Intrinsic approximations in the first principles calculations and inadequate knowledge in the excited state density functionals lead to the inaccuracy of quantum calculations. Here we attempt to build up an entirely new kernel for the quantum computation by implementing the neural network in the first principles calculation to solve this accuracy problem. Oligomers are chosen as the model to study and to validate our strategy in practical situation.

The combined kernel shows desirable predictive power to the experimental result. This can potentially replace many parts of experiments which are difficult and expensive to be carried out in the laboratory.

HPC Cluster provides excellent computational resources and services including Linux clusters, computational software (GUASSIAN98) and technical consultations (in particular, Mr. Woo & Mr. Kwan).

- Back to top- / Contents