Hi Brandon, I am trying to use pycalphad to simulate solidification behaviour of alloys 625 and Monel. I noticed that simulation ends at some temperature dependent on composition and temperature step, even if solidification is far from complete. I attach an extreme example for Monel, comparing temperature steps of 5K and 2K. The simulation with 2K stops prematurely. I noted a similar behavior with alloys 625, even if not so dramatic. Also changes in composition affects the end point.

I attach the code and the database I am using, derived from mc_ni_v2.034.tdb Thank you very much for your help and all you did with pycalphad!

MonelScheil.txt ARNi.txt

I am using a revised version of Ni-database from MatCalc. This database works with pycalphad 0.8.3. When I run scheil 0.1.1, I got error message: ALN: Sublattice frozenset({Species('N', 'N1')}) of (frozenset({Species('AL', 'AL1')}), frozenset({Species('N', 'N1')})) has no components in {Species('AL', 'AL1')}

Any suggestion?

Currently there are no checks for order/disorder phase name checking in simulate_equilibrium_solidification, so a order/disorder modeled phase that is in a disordered configuration will still show as the ordered phase name.

The main challenge in making this change that the order/disorder check assumes a pycalphad LightDataset, rather than xarray Dataset.

The code should be profiled and any optimizations made if possible and reasonable.

For example, equilibrium(..., to_xarray=False) could be used, but the indexing into the LightDataset objects would have to be updated

Peritectic-type reactions in the ternary go from L+A -> A+B, terminating the current Scheil algorithm. Section 12.8 in Hillert's Phase Equilibria, Phase Diagrams, and Phase Transformations (2nd Edition) may be insightful. Specifically, the idea below might be able to be used to determine when a peritectic region is found and how to update the composition of the liquid so that the liquidus can continue to be followed down the liquidus projection to a eutectic.

Introduces an OrderingRecord object to track the relationship between ordered and disordered phases, replacing the old ordering code. The new ordering approach can be used for arbitrary equilibrium calculations and is used to fix #8.

The logic for defining how to compare the disordered phases is currently the same as before, considering the site ratios to determine equivalent sublattices. I am deferring action on this with the issue filed in https://github.com/pycalphad/scheil/issues/23.

• Points dictionaries are now created for the user by default if they are not provided.
• Adaptive sampling does not do any "local" random sampling by default, it just adds the point from the solution
• Don't use pycalphad.core.utils.generate_dof to create the dof_dict (it is deprecated)

Uses the pycalphad API to pass PhaseRecord dictionaries to calculate and equilibrium to improve the tight-loop performance of equilibrium and Scheil solidification simulations https://github.com/pycalphad/pycalphad/pull/361.

Building phase records can give a significant performance boost, roughly 10x to 100x speedup in some cases.

• Add pyproject.toml
• Switch to setuptools_scm instead of versioneer
• Update installation instructions in README
• Add automated deployment on Releases
• Update docs theme to furo

Fixes a regression introduced in #11, where filter_phases was introduced to remove phases that cannot be stable or phases where the disordered part is given as a candidate phase. The change in #11 made it so the disordered phase was not a part of the list of solid_phases that could form during the solidification simulations. With the disordered phase name not included in the solid_phases, the order_disorder_eq_phases function that renamed the ordered phases with disordered configurations to their disordered counterpart would lead to the phase amount of disordered configurations to not be saved.

This PR fixes that and adds a parameterized test case that tests that the disordered phase always is accounted for regardless of whether or not the ordered phase is in the candidate phases or not.

• Exit the equilibrium solidification while loop after a binary search, because the solidification must be converged then by definition (fixes a floating point precision issue where 0.999... < 1 caused the loop to be infinite on some platforms.
• Update CI for pip version of pycalphad
• Remove environment.yml development file
• Drop Python 3.6

Currently, sublattices in ordered (partitioned) phases are considered equivalent by symmetry if their site ratios are equal. This is a poor heuristic and will fail to determine that sublattices 0 and 1 are equivalent in an FCC_L12 phase with sublattice model [['A', 'B'], ['A', 'B'], ['VA']] with site ratios [0.25, 0.75, 3].

Relevant for this fix will be the current heuristic in pycalphad models https://github.com/pycalphad/pycalphad/pull/311, and the changes that will occur as a result of fixing https://github.com/pycalphad/pycalphad/issues/345.