ulrikfriisjensen / chili Goto Github PK

In order to free memory, we bin the histogram of interatomic distances when calculating the Debye equation. This will affect the precision. Make a figure showing the scattering with and without binning and calculate the precision.

See Colab notebook, where it is done.

In general, there are two ways to simulate scattering data: Using the Debye equation which is slow but accurate, or using an assumption of periodic boundary conditions.
For PDF, we make the assumption but add a variable taking into account the size of the material. Make a figure showing that it is a good assumption and it is significantly faster to calculate the PDF.
For SAS, we use the Debye Equation as the assumption is not good. Make a figure showing that.

Use the equations in simulateSAS.py to also simulate diffraction data. Change the Q-step size and the Q-range such that the Q-range is significantly larger and use a larger step size.

simulateSAS.py calculates the scattering pattern from a XYZ file.

Instead, cut out a discrete particle from a CIF file first, which is used to calculate the scattering pattern using the Debye equation.

For now, the code in the h5-constructor class simply ignores these double bonds. I have a failed attempt in the code reference below. We need to figure out either how to implement this in a consistent manner, or if we even need it, or if the supercell representation idea where all the input structures have roughly the same number of atoms (from meeting, 12/06-23) is a better idea overall.

https://github.com/UlrikFriisJensen/InOrgMatDataset/blob/1c5e703e2293807139e708e3ce4e0d1e3a3c4592/Code/h5Constructor.py#L93-L106

https://github.com/UlrikFriisJensen/InOrgMatDataset/blob/37d5c32b24b2b270509d8d6b4509330e24ee0a13/Code/simScatteringPatterns.py#LL57C9-L57C107

After the implementation of X-ray, Neutron and Electron PDF all of them produce monotonically increasing lines.

The simulatePDF.py and simulateSAS.py can only simulate X-ray scattering patterns.

Add neutron (and electron) patterns to it. How to do it:

• For PDF: We can do it using the PDFGenerator used to do fitting (stuff like PDFcalc.radiationType="N"), however, for the PDFCalculator for simulation, it does not work. make an issue in the DiffPy forum: https://groups.google.com/g/diffpy-users
• We using cromer-mann coefficient to calculate the Q-dependency of the formfactors for each atom. For neutrons it is way simpler. Find the form factor of each element and remove the Q-dependency.

Different diffpy elements are installed and from different channels when using the environment file vs doing a normal conda install. The top of the picture is with the environment file and bottom is with conda install