Development of computational methods for materials modelling

The computational methods for modelling the quantum materials developed by the Quantum Chemistry Group at Lodz University Technology have been implemented in popular computational software packages produced in the United States, including Q-Chem, Psi4, NWChem, and Gamess US. They have also been implemented in a library of density functionals called libxc, available in 36 software packages including such popular products as Molpro, Turbomole, and Orca. Consequently, the approaches are available to research centers all over the world specializing in drug and material design, as well as to companies in the chemical and pharmaceutical industry.

The analysis prepared by Goldberg Consulting for the Psi-k scientific network on “Industry interactions of the electronic structure research community in Europe” in 2012 characterized the industry’s view of DFT (density functional theory) as “the most widely accepted quantum, atomistic, and molecular modelling method” and pointed out that the “software industry has emerged from a ‘hype cycle’ into a phase of sustained growth”.

On the other hand, one of the greatest drawbacks of DFT, especially for the purposes of modelling drugs and catalysts, which are often very large molecules, was its inability to correctly describe non-covalent interactions (NCIs). This changed with the development of approaches combining DFT with other methods to enable accurate modelling of NCIs. Our methods were part of the first generation of such approaches, and remain among the most accurate available today.

The methods described in [1] and [2] were featured in several popular science news platforms, including EurekAlert! where their importance for the pharmaceutical industry was highlighted. In addition, most of the major DFT software packages have incorporated at least one of the approaches into their code.

One of the codes which implemented dlDF, is Q-Chem, a commercial computational package from the US. The revenue of E-Chem Inc., whose only product is the Q-Chem code, is estimated to be between $3.4 and $4 million. Financial growth data for the company are not available. However, Q-Chem reports that from 2016 (the first implementation of dlDF in their code) to 2019 the number of citations of Q-chem per year grew by 20%, from 400 to 500. The Director of Q-Chem Inc. Peter Gill writes: “The inclusion of dlDF has expanded the functionality of Q-Chem, making it a more attractive tool for other researchers, and has helped to increase the number of academic and industrial Q-Chem licenses sold worldwide.”

Less straightforward to prove, but equally important, is the impact of the implementation of dlDF in open-source packages, including NWChem. Marat Valiev, Team Lead in Systems Modeling and Computational Sciences for NWChem, states that “availability of the dlDF method in NWChem has increased the versatility of the code. This and other recent developments in NWChem contribute to the continuing growth of the number of NWChem users worldwide.”

The dlDF functional is also implemented in the open-source software (OSS) Psi4 and Gamess US, as well as in the library libxc which is used in multiple codes. The beneficial impact of OSS on the reproducibility of chemical research and chemical education is well established. Up until recently, the economic influence of OSS had not been quantified. However, a study ordered by the European Commission revealed that “companies located in the EU invested around €1 billion in Open Source Software in 2018, which brought about a positive impact on the European economy of between €65 and €95 billion.” The study predicts that “an increase of 10% in contributions to Open Source Software code would annually generate an additional 0.4% to 0.6% GDP”.

The dlDF functional obviously contributes only a fraction to this growth. However, it does  enhance the value of the codes which have implemented it, and moreover it is accessible worldwide both commercially and for free.