d-SEAMS

Author:

d-SEAMS core team

Read

Load a LAMMPS frame and print nop, frame, and box.

Quickstart
Classify

CHILL+ four-neighbour star: cubic, hexagonal, interfacial, water.

Classify ice on the mixed TIP4P example
Score

Cage membership: HC is ice Ih, DDC is ice Ic, else water.

CHILL+ versus cages

Overview

d-SEAMS 2.14 reads a molecular-dynamics trajectory and names the ice in it. libseams is the library: CHILL+, cages, primitive rings, and \(q_6\) share one cutoff neighbour list. The seams command, the C and C++ APIs, and the language bindings call that library.

Install the engine with one line.

meson setup build && meson compile -C build

Ten lines classify the cubic cell shipped in input/traj.

dump=input/traj/mW_cubic.lammpstrj
seams read "$dump" --type 1
seams chill-plus "$dump" --type 1 --cutoff 3.5
seams ovito-chill+ "$dump" --type 1 --cutoff 3.5
seams rings "$dump" --type 1 --cutoff 3.5
seams steinhardt "$dump" --type 1 --l 6
seams cages "$dump" --type 1
seams cages input/traj/genice_sI.lammpstrj --type 1 --signature 512 --cage-edge none
seams rdf "$dump" --pair-types 1,1 --cutoff 6 --bins 4
seams prisms "$dump" --type 1

The videos, the 1.x migration, the performance tables, and the 2.14.0 notes are the rest of this release.

d-SEAMS (Deferred Structural Elucidation Analysis for Molecular Simulations) classifies ice in molecular-dynamics trajectories. This repository is the C++ engine (libyodaLib) and the seams CLI.

The first classification is five lines. Install with pip install 'pydseamslib==2.14.0' for the Frame package, with conda-forge’s Meson toolchain, or with meson setup build && meson compile -C build, then run the Python and C programs on the quick start. Both report 8 cubic molecules.

seams read input/traj/ice1h.lammpstrj
seams chill-plus input/traj/ice1h.lammpstrj --cutoff 3.5
seams cages input/traj/ice1h.lammpstrj

2.14.0 ships libseams (seams.h), libseams_lang, the C and C++ APIs, and the LAMMPS, OpenMM, PLUMED, and in-situ engines. gmx-seams reads the trajectory after gmx mdrun. Each language binding is its own repository, released with the same version against this tag:

The repository map says what lives where.

The header Ecosystem menu on every page jumps between the engine, the bindings, and linkcell.

Build with a Nix flake (nix build, nix develop) or with pixi run setup && pixi run build && pixi run test.

Suite stack

        flowchart TD
  DUMP["LAMMPS and XYZ<br/>chemfiles and con"] --> LIB[libyodaLib]
  LC[linkcell] --> LIB
  LIB --> API[libseams]
  API --> CLI["seams command"]
  CLI --> OUT["stdout counts"]
  API --> LANG["C and C++ here<br/>Python, Lua, R, Julia,<br/>Rust, Fortran repositories"]
    

Documentation structure

This documentation follows the Diataxis framework.

Tutorials

Ice, clathrates, brine, and nucleation, each with a runnable check.

Worked examples
How-to

Cutoff, neighbours, CHILL+, rings, and the engine builds.

Tasks
Reference

C, C++, CLI, and the language bindings.

Lookup pages
Comparisons

OVITO, freud, ClavisCage TRACE, SQQ, and PLUMED, next to the tests that recompute them.

Head-to-head
Videos

Odysee recordings of the bindings and the tutorials.

Videos

The in-book C++ pages under api/ are Doxygen via doxyrest. A sibling doxyYoda HTML tree (inline source, treeview) is at *api-cpp* after pixi run -e docs docbld.

d-SEAMS 2.0 paper

The d-SEAMS 2.0 paper, its benchmark campaign, the record tables and the reproduction rules are in a private paper repository. The benchmark records and figure data are the Hugging Face datasets d-SEAMS/dseams2-benchmark-records and d-SEAMS/dseams2-paper-figure-data. The performance pages here show tables rendered from those records (docs/records/README.md names the revision).

License

MIT. Cite the 2020 JCIM paper (DOI 10.1021/acs.jcim.0c00031). See Citation.