AI Wave Goes Quantum

AI makes iconic Quantum Chemistry breakthrough

It was back in 1925 that an Austrian-Irish physicist named Erwin Schrödinger first postulated a linear partial differential function that not only described the wave function of a quantum mechanical system but also managed to completely transform the way that we knew and understand quantum chemistry at the time. About a century since, we have now managed to create an Artificial Intelligence system that can successfully solve the iconic equation of its own accord.

Waves of a Quantum Nature

“Central to both quantum chemistry and the Schrödinger equation is the wave function – a mathematical object that completely specifies the behaviour of the electrons in a molecule. The wave function is a high-dimensional entity, and it is therefore extremely difficult to capture all the nuances that encode how the individual electrons affect each other. Many methods of quantum chemistry in fact give up on expressing the wave function altogether, instead attempting only to determine the energy of a given molecule. This, however, requires approximations to be made, limiting the prediction quality of such methods.” (SciTechDaily)

The reason why this is news of significantly important measure, of course, is because this is a rather thankless task. In essence, the goal of quantum chemistry is to predict two uncertain states simultaneously: the “chemical and physical properties of molecules based solely on the arrangement of their atoms in space.” While this is something solving the Schrödinger’s equation gives us, this is a rather resource-intensive and time-consuming affair, requiring adept usage of laboratory experiments and technical knowhow.

This essentially reasserts the importance of the work being carried out by the scientists at Freie Universität at Berlin, making it all the more commendable. “Escaping the usual trade-off between accuracy and computational cost is the highest achievement in quantum chemistry,” according to Dr Jan Hermann of Freie Universität Berlin, a man deeply involved in the design of several key features of the method of study. “As yet, the most popular such outlier is the extremely cost-effective density functional theory. We believe that deep ‘Quantum Monte Carlo’ the approach we are proposing, could be equally, if not more successful. It offers unprecedented accuracy at a still acceptable computational cost.”

Enter: PauliNet

The use of a deep neural network in representing wave functions is a rather novel affair. It is essentially an artificial intelligence-based system that can compose a wave function based off relatively simple mathematical concepts to appropriately predict the complex placement of electrons around the nucleus of an atom. Called ‘PauliNet’ in commemoration of Austrian physicist Wolfgang Pauli, the AI system utilises the idea behind his famous ‘Exclusion Principle’: where the neural network architecture depicting the wave function must change signs when two electrons are exchanged – akin to the asymmetry noted in real-life electronic wave functions.

SciTechDaily writes: “Up to now, it has been impossible to find an exact solution for arbitrary molecules that can be efficiently computed. But the team at Freie Universität has developed a deep learning method that can achieve an unprecedented combination of accuracy and computational efficiency. AI has transformed many technological and scientific areas, from computer vision to materials science. “We believe that our approach may significantly impact the future of quantum chemistry,” says Professor Frank Noé, who led the team effort. The results were published in the reputed journal Nature Chemistry.”

There is, however, still a while to go until this system reaches industrial application. Yet, the authors agree that this is ‘fundamental research’ in exacting “a fresh approach to an age-old problem in the molecular and material sciences; and (are) excited about the possibilities it opens up.”

Acknowledgement: “Deep-neural-network solution of the electronic Schrödinger equation” by Jan Hermann, Zeno Schätzle and Frank Noé, 23 September 2020, Nature Chemistry.