Skip to content

VMEX 0.5.0

Choose a tag to compare

View all tags
@rogeriojorge rogeriojorge released this 11 Aug 19:13
· 854 commits to main since this release
v0.5.0
5d6ceff
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

VMEX 0.5.0 makes converged equilibria optimizer-neutral building blocks.

  • Adds consistent SciPy and JAX residual, Jacobian, scalar, and value/gradient contracts for QI, QS, geometry, profile, stability, and user-authored objectives.
  • Makes exact implicit derivatives, automatic Jacobian direction, hot and perturbation restarts, ESS scales, bounded Jacobian batches, and smooth failed-trial handling production defaults.
  • Exposes forward convergence and derivative-certification controls, including FSQ diagnostics.
  • Adds shared Boozer evaluation for J-invariant QI and maximum-J objectives while retaining free-boundary, virtual-casing, bootstrap, stability, plotting, CLI, and standard wout workflows.
  • Makes automatic worker counts respect process CPU limits on shared, containerized, and scheduler-managed machines.

Upgrade note: tuple weights in VmecProblem.from_tuples use cost semantics. The compatibility optimize.least_squares wrapper retains its documented residual-weight convention.

See the full changelog.

Assets 2
Loading