Use a language binding¶
This repository builds the engine, the C API libseams (include/seams.h),
the C ABI libseams_lang (bindings/cxx/seams_lang.h), and the seams
command.
C and C++ document those libraries.
Every other language is its own repository, with its own documentation,
tests, and releases. A binding’s version is the seams-core version it was
built against. The Python release for this engine is pydseamslib 2.14.0.
Each repository’s README has the install command.
Languages¶
Language |
Repository |
Package |
Links |
|---|---|---|---|
Python |
|
|
|
Lua, Fennel |
Git source, |
|
|
R |
|
|
|
Julia |
|
|
|
Rust |
crates |
|
|
Fortran |
fpm Git dependency |
|
import dseams is the array API over libseams: dseams.Frame,
chill_plus, q4q6, neighbours, rings, and from_ase.
import pydseams and import pydseamslib are the trajectory package.
pydseams.read opens a dump. dseams.userframe is the Frame over
libseams_lang, documented on the
Python Frame page.
Dseams calls libseams_lang for a frame and libseams for the
re-exported seams_* functions. libseams_jll supplies the engine
binary. The same repository includes Seams, an example of the
libseams C API.
The R package compiles the engine it ships, and the calls in its README
use libseams_lang.
Ice Ih¶
input/traj/ice1h.lammpstrj is one orthorhombic frame of ice Ih: 128
oxygens, every atom LAMMPS type 1. The path is relative to a seams-core
checkout. dseams.Frame stores the coordinates and the three edge
lengths. Published CHILL+ labels every oxygen hexagonal, so counts[1]
is 128. The ring histogram holds 256 six-rings, so histogram[6] is 256.
The mean local q6 at 3.5 Å is 0.625855. The cutoff neighbour list holds
512 directed pairs, four per oxygen, so offsets[-1] is 512.
>>> import numpy as np
>>> import dseams
>>> def ice():
... xyz, edges, reading = [], [], False
... handle = open("input/traj/ice1h.lammpstrj")
... try:
... while True:
... line = handle.readline()
... if line == "":
... break
... if line.startswith("ITEM: BOX BOUNDS"):
... edges = []
... for _ in range(3):
... lo, hi = handle.readline().split()[:2]
... edges.append(float(hi) - float(lo))
... elif line.startswith("ITEM: ATOMS"):
... reading = True
... elif reading:
... parts = line.split()
... if len(parts) < 5:
... break
... xyz.append([float(v) for v in parts[2:5]])
... finally:
... handle.close()
... return xyz, edges
>>> xyz, edges = ice()
>>> len(xyz)
128
>>> frame = dseams.Frame()
>>> frame.set_frame(np.asarray(xyz), np.asarray(edges))
>>> frame.count()
128
>>> offsets, columns = frame.neighbours(3.5)
>>> int(offsets[-1])
512
>>> sizes, symmetry, histogram, fraction = frame.rings(3.5, 6)
>>> int(histogram[6])
256
>>> q4, q6 = frame.q4q6(3.5)
>>> round(float(q6.mean()), 6)
0.625855
>>> labels, counts = frame.chill_plus()
>>> int(counts[1])
128
These four commands print that census. seams rings prints two lines,
and the other three commands print one line each. seams steinhardt
also prints the Lechner-Dellago q6-bar. seams cn with this cutoff and
type prints nI 128, the number of atoms in the coordination sum.
seams read input/traj/ice1h.lammpstrj
seams rings input/traj/ice1h.lammpstrj --cutoff 3.5 --type 1
seams steinhardt input/traj/ice1h.lammpstrj --cutoff 3.5 --l 6 --type 1
seams chill-plus input/traj/ice1h.lammpstrj --cutoff 3.5 --type 1
nop 128 frame 1 box 15.6457 14.7071 18.073
nop 128 rings 256
6 256
nop 128 q6 0.625855 q6bar 0.562449
nop 128 hexagonal 128
dseams.from_ase returns positions and a nine-number LAMMPS cell for
set_frame. The Python Frame page
runs that round trip on this file.
Recordings¶
Each binding below has a short screencast. The full set is on the videos page.