Machine Learning for the Acceleration of Quantum Chemical Simulations
AI and Machine Learning can accelerate quantum chemical simulations even on a budget
For many applications, it is not necessary to use very large and complex AI models. My research shows that relatively simple models for specific quantum chemical problems can already yield good results.
My work is at the intersection of computational chemistry, quantum chemistry and artificial intelligence. Quantum chemical simulations can be used to investigate the properties of molecules without the need for a laboratory experiment for every question. Such simulations can support research, but accurate calculations often require a lot of computing time and computer power.
Therefore, I focused on the question of how artificial intelligence can be used to perform certain quantum chemical calculations more efficiently. The goal was to obtain results of similar quality with less computational work.
My research shows that the most complex AI method is not automatically the best choice. Which approach is most suitable depends on the problem, the available data and the desired accuracy. For some quantum chemical applications, simple models may be sufficient. For other applications, a combination of artificial intelligence and existing scientific models works better.
More information on the thesis to follow.