Trajectories and Frames

An N-body simulation, such as a molecular dynamics simulation, fundamentally consists of a set of frames - snapshots of the system describing the particles and their positions. A trajectory is a set of these frames, forming a three dimensional ‘movie’. Simulations are essentially trajectories which are being generated on the fly.

Each frame consists of information about the system and the atoms within it. This data takes the form of values related to the frame (such as the system’s temperature or energy) and arrays storing information about the atoms themselves (such as position or element).

The approach taken by Narupa is to use a key value system to store frames, as what is available in a frame varies dramatically based on what is being simulated. Narupa applications use a set of standard keys to define common fields, such as ‘particle_positions’ for the the field containing the coordinates of the particles in the system. See the python implementation of FrameData or some of our conversion utilities for more examples.

For a basic interactive molecular simulation, a frame typically has the following fields:

• Particle positions (we always use nanometers!)
• Particle elements
• Bond pairs

For a simulation of a protein, a frame will have additional fields, such as:

• Particle residues
• Residue names
• Residue IDs
• Residue chains
• Chain names

While in a QM calculation, we may have:

• Bond orders

If you need to customize a frame with your own data, you can - just pick a good key by which to identify it, and convert your data into either an array of integers, bytes, floats, strings, or a single value type such as integer, byte, float, string, or a JSON like Struct.