[algorithmiq-release-bot]

Merging this PR releases v0.2.0.

It is updated automatically whenever a change lands on release/0.2.
Wording fixes to CHANGELOG.md can be pushed to this PR, but make them
just before merging: the next update overwrites them.

---

0.2.0 (2026-11-04)

Breaking changes

• entropy: Rank single-orbital entropies over the whole window by default (#112)

  Code that builds EntropyConfig without ranking now thresholds over
  the whole window.
  To keep the previous selection, pass ranking="subset":

  # before (implicit subset)
  EntropyConfig(fraction=0.1, fraction_type="r")
  # after, same selection as before
  EntropyConfig(fraction=0.1, fraction_type="r", ranking="subset")

• projection: Reject AO labels that match no projection AO (#113)

  A label that matches no AO used to give an empty selection and now
  raises ValueError.
  Fix the label using the shells listed in the message. For a 4d/5d metal,
  either give the
  molecule an ECP or project onto the molecular basis:

  # before: silently empty when Ru has no ECP
  ProjectionConfig(ao_labels=["Ru 4d"])
  # after: give Ru an ECP when building the molecule (as in the paper) ...
  mol = gto.M(atom=..., basis={"Ru": "cc-pvdz-pp", "default": "def2-svp"}, ecp={"Ru": "cc-pvdz-pp"})
  # ... or project onto the molecular basis
  ProjectionConfig(ao_labels=["Ru 4d"], basis=None)

• projection: Default select_by_ao_projection to a contribution of 0.5 (#114)

  Calls to select_by_ao_projection without contribution (and without
  top_n) now keep
  only orbitals with a weight of at least 0.5. To keep every orbital, as
  before, pass
  contribution=0.0:

  # before (implicit 0.0: every orbital kept)
  select_by_ao_projection(mf, ao_labels=["C 2pz"])
  # after, same result as before
  select_by_ao_projection(mf, ao_labels=["C 2pz"], contribution=0.0)

Fixes

• projection: Warn when per-label thresholds do not match the labels (#116)

Documentation

• parameters: How to choose the AEGISS parameters (#117)
• uhf: Define the UHF active space as two independent lists (#115)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 74.64%. Comparing base (d92fc9e) to head (37ad093).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@             Coverage Diff              @@
##           release/0.2     #134   +/-   ##
============================================
  Coverage        74.64%   74.64%           
============================================
  Files               36       36           
  Lines             2390     2390           
  Branches           351      351           
============================================
  Hits              1784     1784           
  Misses             567      567           
  Partials            39       39