While quantum chemical calculations supply fast and accurate predictions of atomistic chemical properties, they do not explicitly capture the electronic degrees of freedom of a molecule, which limits their applicability for reactive chemistry and chemical analysis. This paper presents a deep learning framework for the prediction of the quantum mechanical wavefunction on a local basis of atomic orbitals. This approach retains full access to the electronic structure via the wavefunction at force-field-like efficiency and captures quantum mechanics in an analytically differentiable representation.
Why did we not directly train on wave functions and eigenvalues, but rather on Hamiltonians in local basis representation? What other representations have been proposed and how do they differ? (a few other works have been published in the meantime) What is the rationale behind the chosen DNN layout and what are the crucial elements? What does the future of combined ML/quantum theory approaches hold?
Training on wave functions and eigenvalues has been tried but it,s not working.,That is why training is done on Hamiltonians in local basis representation.