single-reference [i.e., time-dependent density functional theory or configuration interaction singles (CIS)] and multireference (i.e., CASSCF/CASPT2) electronic structure levels theory. It relies on truncating single- WFs CIS-like expansions spanning same configurational space maximizing molecular orbital by means unitary transformation. To demonstrate functionality we calculate transient spectrum acetylacetone evaluating excited state absorption signals with quality top on-the-fly dynamics simulations. Semiautomatic spectra generation is facilitated interfacing COBRAMM package, which also allows one use WFOT several quantum chemistry codes such as Gaussian, NWChem, OpenMolcas. Other exciting possibilities for utilization code beyond simulation spectroscopy are eventually discussed.

Язык: Английский

---

Assessing and applying DFT approaches for geometries and UV–Vis absorption spectra of tetragonal iron(II) complexes

Theoretical Chemistry Accounts, Год журнала: 2024, Номер 143(11)

Опубликована: Окт. 12, 2024

Язык: Английский

---