Commonly used parameters
This is a detailed explanation of a few input parameters that are most commonly used by MDANSE. A full list can be found in section Input Parameter Types.
Frames
Parameter |
Format |
Default Value |
Description |
|---|---|---|---|
First frame |
int |
0 |
The frame from which the analysis will begin, the first frame taken into account. |
Last frame |
int |
Last frame |
The frame until which the analysis proceeds. The last frame taken into account. |
Frame step |
int |
1 |
Determines the periodicity of which steps are used and which are skipped. 1 means that all frames are read, 2 means every other is read, etc. |
Correlation Frames
Parameter |
Format |
Default Value |
Description |
|---|---|---|---|
First frame |
int |
0 |
The frame from which the analysis will begin, the first frame taken into account. |
Last frame |
int |
Last frame |
The frame until which the analysis proceeds. The last frame taken into account. |
Frame step |
int |
1 |
Determines the periodicity of which steps are used and which are skipped. 1 means that all frames are read, 2 means every other is read, etc. |
Correlation frames |
int |
Last frame / 2 |
Determines the size of the correlation window (in frames). This determines the time range of the output of the analysis. See also Time Series. |
Q shells
Input |
Format |
Default |
Description |
|---|---|---|---|
from |
float |
0 |
The lowest value of \(q\) to be used in \(\mathbf{q}\)-vector generation. |
to |
float |
10 |
The highest value of \(q\) to be used in \(\mathbf{q}\)-vector generation. |
by step of |
float |
1 |
The step by which \(q\) is incremented when changing from one \(\mathbf{q}\)-shell to the next one. Please adjust the width input parameter accordingly when changing the step. |
The unit of the \(\mathbf{q}\)-vector length in MDANSE is \(\text{nm}^{-1}\).
Output trajectory
This is used in every converter, and a few analysis jobs that also output a trajectory (CenterOfMassesTrajectory, TrajectoryEditor).
Input |
Format |
Default |
Description |
|---|---|---|---|
filename |
str |
The full path to the new file that will be created by this run. |
|
precision |
int |
64 |
Number of bits used for writing out floating point numbers. 64, 32 and 16 are possible |
chunk size |
int |
128 |
Number of atoms to be included in a single chunk of the HDF5 datasets in the output trajectory. An entire chunk must be loaded to access any number in that chunk. Smaller number means faster analysis, but also larger files. There rarely any advantage to chunks smaller than 128. |
compression |
str |
gzip |
Can be ‘none’, ‘gzip’ or ‘lzf’. ‘none’ means no compression. For most trajectories, compressing the atom coordinate arrays visibly reduces the file size at an expense of only a minor slowing down. |
log level |
str |
no logs |
Can be “no logs”, “DEBUG”, “INFO”, “WARN”, “ERROR” or “CRITICAL”. Not relevant to the output trajectory itself, but will change the amount of output in the log file of the converter run. |
Output files
Most analysis types define their output file using this.
Input |
Format |
Default |
Description |
|---|---|---|---|
filename |
str |
The full path to the new file that will be created by this run. Since multiple formats can be selected, it can be used as the base name to which different exten- sions will be appended for different output formats. |
|
format |
list[str] |
MDAFormat |
Can be [“MDAFormat”], [“TextFormat”] or both can be used with [“MDAFormat”, “TextFormat”]. MDA format is the only format used by the MDANSE_GUI. |
log level |
str |
no logs |
Can be “no logs”, “DEBUG”, “INFO”, “WARN”, “ERROR” or “CRITICAL”. Not relevant to the output trajectory itself, but will change the amount of output in the log file of the converter run. |
Atom Selection
Atom Selection allows you to select any set of atoms and/or other particles. These selected particles are then the ones that are made the target of the analysis. There is no limit to which particles can be included in a selection, or to how many selections can be used simultaneously. Atom Selection is entirely optional; if it is omitted, all the particles in the simulation are used.
More information about atom selection can be found here: Atom Selection.
Atom Transmutation
Atom transmutation uses the same interface as atom selection. Once you have selected the atoms you wanted to transmute, you can choose what chemical elements to replace them with, and add this change to the total transmutation mapping.
Atom Charges
The partial charge setting uses the same interface as the atom selection. You will most likely need to create several selections one after another. Every time you have selected atoms which should have the same charge assigned, set their charge and reset the selection. Once all the charges have been set, confirm the changes by clicking “Use setting”.
q-vectors
\(\mathbf{q}\)-vectors can be created using several generators, which sample the reciprocal space in different ways. Specific generators are described in List of q-vector generators.
For vector generators requiring “shells” and “width” input, the “shell” input defines a range of bin centres, and “width” gives the bin width used for assigning vectors to shells. For example, shells = (4.0, 8.0, 2.0) will generate 3 bins, centred on 4.0, 6.0 and 8.0 \(\text{nm}^{-1}\). The bins will contain vectors with \(q\) in the ranges of 3.5-4.5, 5.5-6.5 and 7.5-8.5, respectively. That is, for each bin the range of accepted \(q\) is \((\text{centre}-0.5*\text{width},\, \text{centre}+0.5*\text{width})\).
Vector generators with “Lattice” in their name generate vectors commensurate with the reciprocal lattice of the simulation box. This is important for analysis types which calculate correlations of pairs of atoms (Current Correlation Function and Dynamic Coherent Structure Factor).
While Dynamic Incoherent Structure Factor does not require a Lattice vector generator, it is still necessary to use one if you intend to combine the coherent and incoherent parts into the total signal using Neutron Dynamic Total Structure Factor.
Lattice vector generators initially generate a random distribution of vectors allowing also fractional coordinates, then round them to the nearest full lattice vector. Typically, there will be multiple vectors corresponding to a single lattice vector. The number of the initial vectors that ended up rounded to the same value is used as a weight factor, increasing the contribution of the vectors that were more frequently sampled. This should improve the results for low-symmetry unit cells, where lattice vectors are not uniformly distributed. Also, this means that the number of generated vectors can be increased significantly for lattice vector generators. The calculation will only be performed for the unique vectors, so once all the possible vectors in a shell have been generated, increasing the number of vectors further will improve the accuracy of the weight factors without increasing the computational effort.
In the MDANSE GUI it is possible to preview the number and distribution of generated vectors before running the analysis.
Instrument resolution
This option is available in all the analyses performing a time Fourier Transform, e.g. for the calculation of the density of states or the dynamic structure factor. The following resolution shapes are supported in MDANSE at the moment:
Gaussian
('gaussian', {'mu': 0.0, 'sigma': 1.0})Lorentzian
('lorentzian', {'mu': 0.0, 'sigma': 1.0})Pseudo-Voigt
The corresponding MDANSE input is:
('pseudo-voigt', {'eta': 0.5, 'mu_lorentzian': 0.0, 'sigma_lorentzian': 1.0, 'mu_gaussian': 0.0, 'sigma_gaussian': 1.0})square
The corresponding MDANSE input is:
('square', {'mu': 0.0, 'sigma': 1.0})triangular
The corresponding MDANSE input is:
('triangular', {'mu': 0.0, 'sigma': 1.0})ideal The corresponding MDANSE input is:
('ideal', {})