Model-free simulation approach to molecular diffusion tensors: Water

This file contains part 2 (water) of the supplementary material for the following publication:

Title: Model-free simulation approach to molecular diffusion tensors
Authors: Guillaume Chevrot, Konrad Hinsen, Gerald R. Kneller
Journal: Journal of Chemical Physics 139, 154110 (2013)

It contains the software implementing the computations described in the article, the input datasets for water, the resulting output datasets, and the figures. A detailed list is given below.

The file water_diffusion.ap is a HDF5 file that can be read with any HDF5-compatible software, including the free HDFView package (http://www.hdfgroup.org/hdf-java-html/hdfview/). HDFView can be used to inspect the arrays and tables contained in this file. Reading the molecular structure for water requires software that understands the Mosaic data model (http://bitbucket.org/molsim/mosaic/).

The file has been prepared using the ActivePapers framework (http://bitbucket.org/khinsen/active_papers_py) for reproducible research. It contains all the software used in the study that is described in the article, as well as most of the data, starting from the rigid-body trajectories. The original all-atom Molecular Dynamics trajectories are not included because of their size. All software is written in the Python language. The ActivePapers framework keeps track of which data was generated using which script and also notes the user name, machine name, and versions of all software packages used when running each script.

Readers wishing to modify and re-run the scripts, or to run them on different input data, should download and install the ActivePapers framework.

 

Datasets contained in the file water_diffusion.ap

The input data to the computations in this file are rigid-body trajectories for each of the 511 water molecules in the original MD trajectory. This input data is too big to be provided here. The correlation functions and mean-square displacements are first computed individually for each water atom and then averaged. The single-molecule functions are also too big to be provided here, except for a subset of five water molecules for demonstration.

All single-molecule data has a numerical suffix. In the following list, only the first one ("_0") is shown.

1) Input data.

1.1) The reference structure of water, with the atomic masses, in Mosaic format.

  /data/reference_structures
    atom_masses
    configuration
    universe

1.2) The rigid-body trajectory, consisting of the time axis, the center-of-mass positions, and the orientation stored as a quaternion trajectory.

  /data/center_of_mass
  /data/orientation
  /data/time

These data sets are links to a separate file which is *not* provided because of its size (16 GB).

2) Correlation functions, mean-square displacements, and trajectory statistice (calclet "correlation_functions")

2.1) Means and variances of the linear and angular velocities. Ideally, the means should be zero and the variances equal to their thermal equilibrium values.

  /data/trajectory_statistics
    mean_angular_velocity_laboratory_frame
    mean_angular_velocity_molecular_frame
    mean_velocity_laboratory_frame
    mean_velocity_molecular_frame
    variance_angular_velocity_laboratory_frame
    variance_angular_velocity_molecular_frame
    variance_velocity_laboratory_frame
    variance_velocity_molecular_frame

2.2) Correlation functions and mean-square displacements.

The time axes, which are the same for all correlation functions and for all mean-square displacements.

  /data/correlation_function_time
  /data/mean_square_displacement_time

The functions averaged over all 511 water molecules.

  /data/averaged
    correlation_function_laboratory_frame
    correlation_function_molecular_frame
    mean_square_displacement_laboratory_frame
    mean_square_displacement_molecular_frame

The single-molecule functions for the first five molecules.

  /data/single_molecule
    correlation_function_laboratory_frame
    correlation_function_molecular_frame
    mean_square_displacement_laboratory_frame
    mean_square_displacement_molecular_frame

3) Manually set parameters used in computing the diffusion tensors and in prepapring the plots.

  correlation_function_integration_limit
  msd_fit_range
  msd_plot_range

4) Diffusion tensors from the Kubo relations (integral over the velocity correlation functions) and from the slope of the mean-square displacement.

4.1) The diffusion tensors computed from the data averaged over all molecules.

  /data/averaged
    diffusion_tensor_kubo_laboratory_frame
    diffusion_tensor_kubo_molecular_frame
    diffusion_tensor_msd_laboratory_frame
    diffusion_tensor_msd_molecular_frame

4.2) The diffusion tensors computed from single-molecule data for five molecules.

  /data/single_molecule/molecule_0
    diffusion_tensor_kubo_laboratory_frame
    diffusion_tensor_kubo_molecular_frame
    diffusion_tensor_msd_laboratory_frame
    diffusion_tensor_msd_molecular_frame

4.3) The variance of the diffusion tensor elements, computed over the five molecules but relative to the average over all 511 molecules.

  /data/single_molecule
    diffusion_tensor_variance_kubo_laboratory_frame
    diffusion_tensor_variance_kubo_molecular_frame
    diffusion_tensor_variance_msd_laboratory_frame
    diffusion_tensor_variance_msd_molecular_frame

4.4) The relative statistical error of the diffusion tensor elements, computed from the variances.

  /data/single_molecule
    diffusion_tensor_relative_error_kubo_laboratory_frame
    diffusion_tensor_relative_error_kubo_molecular_frame
    diffusion_tensor_relative_error_msd_laboratory_frame
    diffusion_tensor_relative_error_msd_molecular_frame