Deferred Structural Elucidation Analysis for Molecular Simulations

Representative image of various ice types (Ic, Ih and Interfacial) forming on a surface


Developed actively at the Computational Nano Science Group of IIT Kanpur, under the aegis of Jayant K. Singh and is being prepared for publication.

d-SEAMS is a collection of analysis tools for molecular simulations. It has been implemented as a High Performance Cluster enabled engine in C++, with extensions via the Lua scripting interface. The code sheds light on the structures from simulation trajectories, in terms of their topology and it has been tailored for the analysis of nucleating systems. The framework is meant to be interfaced to a large variety of external tools and software, including R and other libraries. It is also the first and at-present only scientific analysis tool to use reproducible builds via nix.

The full documentation and installation instructions are here: https://docs.dseams.info

Radial Distribution Plots of a bulk system at various temperatures.

Software Features

d-SEAMS, as a toolkit has features intended both for long-term mantainance and extensibility, but also high performance insightful execution.

  • Embedded Lua interpreter
  • High Performance Cluster Ready
  • Plotting plugins
  • YAML input
  • Deterministic OS-Independent Execution (nix)
  • C++ engine with native LAMMPS support

Plugin Overview

d-SEAMS, as a scientific tool was initially designed to overcome analysis bottlenecks for nucleating systems simulated with LAMMPS.

  • Radial Distribution Function (+in-plane)
  • Mean First Passage Method
  • Connectivity Analysis
  • Ice Type Distribution (CHILL, CHILL+)

First Installation

The documentation and README are very detailed, but given the nature of the nix setup, the following first installation video is pertinent and serves as a brief tutorial.

Image Overview

Some sample images encountered during use.

Get The Code

Contributors welcome! As a FOSS project, we welcome contributions in any form, especially documentation and analysis methods.

Get in touch

We prefer GitHub issues. However, for joint research endeavors, use the form below.