Deferred Structural Elucidation Analysis for Molecular Simulations

Classify ice in MD trajectories.

Version 2 is one C++ engine and three packages. The 2020 yodaStruct binary is gone. 2.x has an expanded team.

2.2

One engine. Three packages.

The classifier is libyodaLib. Each language takes it as a Meson subproject. 2.x has an expanded team; names for the Python binding live on the pydseams book.

1.x / 2020

One yodaStruct executable. YAML plus Lua. JCIM paper.

2.x

seams CLI. pydseams and dseams in their own repos. linkcell for periodic kNN.

Elja exclusive Xeon Platinum 8358, job 1785712. Base `1c8efa3` vs tip `13c95ce`. Times in ms, best-of. Manifest: `repro/reference/paper_manifest-elja-1785712.json`.
Stage	N	1.x scan	2.x
neighbour list	32000	862.2	43.5
ring index	32000	5955.5	42.4
CHILL+	32000	111.5	44.8
six-rings	4096	100.7	13.9
findHC	4096	3582.9	45.2

The suite

Call it

CLI

seams read water.lammpstrj
seams chill-plus water.lammpstrj --cutoff 3.5
seams cages water.lammpstrj

Python

import pydseams as ds
frame = ds.read("water.lammpstrj")
print(frame.chill_plus())
print(frame.cages())

Lua

local dseams = require("dseams")
local cloud = dseams.read("water.lammpstrj")
print(dseams.chill_plus(cloud, {cutoff = 3.5}))

Two classifiers

CHILL+

Four-neighbour star. Labels cubic, hexagonal, interfacial, clathrate, and water. It does not assign cage membership.

Cages

Hexagonal cages (HC) are ice Ih. Double-diamond cages (DDC) are ice Ic. Everything else is water on that score.

Cubic, hexagonal, and interfacial ice forming on a surface — Ic, Ih, and interfacial ice on a surface.

Hexagonal ice as a topological ring network — Pure hexagonal ice as a ring network.

Radial distribution plots of a bulk system at several temperatures — Radial distribution of a bulk system at several temperatures.

Lua modules loaded without recompilation — Lua modules without recompilation.

Demos

1.x

The published nix install of the single yodaStruct binary. Keep this recording. Do not retitle it as 2.x.

2.x

Needs its own recording: seams, pip install pydseams, require("dseams"). Until that exists, the snippets above are the demo.

1.x recording

Cite

Goswami, R.; Goswami, A.; Singh, J. K. d-SEAMS: Deferred Structural Elucidation Analysis for Molecular Simulations. J. Chem. Inf. Model. 2020. 10.1021/acs.jcim.0c00031 / arXiv:1909.09830

Software 2.2 authors are on each package CITATION.cff.