A5001121489- Machine Learning in Materials Science
- Advanced Chemical Physics Studies
- Crystallography and molecular interactions
- X-ray Diffraction in Crystallography
- Spectroscopy and Quantum Chemical Studies
- Solid-state spectroscopy and crystallography
- Surface Chemistry and Catalysis
- Electron and X-Ray Spectroscopy Techniques
- Catalysis and Oxidation Reactions
- Machine Learning and Data Classification
- Computational Drug Discovery Methods
- Organic Light-Emitting Diodes Research
- Photonic and Optical Devices
- Analytical Chemistry and Sensors
- AI and HR Technologies
- Topic Modeling
- Nuclear Materials and Properties
- Luminescence and Fluorescent Materials
- Nonlinear Optical Materials Research
- Electronic and Structural Properties of Oxides
- Luminescence Properties of Advanced Materials
- Cold Atom Physics and Bose-Einstein Condensates
- Molecular Junctions and Nanostructures
- Artificial Intelligence in Healthcare
- Physics of Superconductivity and Magnetism
University of Toronto
2012-2025
Structural Genomics Consortium
2025
Dalhousie University
2020-2024
We present a comprehensive analysis of the capabilities modern machine learning force fields to simulate long-term molecular dynamics at near-ambient conditions for molecules, molecule-surface interfaces, and materials within TEA Challenge 2023.
Assessing the performance of modern machine learning force fields across diverse chemical systems to identify their strengths and limitations within TEA Challenge 2023.
Post-self-consistent dispersion corrections are now the norm when applying density-functional theory to systems where non-covalent interactions play an important role. However, there is a wide range of base functionals and available from which choose. In this work, we opine on most desirable requirements ensure that both functional correction, individually, as accurate possible for non-bonded repulsion attraction. The should be dispersionless, numerically stable, involve minimal...
Abstract For many machine learning applications in science, data acquisition, not training, is the bottleneck even when avoiding experiments and relying on computation simulation. Correspondingly, order to reduce cost carbon footprint, training efficiency key. We introduce minimal multilevel (M3L) which optimizes set sizes using a loss function at multiple levels of reference minimize combination prediction error with overall acquisition costs (as measured by computational wall-times)....
Molecular crystals are important for many applications, including energetic materials, organic semiconductors, and the development commercialization of pharmaceuticals. The exchange-hole dipole moment (XDM) dispersion model has shown good performance in calculation relative absolute lattice energies molecular crystals, although it traditionally been applied combination with plane-wave/pseudopotential approaches. This limited XDM to use semilocal functional approximations, which suffer from...
While density-functional theory (DFT) remains one of the most widely used tools in computational chemistry, functionals fail to properly account for effects London dispersion. Hence, there are many popular post-self-consistent methods add a dispersion correction DFT energy. Until now, have never been compared on equal footing due not being implemented same electronic structure packages. In this work, we performed large-scale benchmarking study, directly comparing accuracy exchange-hole...
A seventh blind test of crystal structure prediction has been organized by the Cambridge Crystallographic Data Centre. The results are presented in two parts, with this second part focusing on methods for ranking structures order stability. exercise involved standardized sets seeded from a range generation methods. Participants 22 groups applied several periodic DFT-D methods, machine learned potentials, force fields derived empirical data or quantum chemical calculations, and various...
Abstract Two common methods for recording absolute photoluminescence quantum yields using integrating spheres are discussed. These developed from a theoretical standpoint, with discussion of the assumptions made in each and their principle differences, practical comparisons between two range different materials sample types. It is shown that despite underlying differences both ultimately yield very similar experimental results. Additionally, concept time‐dependent examined preliminary...
Pairing the XDM dispersion model with hybrid density functionals shows significant improvements in computed crystal energy landscapes for 4 of 26 compounds appearing first six blind tests structure prediction.
Exact exchange contributions significantly affect electronic states, influencing covalent bond formation and breaking. Hybrid density functional approximations, which average exact admixtures empirically, have achieved success but fall short of high-level quantum chemistry accuracy due to delocalization errors. We propose adaptive hybrid functionals, generating optimal admixture ratios on the fly using data-efficient machine learning models with negligible overhead. The...
We present the second part of rigorous evaluation state-of-the-art machine learning force fields (MLFFs) within TEA Challenge 2023. This study provides an in-depth analysis performance MACE, SO3krates, sGDML, SOAP/GAP, and FCHL19* in modeling molecules, molecule-surface interfaces, periodic materials. compare observables obtained from molecular dynamics (MD) simulations using different MLFFs under identical conditions. Where applicable, density-functional theory (DFT) or experiment serves as...
Atomistic simulations are routinely employed in academia and industry to study the behavior of molecules, materials, their interfaces. Central these force fields (FF), whose development is challenged by intricate interatomic interactions at different spatio-temporal scales vast expanse chemical space. Machine learning (ML) FFs, trained on quantum-mechanical energies forces, have shown capacity achieve sub-kcal/(mol*A) accuracy while maintaining computational efficiency. The TEA Challenge...
Many crystal structure prediction protocols only concern themselves with the electronic energy of molecular crystals. However, vibrational contributions to free (Fvib) can be significant in determining accurate stability rankings for candidates. While force-field studies have been conducted gauge magnitude these free-energy corrections, highly results from quantum mechanical methods, such as density-functional theory (DFT), are desirable. Here, we introduce PV17 set 17 polymorphic pairs...
Polysilafluorenes (PSFs) are an important class of light-emitting conjugate polymers noted for their characteristic brilliant solid state blue luminescence, high quantum efficiency, excellent solubility, and improved thermal stability. These also reported to have superior electron conductivity polyfluorenes. The higher affinity conductivity, which particularly promising OLEDs, originate from σ*−π* conjugation between the σ* antibonding orbital exocyclic Si–C bond π* butadiene fragment. In...
Differences in London dispersion, the weakest intermolecular interaction, can be sufficient to impart enantioselectivity for amino-acid adsorption on quartz.
Exact exchange contributions are known to crucially affect electronic states, which in turn govern covalent bond formation and breaking chemical species. Empirically averaging the exact admixture over configurational compositional degrees of freedom, hybrid density functional approximations (DFAs) have been widely successful, yet fallen short reach explicitly correlated high level quantum chemistry accuracy general. Using machine learning, we adaptified DFAs by generating optimal ratios "on...
Despite dominating industrial processes, heterogeneous catalysts remain challenging to characterize and control. This is largely attributable the diversity of potentially active sites at catalyst-reactant interface complex behaviour that can arise from interactions between sites. Surface-supported, single-site molecular aim bring together benefits both homogeneous catalysts, offering easy separability while exploiting design reactivity, though presence a surface likely influence reaction...
The seventh blind test of crystal structure prediction (CSP) methods substantially increased the level complexity target compounds relative to previous tests organized by Cambridge Crystallographic Data Centre. In this work, performance density-functional is assessed using numerical atomic orbitals and exchange-hole dipole moment dispersion correction (XDM) for energy-ranking phase test. Overall, excellent was seen two rigid molecules (XXVII, XXVIII) organic salt (XXXIII). However,...
Molecular crystals are important for many applications, including energetic materials, organic semi-conductors, and the development commercialization of pharmaceuticals. The exchange-hole dipole moment (XDM) dispersion model has shown good performance in calculation relative absolute lattice energies molecular crystals, although it traditionally been applied combination with plane-wave/pseudopotential approaches. This limited XDM to use semilocal functional approximations, which suffer from...
For many machine learning applications in science, data acquisition, not training, is the bottleneck even when avoiding experiments and relying on computation simulation. Correspondingly, order to reduce cost carbon footprint, training efficiency key. We introduce minimal multilevel (M3L) which optimizes set sizes using a loss function at multiple levels of reference minimize combination prediction error with overall acquisition costs (as measured by computational wall-times). Numerical...
Accurate and efficient computation of relative energies molecular-crystal polymorphs is central importance for solid-state pharmaceuticals in the field crystal engineering. In recent years, dispersion corrected density-functional theory (DFT) has emerged as pre-eminent energy-ranking method structure prediction (CSP). However, planewave implementations these methods are hindered by poor scaling large unit cells limited to semi-local functionals that suffer from delocalisation error. this...
Molecular crystals are important for many applications, including energetic materials, organic semiconductors, and the development commercialization of pharmaceuticals. crystal structure prediction (CSP) relies on use accurate inexpensive computational methods to rank candidate structures. The exchange-hole dipole moment (XDM) model has shown excellent performance in calculation relative absolute lattice energies molecular past. XDM traditionally been applied combination with...