Chemistry students learn early that a reaction proceeds from reactants to products via a transition state connecting them. This simple picture does not always hold though. The reactions never pass exactly through the transition state and we must propagate molecular dynamics trajectories …
If you want to use machine learning for potential energy surfaces, one of the biggest obstacles is getting the data to train machine learning potential. We have recently developed the physics-informed active learning protocol for efficient data sampling and training …
Active learning for building data-efficient machine learning potentials Read more »
Choosing a quantum chemical method suitable for your simulations is not an easy task, because you need to balance accuracy andcomputational cost requirements. Unless you use B3LYP all the time, of course. Generally, the more time you spend, the more …
Supercharge your computational chemistry with the universal and updatable AI models Read more »
Molecular IR spectra can now be calculated online with MLatom@XACS with DFT using a simple input file: You can find the frequencies and the corresponding intensities in the output file: You can check out the online tutorial for the theoretical …
Want to run DFT calculations in an easy way? Search no more! You can easily run DFT calculations online on XACS cloud with simple inputs such as this for geometry optimization. We also provide extensive online tutorials guiding you through various types …
Transfer learning (TL) is an often-used technique in machine learning that helps you train better AI models. It was used to create accurate universal ML models for computational chemistry – AIQM1 (Nat. Commun. 2021, 12, 7022) and ANI-1ccx (Nat. Commun. …
Transfer learning for better AI models with less data Read more »
Recently, the Chung group at Southern University of Science and Technology (SUSTech) has combined efficient machine learning potentials (MLPs) with multi-scale quantum refinement methods to enhance computational efficiency and reliability. Their results are published in Nature Communications. The CC-quality AIQM1 method in MLatom@XACS software was …
MLatom supports a wide range of universal machine learning (ML)-based models including ML potentials and hybrid ML-enhanced quantum mechanical (QM) methods. They can be used out-of-box without training. In short, we have there several new universal (no training needed) ML …
Easy-to-use universal AI models for modern computational chemistry Read more »
MLatom@XACS is a powerful tool for training and using machine learning potentials. It supports a wide variety of representative potentials. These potentials include: ·Equivariant neural network MACE ·Popular ANI with a good cost/performance ratio ·Kernel methods such as KREG and …
Training and using machine learning potentials with MLatom@XACS Read more »
Check out our tutorial on how to run TS geometry optimizations and also analysis (freq and IRC) with MLatom@XACS.We demonstrate on an example of the Diels–Alder reaction how to use fast GFN2-xTB and AIQM1 methods providing reasonable TS geometries and …
Transition state search and analysis with MLatom@XACS Read more »