9. Workflow of the Analysis

Most analysis jobs offered by MDANSE follow the same pattern of execution:

9.1. Input

9.1.1. Trajectory

The trajectory must be in the MDANSE format, saved as a NetCDF file.

(Most likely your trajectory is in whatever format was output by your preferred Molecular Dynamics simulation software, and you need to convert it first. Once you have converted your trajectory to the MDANSE NetCDF format, you can use it as input for all kinds of analysis. See also Trajectory Converters)

9.1.2. Frames

It is not necessary to use all the time frames of your MD simulation. You can decide to limit the range of simulation time, and reduce the number of the frames taken in that range by increasing the step between them. Only the frames you selected will be passed to the analysis. See also Frames.

9.1.3. Atom selection

Just as it is not necessary to include all the time frames in the analysis, it is also possible to select only a subset of all the atoms present in the trajectory. Once you have defined a selection, you can decide to run an analysis on the selected atoms, and ignore the rest. This is useful if you are trying to determine which atoms contribute to a specific feature in your signal. See also Atom Selection.

9.2. Analysis

The analysis is run in steps, iterating over parts of the trajectory.

If you chose to determine the atom velocities by interpolation, it will be done at this stage.

The iterations over steps will produce partial results. This is where the specific equations described in the documentation of an analysis type are applied. The partial results will be combined into the final result in the next step of the workflow.

9.3. Finalising

At this stage the partial properties have been calculated, typically per atom type, or per pair of atom types. They will now be combined into the final result.

9.3.1. Resolution

If the analysis allows for applying instrumental resolution, it will be done first. The resolution is enabled only for the analysis types which calculate an energy spectrum. This is normally done by calculating a Fourier transform of a correlation function. The resolution is applied by multiplying the time-dependent function by a window function before applying the Fourier transform. The details are given in the section Instrument resolution.

9.3.2. Normalisation

If the analysis offers the option of normalising the results, it is done at this stage. The normalisation is described in the section Normalize.

9.3.3. Weighting

The partial properties calculated so far will now be combined using the weights chosen by the user, as described in the section Weights. Please remember that the MDANSE GUI normally recommends the weighting scheme appropriate to the type of analysis performed.

9.4. Writing output

All the output arrays created in the analysis are written to the filesystem in the format chosen by the user. (If you intend to continue visualising the results within the MDANSE GUI, you will have chosen the HDF5 format. If, however, you were planning to process the results further using other software, then you have most likely picked the ASCII output. See also Output files)