Releases: molssi-seamm/diffusivity_step
Releases · molssi-seamm/diffusivity_step
Bugfix: Reinitialization of data in loops
- Fixed a bug where the data was not being correctly initialized if the step was in a loop. This caused the timing results to be incorrect.
- Added actual number of samples for the MSD and length of the Helfand integral to the results.
Added optional correction for cell size
- Added an option to use the Yeh-Hummer hydrostatic correction for the effects of the finite cell size. The viscosity is required as an input, but the correction eliminates the need to extrapolate to 1/L = 0.
- Added control parameters and timings to the available results.
Significant improvements!
- Simplified error analysis to safe approach of analyzing the diffusion constants over runs.
- Improved fitting of the curves to focus on the central linear portion. There are reasonable defaults but the user can adjust as needed.
- Provided a combined average and error bars when both the MSF approach and Helfand moments are used.
- Capture temperature, pressure, and cell size from the MD step, providing 1/L as a result since the true diffusion constants are found by extrapolating to 1/L = 0.
- Provided control over the number of steps for the expensive numerical integration in the Helfand moments, providing a reasonable default of 1000.
Bugfix: Significant error in Helfand Moment approach
- Now fixed and seems to be working. Has been tested reasonably well.
Improved fitting of curves
- Removed weighting of the fit by the stdev since it is too biased to the beginning
- Added control over the portion of the data to fit in order to avoid the initial curvature and poor data towards the end.
Bugfix: handling of options for subflowchart
- Fixed a bug where the options for the subflowchart were not being parsed correctly.
Updated for new task handling
- The new handling of running tasks such as LAMMPS required a small change in the code.
Changed default to MSD
The Helfand moments approach seems give incorrect results if the sampling time is too long. It is not dramatic, but gives increasingly incorrect results as the sampling time is increased. Thus using the Helfand moments is dangerous because the results may be wrong, but not obviously so.
Initial working release
2023.8.30 Fixed documentation requirements