Master of Science in Computer Science (MSCS)


Computer Science

Document Type



Extracting and interpreting the information contained in large sets of time-varying three dimensional positional data for the constituent atoms of simulated material system is a challenging task. This thesis work reports our initial implementation of a web-based visualization system and its use-case study. The system allows the users to perform the desired visualization task on a web browser for the position-time series data extracted from the local or remote hosts. It involves a pre-processing step for data reduction, which involves skipping uninteresting parts of the data uniformly (at full atomic configuration level) or non-uniformly (at atomic species level or individual atom level). Atomic configuration at a given time step (snapshot) is rendered using the ball-stick representation and can be animated by rendering successive configurations. The entire atomic dynamics can be captured as the trajectories by rendering the atomic positions at all time steps together as points. The trajectories can be manipulated at both species and atomic levels so that we can focus on one or more trajectories of interest. They can be color-coded according to the additional information including the time elapsed and the distance traveled. The instantaneous atomic structure and the complete trajectories can be superimposed to help assess the 3D geometries and extents of the selected trajectories. The implementation was done using WebGL and Three.js for graphical rendering, HTML5 and Javascript for GUI, and Elasticsearch and JSON for data storage and retrieval within the Grails Framework. We have demonstrated the usefulness of our visualization system by analyzing the simulated position-time series for proton-bearing forsterite (Mg2SiO4) system – an abundant mineral of Earths upper mantle. Visualization reveals that protons (hydrogen ions) incorporated as interstitials are much more mobile than protons substituting the host Mg and Si cation sites. The proton diffusion appears to be anisotropic with high mobility along the x-direction, showing limited discrete jumps in other two directions. Our work at the present represents a simplistic (direct) web-based rendering of large atomic data sets. While the atomic structure can be animated at an interactive rate, the trajectory processing is slow, taking several minutes. We anticipate to further improve the system and use it in gaining useful structural and dynamical information from more materials simulation data.



Document Availability at the Time of Submission

Release the entire work immediately for access worldwide.

Committee Chair

Karki, B., Bijaya