Understand-Predict-Design is the motto of any computational research, and so is ours. Our group works on two main verticals. The first one is employing DFT-based models to understand reaction mechanisms at the atomistic level. The second one is to build ML models to predict novel materials.

With computers as a tool, we design virtual experiments to understand various aspects of heterogeneous catalysis. We model systems to understand reaction mechanisms. [1,2,3] We scan chemical space to design catalysts suitable for a specific reaction like CO2 to value added products [4, 5] or methane activation.[6] We do controlled experiments like investigating the interaction of methanol with various Zn-based systems to understand what leads to the reactivity of a specific composition/facet in terms of underlying electronic structure and develop fingerprints of reactivity.[7,8,9,10] In short, we use computers as a tool to explore the chemical world. Density functional theory is our probe for this exploration.
 
For those unaware of DFT, Density functional theory (DFT) is a computational method used to study the electronic structure of molecules and materials. It helps understand the behaviour of electrons in atoms, molecules, and solids and thus plays a crucial role in computational materials science and chemistry. DFT calculations are based on the principles of quantum mechanics, and they can provide valuable insights into the properties of materials and their behaviour under different conditions. DFT is a powerful tool for researchers to study the chemical world and design new materials with desired properties.
 
Recently, we have also ventured into developing machine learning-based models to predict novel materials or materials with targeted properties. Machine learning models learn and improve from the data without being explicitly programmed. It involves developing algorithms and statistical models that enable computers to recognize patterns in the data and make predictions. Machine learning has become an important tool in materials science research. ML has the potential to accelerate the discovery of new materials with desired properties when combined with traditional methods.
 
We have employed ML-based models to predict the interaction between a cluster and a molecule [11], the energy of a given cluster [12] or generating isomer families of an n-atom cluster from (n-1) atom cluster[13]. We demonstrate the advantages of our approach over conventional methods for scanning local minima on potential energy surface. Our approach generates isomer of gallium clusters at a minimal computational cost and predicts isomer families not discovered through previous density-functional theory (DFT)--based approaches. Last but not the least, we are working towards predicting alloys suitable for solid-state hydrogen storage [14]
 
We also extensively collaborate with various experimental groups to either explain their experiments results or validate our predictions.