[dgaines2]

Summary of Changes

This adds a calculator-agnostic equation of state workflow in quacc.recipes.common.eos, modeled on atomate2's CommonEosMaker. The structure is volume-strained isotropically, each strained cell gets a fixed-cell relaxation and optionally a static, and the energy-volume data are fit with emmet's EOSDoc. That gives V0, E0, B0 and B1 for the Murnaghan, Birch, Birch-Murnaghan and Vinet models.

• eos_flow(atoms, relax_job, static_job, pre_relax=True, run_static=False, ...) optionally relaxes the bulk cell, then runs eos_subflow. It has the same shape as elastic_tensor_flow.
• eos_subflow(undeformed_result, relax_job=None, static_job=None, ...) starts from an existing result. Each strained structure is relaxed at fixed cell (relax_cell=False) if relax_job is given, then run with static_job if given. The fit uses the static energies when available.
• make_volume_strains_from_bulk in quacc.atoms.deformation scales the cell by (1 + s)^(1/3) with fractional coordinates fixed. The defaults are ±1.5% to ±7.5% volume in 10 steps.
• EOSSchema is a generic return type, so the deformed results keep the schema of the jobs passed in (e.g. VaspSchema).
• A ValueError is raised before any jobs launch if fewer than 4 points would be fit.

from ase.build import bulk

from quacc.recipes.common.eos import eos_flow
from quacc.recipes.emt.core import relax_job, static_job  # or any code's jobs

atoms = bulk("Cu")
result = eos_flow(atoms, relax_job, static_job)  # run_static=True recommended for VASP

fit = next(f for f in result["eos_doc"].eos if f.model.value == "birch_murnaghan")
print(f"V0 = {fit.V0:.3f} Å^3, B0 = {fit.B0:.1f} GPa")  # V0 = 11.566 Å^3, B0 = 134.3 GPa

Questions for maintainers

1. Strain convention and range: we use zero-centered volumetric strains (-0.06), the same as ASE's calculate_eos eps, not scale factors (0.94). Our range (±7.5% volume) is narrower than atomate2's (±5% linear, about ±15% volume). Any preference?
2. run_static default: it's False, matching elastic. For VASP, True is better because the relax's final energy is less accurate; for EMT and MLIPs the static repeats the last relax step.
3. Per-code wrappers: none are included. Elastic has EMT and MLIP wrappers, so it appears in each code's docs. Should I add EMT/MLIP (and VASP?) wrappers here or in a follow-up?
4. Generic EOSSchema: this is the first generic TypedDict in types.py. It follows Fix type annotations across recipes, runners, and schemas #3535's approach for common helpers like common/slabs.py. tests/typecheck/api.py covers it.

[buildbot-princeton]

Can one of the admins verify this patch?

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (88f951d) to head (9cef629).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main     #3552   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files          103       104    +1     
  Lines         4548      4596   +48     
=========================================
+ Hits          4548      4596   +48     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Andrew-S-Rosen]

Thanks, @dgaines2! I will try to review this weekend.