GROMACSΒΆ

Build gmx-seams against conda-forge GROMACS 2026.3, or a later 2026 release. Eigen and a C++ compiler need to be on the path. The exports are the include and library paths of that prefix, including the AVX directories conda-forge uses.

pixi init -c conda-forge dseams-gromacs
pixi add --manifest-path dseams-gromacs/pixi.toml "gromacs>=2026.3,<2027"
eval "$(pixi shell-hook --manifest-path dseams-gromacs/pixi.toml)"
export PKG_CONFIG_PATH="$CONDA_PREFIX/lib/pkgconfig:$CONDA_PREFIX/lib.AVX2_256/pkgconfig:$CONDA_PREFIX/lib.AVX_256/pkgconfig${PKG_CONFIG_PATH:+:$PKG_CONFIG_PATH}"
export LIBRARY_PATH="$CONDA_PREFIX/lib:$CONDA_PREFIX/lib.AVX2_256${LIBRARY_PATH:+:$LIBRARY_PATH}"
export CPATH="$CONDA_PREFIX/include${CPATH:+:$CPATH}"
meson setup build -Dwith_gromacs=enabled
meson compile -C build

gmx-seams reads a trajectory in nanometers, multiplies the coordinates and the box by 10, and classifies the frame. Convert input/traj/ice1h.lammpstrj to a one-frame .gro and run:

gmx-seams -f ice1h.gro -cutoff 3.5

The tool prints two lines. ice on the first is the cubic total over every frame. hexagonal, six-rings, and q6 are the last frame. overhead_s is on the frames line and depends on the machine. This cell has 128 molecules, all hexagonal, so the cubic total is 0 and the last frame is:

gromacs hexagonal 128 six-rings 256 q6 0.625855

engines/gromacs/run_mdrun.sh is a second path. It keeps the first frame of engines/gromacs/ice.gro (the file has two), writes a frozen .tpr with nsteps 4, runs gmx mdrun, and then runs gmx-seams on the .xtc mdrun wrote. gmx-seams is a separate process after mdrun. That first frame, converted from nanometers to angstroms, is 64 cubic atoms, 128 primitive six-rings, and mean local q6 0.628539. The frames line sums the cubic count over the frames in the .xtc.