Seminar #3 - Force and Energy Computation

2024 Winter Molecular Simulation Seminar
Room 302, the 2nd experimental building at POSTECH
Presenter: Seunghyun Kang

The physical quantity which contain the crucial information in molecular simulation is the PES (Potential Energy Surface) of interested system, considering that the evaluation of potential energy would lead to the force, velocity and acceleration of each atom in the system. Therefore, the accuracy and speed of computing potential energy is directly related to reliability and efficiency of simulation. This inspires the development of several algorithms to compute the potential energy given coordinates, in which some of them are adopted in MD simulation packages nowadays. However, the intrinsic nature of trade-off relation between accuracy and efficiency along with the inherent limitation and numerical error in energy computation algorithms would lead to the artifacts in simulation that describes unphysical physics from real-world physics. To ensure the accuracy and efficiency of simulation in computational chemistry research, therefore, understanding the theory and bottlenecks of several energy computation algorithms is of central request for establishing reasonable model system. Moreover, we can derive much more accurate, acceptable result and conclusions when we recognize the artifacts of the model system.

In this seminar, I’ll present ‘How potential energy calculation algorithm works and how it affects to result of the system?’. Addressing these questions are divided into four sections: First, I’ll introduce PBC (Periodic Boundary Condition)[1], minimum image convention, and finite size effect[2] of MD simulation. Then, the advantages and limitations of popularly used potential energy computation methodologies[3] (PME, P3M, Reaction Field, MEMD, especially sPME[4] and Reaction Field[5] which is being implemented in OpenMM) will be treated. Most of MD simulation packages, utilizes PME for nonbonding energy computation. The reasons and crucial artifacts of PME method[6,7] will be discussed in third question. Lastly, comparison between CPU work and GPU work in perspective of computation and acceleration strategies of nonbonding potential energy calculation by using GPUs[8,9] will be the last topic for this presentation.

Reference

[1] Stenhammar, J., Karlström, G., & Linse, P. Structural anisotropy in polar fluids subjected to periodic boundary conditions. J. Chem. Theory Comput., 2011, 7(12), 4165-4174.
[2] Celebi, A. T., Jamali, S. H., Bardow, A., Vlugt, T. J., & Moultos, O. A. Finite-size effects of diffusion coefficients computed from molecular dynamics: a review of what we have learned so far. Mol. Sim., 2021, 47(10-11), 831-845.
[3] Allen, M. P., & Tildesley, D. J. Computer simulation of liquids 2thed. Oxford univeristy press, 2017.
[4] Shirts, M. R., Mobley, D. L., Chodera, J. D., & Pande, V. S. Accurate and efficient corrections for missing dispersion interactions in molecular simulations. J. Phys. Chem. B, 2007, 111(45), 13052-13063.
[5] Tironi, I. G., Sperb, R., Smith, P. E., & van Gunsteren, W. F. A generalized reaction field method for molecular dynamics simulations. J. Chem. Phys., 1995, 102(13), 5451-5459.
[6] Hub, J. S., De Groot, B. L., Grubmüller, H., & Groenhof, G. Quantifying artifacts in Ewald simulations of inhomogeneous systems with a net charge. J. Chem. Theory Comput., 2014, 10(1), 381-390.
[7] Boresch, S., & Steinhauser, O. Presumed versus real artifacts of the Ewald summation technique: The importance of dielectric boundary conditions. Berichte der Bunsengesellschaft für physikalische Chemie, 1994, 101(7), 1019-1029.
[8] George, A., Mondal, S., Purnaprajna, M., & Athri, P. Review of Electrostatic Force Calculation Methods and Their Acceleration in Molecular Dynamics Packages Using Graphics Processors. ACS omega, 2022, 7(37), 32877-32896.
[9] Eastman, P., & Pande, V. S. Efficient nonbonded interactions for molecular dynamics on a graphics processing unit. J. Chem. Theory Comput., 2010, 31(6), 1268-1272.