MDANSE icon indicating copy to clipboard operation
MDANSE copied to clipboard

Beta release 2024

Open MBartkowiakSTFC opened this issue 1 year ago • 1 comments

Description of work Final changes before the beta release of MDANSE 2.

Fixes

  1. Updated version numbers of MDANSE and MDANSE_GUI,
  2. Splash screen text changed to 'beta release',
  3. Correction to the job progress bar, to ensure that a finished job is 100% complete.
  4. Initialised the primary axis field in the plotting GUI with the correct axis name.
  5. Correction to the Grid.py file, allowing to plot curves with different primary axes.

To test Run a DOS job on a trajectory. Check that it reached 100% when finished. Plot the results of DOS and VACF at the same timer. Switch the plot to Grid. All selected curves should be visible.

MBartkowiakSTFC avatar Oct 10 '24 14:10 MBartkowiakSTFC

Looks, good so far.

I've found one issue with the project coordinates configurator. For the axial and planar projection, you can input the vector of zeros, this causes the jobs to fail. Also, non-numerical numbers can be entered into the boxes, hitting run doesn't do anything but we should stop the user from entering these like the other configurators.

image

ChiCheng45 avatar Oct 14 '24 08:10 ChiCheng45

The projection widget should work correctly now.

MBartkowiakSTFC avatar Oct 14 '24 12:10 MBartkowiakSTFC

So far I have found some minor issues (#577, #576, #575 and #574) which I think we need not fix for this release.

There is a lammps trajectory conversion failure with the p_lammps_10000 test file from one drive. This does not work on protos either and the issue appears to have been introduced in #546. I tested MDANSE at a98af806e87e9c9ba7bdbcf208e4142a8d38af8f and the conversion works fine at this point. I think we should try to fix this before the release.

ChiCheng45 avatar Oct 14 '24 17:10 ChiCheng45

The problem is that LAMMPS has multiple possible formats of the configuration file, and specifically this section: https://docs.lammps.org/atom_style.html is what creates the problem. Depending on which one is selected in the LAMMPS script, the number of columns in the Atoms section and their meaning will be different.

Some of the files conveniently contain a comment after the word 'Atoms' which specified which atom_style is being used. In the long run we should probably aim to support all of them, but there are quite a few possibilities. We can't really rely on the comment being there, and being correct, so some other mechanism of recognising the atom_style will have to be implemented.

Having said that, I think I fixed the problem with our example trajectory. It should work again.

MBartkowiakSTFC avatar Oct 14 '24 19:10 MBartkowiakSTFC

It looks good all my lammps test trajectories work fine now.

Can we update the self-part of the van Hove function so that it gives the same error as the distinct part when a simulation without a unit cell is used?

This is error in the self-part

179, in run_step
.unit_cell.volume
^^^^^^^^^
AttributeError: 'RealConfiguration' object has no attribute 'unit_cell'

This is the error with the distinct-part

line 81, in detailed_unit_cell_error
raise ValueError(
ValueError: This analysis job requires a unit cell (simulation box) to be defined. The box will be used for calculating density in the analysis. You can add a simulation box to the trajectory using the TrajectoryEditor job. Be careful adding the simulation box, as the wrong dimensions can render the results meaningless.

ChiCheng45 avatar Oct 15 '24 08:10 ChiCheng45

I have identified two new issues we may want to try to fix #579 and maybe #581. For #581 It might be better to deactivate the unfolded trajectory job for this release.

ChiCheng45 avatar Oct 15 '24 10:10 ChiCheng45

Thanks for finding these problems! In the meantime I added the error message to the van Hove self function.

Also, I think that the commits I added today should close #580 #576 #577 and make the AverageStructure analysis fail sooner, which does not solve the problem, but is an improvement.

MBartkowiakSTFC avatar Oct 15 '24 10:10 MBartkowiakSTFC

Looks like #580, #576 and #577 are fixed now.

Still one minor visual issue with the q_vector generator is that when you use a vector or range of 4 or more elements the cell in the table is not highlighted. The run button is blocked so it is not much of an issue, so I think we can leave this as it is.

image

ChiCheng45 avatar Oct 15 '24 11:10 ChiCheng45

To address #582 I added placeholder text to the instrument input fields. The values are not very useful (they are all 1) but at least this produces a valid configuration and it is clear what values were used.

MBartkowiakSTFC avatar Oct 15 '24 12:10 MBartkowiakSTFC

Also, #579 should be fixed now. #581 has been addressed by disabling UnfoldedTrajectory for now.

MBartkowiakSTFC avatar Oct 15 '24 12:10 MBartkowiakSTFC