Automated thermodynamic database development within the CALPHAD method using pycalphad
Some in the POSCAR format are available here for some simple phases: http://www.phases.psu.edu/?page_id=77
@bocklund Hi, Brandon, in the calculation of Ag-Cu phase diagram, I use Ag3Cu,AgCu and AgCu3 solid solution structure for phonopy calculation, that is, replace one or more copper atoms in the FCC cell, which are all fcc-A1 phases. May I ask how to configure HM,SM with temperature variation in the input file? In my opinion, sublattice_site_ratios are [1, 1], so how to distinguish Ag3Cu,AgCu and AgCu3 solid solution structure when the variable is T?
Is this correct? If I use phonopy to calculate HM,SM,CPM as a function of temperature, what is the format of the input file when the variable is just T?
My current computer can't use espei = 0.6 again, so I don't know which version to use, I don't know what went wrong.
I always get MPI errors when I use espei = 0.6
That error looks like an old bug
conda list is probably the easiest way
You should be running pycalphad 0.8
@npaulson You (and others) may find this new MCMC package interesting: https://pymcmcstat.readthedocs.io/en/latest/
@bocklund I found that the input data for Ni3Al on github is different from the refence of "Shunli DFT calculations (2010)". Is the reference "First-principles thermodynamics from phonon and Debye model: Application to Ni and Ni3Al"? The HM, SM, CPM data obtained from Fig7 and 8 are not the same as the data on github. For example, the HM value at 1615K, github is 39428.25, while in the literature is 157430.
What is the cause? Do you need a numerical conversion?
What is the cause? Do you need a numerical conversion?
@bocklund @richardotis I've been working on PDUQ (https://github.com/npaulson/pduq) a bit and I believe it is finally in pretty good shape (including documentation, https://pduq.readthedocs.io/en/latest/index.html). If you have the chance, take a look and perhaps even try installing/doing some analyses. Any feedback would be fantastic before I do an official release.
sorry, this should work for the documentation: https://pduq.readthedocs.io/en/latest/
Care to comment on what's going on in this figure? https://pduq.readthedocs.io/en/latest/_images/phase-fraction_1000K.png
I think it makes some sense when looking at this figure at the same time: https://pduq.readthedocs.io/en/latest/_images/phase-diagram_superimposed.png
It's because we are right at 1000K around the invariant
so there's huge uncertainty in the stability of liquid, Laves and CuMg2
It totally looks wrong at first glance, but I think it's real
I have a use case for this, but I need to get some new functionality implemented in pycalphad first. I want to look at uncertainty for a liquidus temperature versus chemical potential plot
This is great. I'm going to work some links into the ESPEI docs that point people towards PDUQ for UQ. Does this follow exactly the Cu-Mg example on espei.org (particularly the espei-datasets data)? Most people who have come across ESPEI and attempted to use it have followed the example, so having continuity where they can verify their plots and diagnostics match the website would be good.
Another thing I think we have seen in pycalphad and ESPEI is that the literate programming notebook style is very informative, but copy-pasting code chunks are how people tend to use the examples. That spawned ESPEI's Recipes page, which collects the useful code to produce one output into one pastable chunk that you can link to when people ask "how do I plot the thermodynamic properties with my data", etc. The recipes I have up there now are commented, but don't have much explanation, since those chunks and pieces are distributed through the example
I don't think that my trace files will be exactly identical to yours unless we happened to use the exact same weights for the various datasets. I had done some messing around to ensure that I was getting good convergence. That said, the code examples should work "out of the box" with those traces
Maybe we can think about converging the examples so they are actually identical
I think similarly to ESPEI, the code chunks in PDUQ should be readily useable
The datasets in the espei.org example haven't been updated since we added the weighting feature last year, so it needs to be redone. If you want to share the datasets you've weighted with me or a make a PR against https://github.com/PhasesResearchLab/ESPEI-datasets, I can rerun the example to produce consistent output