Modern Fortran toolkit for reading in and analyzing DCD simulation trajectories output from LAMMPS
Clone or download
Failed to load latest commit information.
.github update issue template Oct 9, 2017
docs fix and update nframes documentation Feb 28, 2018
example fix undeclard variables Jan 23, 2018
meson update link to docs Feb 14, 2018
src update copyright year Jul 18, 2018
.gitignore update Sep 10, 2017
LICENSE Remove vmd plugin. Create native reader. Oct 9, 2017 typo Feb 28, 2018 require gcc >= 7.1 Feb 8, 2018


Copyright (C) 2017,2018 James W. Barnett



Fortran library for natively reading in DCD trajectory files generated from LAMMPS simulations for analysis. Uses an object-oriented style. For example, you can simply read in all simulation snapshots from a DCD file by adding the following lines:

using dcdfort_trajectory
type(Trajectory) :: trj

call trj%read("mytrajectoryfile.dcd")

Now all information from the trajectory file (atom coordinates, box dimensions) is accessible via object getters. There is additionally support for GROMACS-style index files and groups. Basic utility functions are also provided (e.g., pbc and distance). See the API below.

This is similar to my other project libgmxfort, except that this project does not require any plugins to read the binary trajectory files.

Note: DCD files generated from simulation packages other than LAMMPS will probably not work with this library. LAMMPS outputs DCD files as 32-bit CHARMM files with a unit cell and three dimensions, which is what this library can read in.

Build requirements

  • gfortran >= 7.1 - Required because we use convert=swap with open, which is a GNU-specific extension. Additionally version 7.1 added the ability to use non-constant error stop codes, which we use.
  • coreutils >= 8.23 - Allows the use of -D and -t together in install, which we use to install the Fortran .mod files. You can use an older version; you just have to manually create the include directory where the module files will be installed.
  • meson
  • ninja


After cloning the repository, or extracting the release tarball, cd into the repository. Then:

meson --buildtype=release build
ninja -C build


To test your build, do:

ninja -C build test

If any tests do not pass, please file an issue.


The following will install the library to the location specified by the meson flag --prefix, which is /usr/local by default.

ninja -C build install


Compile your Fortran trajectory analysis program with -ldcdfort. You may also need to use -I to point to where the modules files are even with all of the right environment variables set (by default at /usr/local/include).


A pkg-config file is included, so that it can be used in your program compilations. You may need to set the PKG_CONFIG_PATH environment variable to find the file (by default in the directory /usr/local/lib/pkgconfig). See man 1 pkg-config for more information.


Add use dcdfort_trajectory to your Fortran program in order to use the Trajectory class and use dcdfort_utils in order to use any of the other utilities. There is an example in the example folder on how to do this.

Full API documentation is here:

Reading in trajectory and index files

Typically you will open a trajectory file (and optionally a corresponding GROMACS-style index file). Then you will read in the entire trajectory file at once, or you can read it in in chunks. Then you should close the trajectory file when done.

The simplest way to use this library is to construct a Trajectory object and then use the read() method:

use dcdfort_trajectory
implicit none
type(Trajectory) :: trj
call trj%read("traj.dcd")

If you have a corresponding index file, you can add a second argument to open:

call trj%read("traj.dcd", "index.ndx")

Now information regarding the index groups is stored in memory and can be used in some of the following methods.

The read() method opens the dcd file, reads in all information, and then closes it. The trj object in this example now stores all of the coordinates and information from the .dcd file.

To skip the first portion of a trajectory file with read() use the skip argument. The following skips the first 100 frames before reading them into memory.

call trj%read("traj.dcd", "index.ndx", skip=100)

To only ready in every so many frames, use the every argument. The following reads in only every 10th snapshot into memory:

call trj%read("traj.dcd", "index.ndx", every=10)

To limit the number of frames read in, use the last argument, which specifies the last frame to read in numbered relative to the number of frames in the trajectory file. This is not necessarily the number of frames that you will read in when combined with skip and every.

call trj%read("traj.dcd", "index.ndx", last=1000)

All of these arguments can be used together.

If you want to read in the trajectory file in frame-by-frame use read_next() instead of read(). To use this, you must additionally open and close the dcd file on your own. By default it reads in one frame:

integer :: n
call trj%open("traj.dcd", "index.ndx")
n = trj%read_next()
call trj%close()

To read in more than one, specify an argument. The following reads in 10 frames:

n = trj%read_next(10)

read_next() returns the number of frames actually read in. It is a function, and not a subroutine. This is useful for using it with a do while loop. For example:

use dcdfort_trajectory

implicit none

type(Trajectory) :: trj
integer :: i, n

call trj%open("traj.dcd", "index.ndx")

n = trj%read_next(10)
do while (n > 0)
    do i = 1, n
        ! do some things with the frames read in
    end do
    n = trj%read_next(10)
end do

call trj%close()

To skip a frame without reading it into memory use skip_next(). You can also pass an integer argument to indicate how many frames to skip. The function returns the actual number of frames skipped (you might be near the end of the file and not able to skip all you specified).

Getting simulation information

After calling read() or read_next() every atom's coordinates are accessible via the x() method. For example, to get the coordinates of the first atom in the first frame you would do the following. The frame is the first argument and the atom number is the second argument.

real :: myatom(3)
! ...
myatom = trj%x(1, 1)

Note: Fortran uses one-based indexing, and that convention is retained here.

If you read in an index file, you can get atom coordinates in relationship to that. The following gets the fifth atom in index group C in the 10th frame:

myatom = trj%x(10, 5, "C")

If the index group does not exist, then an error will be thrown, causing the program to stop.

Note: If you have more than one group in your index file with the same name, this will simply use the first group with that name. It's best not to repeat group names in your index file. The library will give you a warning if it finds that an index name is duplicated, but the program will continue.

If you want direct access to the object storing a coordinate, do the following use trj%frameArray(i)%xyz(j,k) where i is the frame number, j are the x, y, and z coordinates (so 1, 2, and 3), and k is the atom number. The x() method is just a convenient way to get this object.

Note that when you use x() you will still have to give it the frame number as the first argument even if you only read in one frame with read_next(). You can always get the total number of frames in a trajectory file object with the nframes member:

integer :: n
! ...
n = trj%nframes

This is distinct from the number of frames read in using read_next(). The frame number passed to the x() method, and other methods here, is always in relationship to the number of frames read in, not the total number of frames in the file. To get the number of frames read in using read() use:

integer :: n
! ...
n = trj%frames_read

To get the timestep corresponding with the first saved frame in the trajectory file do:

integer :: istart
! ...
istart = trj%istart

To get the timestep corresponding with the last saved frame in the trajectory file do:

integer :: iend
! ...
iend = trj%iend

To get how often frames were saved in your simulation to this trajectory file use the nevery object. This corresponds with the fifth column in a LAMMPS dump dcd line where you indicated to dump every this many timesteps. It is the column labeled N in the LAMMPS dump manual page.

real(8) :: nevery
! ...
nsavc = trj%nevery

To get the simulation timestep, use the timestep object. This corresponds to the timestep setting in LAMMPS.

real(8) :: timestep
! ...
delta = trj%timestep

Warning: Some programs such as catdcd overwrite time step information. dcdfort outputs this information whenever it opens a file. If you intend on using this information in your analysis program, double check that it is correct. If you are only using LAMMPS output, you shouldn't have to worry about this.

You can also get the number of atoms with the natoms() method:

integer :: n
! ...
n = trj%natoms()

If you want to know how many atoms are in an index group include the group name as an argument. In this example the group name is "C":

n = trj%natoms("C")

If that index group does not exist, then the method will simply return 0.

To get the box coordinates, use box. The following gets the box of the 2nd frame:

real(8) :: mybox(6)
! ...
mybox = trj%box(2)

The first three elements of the array are the lengths of the unit cell. In an orthogonal simulation these are equivalent to the x, y, and z dimensions. With a triclinic box, these are the length of the unit cell vector along the x-axis (A), the length of the unit cell vector in the xy-plane (B), and the length of the unit cell vector in the yz-plane (C). The last three elements are the cosine of the box angles alpha, beta, and gamma. alpha is the angle between B and C, beta is the angle between A and C, and gamma is the angle between A and B.

Reading in specific groups only

As shown above, the most common use of this library is to use read() or read_next() to save all atom locations and then use getters like x() and natoms() to get information about them by specifying an index group as an argument.

To save memory, you can save just a specific index group with read():

trj%read(xtcfile, ndxfile, "C")

If you do this, you only have access to the group above, and you should never pass an index group name to getters like x(), since only one group is available. If you do specify a group in a getter after already specifying it in read() or read_next(), you will get an error, and the program will stop.


There are several functions and subroutines in the dcdfort_utils module, including periodic boundary and distance calculations. Check out the source file and full API documentation for what is available.


This library will only ever read in and process DCD files and auxillary files that are useful in analyzing trajectories (for example, GROMACS-style index files). Support for a wider range of DCD file formats could be added but is not planned at this time.

Only basic utility functions will ever be provided.


This project is released under the following license.


Copyright (C) 2017,2018 James W. Barnett

This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation; either version 2 of the License, or (at your option) any later

This program is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A
PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc., 51 Franklin
Street, Fifth Floor, Boston, MA 02110-1301 USA.

See the file LICENSE for full details.