jaxvacua.flux_utils.classify_solution

jaxvacua.flux_utils.classify_solution#

classify_solution(model, x, flux, noscale=True, min_tol=1e-06)#

Classify a converged critical-point candidate (x, flux).

Details

Computes the scalar potential \(V\), the F-term norm \(|DW|=\sum_i|D_iW|\), and the Hessian eigenvalues \(\mathrm{eig}(\partial^2 V)\), then labels the point as

  • SUSY iff \(|DW|<10^{-6}\) (fixed threshold);

  • minimum iff every Hessian eigenvalue exceeds min_tol, otherwise saddle.

Eager-evaluation Python routine suitable for one-shot post-Newton classification; a JIT-vmapped batched variant lives on the finder for sample-time use.

Parameters:

  • model (Any) – Finder instance providing _convert_real_to_complex(), scalar_potential(), DW(), ddV_x() and tadpole(). Passed explicitly (rather than via implicit self) so the helper works with any FluxEFT subclass without inheritance assumptions.

  • x (Any) – Converged real-coord solution vector.

  • flux (Any) – Flux vector at the candidate point.

  • noscale (bool) – Pass-through to scalar_potential() and ddV_x(). Defaults to True.

  • min_tol (float) – Minimum-classification threshold. A point is labelled is_minimum=True iff every Hessian eigenvalue exceeds min_tol. Defaults to 1e-6.

Returns:

dict – Mapping with keys

  • 'V' (float): Scalar potential at the point.

  • '|DW|' (float): \(\sum_i |D_i W|\).

  • 'eigenvalues' (np.ndarray): Sorted real Hessian eigenvalues.

  • 'is_susy' (bool): |DW| < 1e-6.

  • 'is_minimum' (bool): All eigenvalues > min_tol.

  • 'Nflux' (float): |tadpole(flux)|.

Return type:

dict