jaxvacua.flux_utils.is_physical

Contents

jaxvacua.flux_utils.is_physical#

is_physical(model, sampler, x, moduli_max=None, verbose=False)#

Decide whether the converged candidate point x lies in the physical region of moduli space.

Details

Five checks are applied in increasing cost order. Hyperplane failure rejects immediately, but hyperplane success is not treated as conclusive: Kähler-metric positivity is still checked.

#

Check

Cost

Rejected iff

0

Runaway bound

O(1)

max(|moduli|, |tau|) > moduli_max (if provided)

1

Dilaton floor

O(1)

Im(tau) <= s_lower

2

Kähler-cone hyperplanes (if available; early-return)

O(h12²)

hp @ Im(z) not all > 0

3

Kähler metric positivity (fallback; early-return)

O(h12³)

any eig(K) <= 0

4

Basic Im(z_i) > 0

O(h12)

any Im(z_i) <= 0

If model.lcs_tree.hyperplanes is set, check 2 provides the exact linear cone test. Check 3 is still tried afterwards; if it raises, the function falls through to check 4 as a last-resort sanity check.

Parameters:

  • model (Any) – Finder instance providing _convert_real_to_complex(), lcs_tree and (for check 3) kahler_metric().

  • sampler (Any) – Sampler instance providing the s_lower attribute (dilaton floor). Passed explicitly so the helper is decoupled from any particular finder configuration.

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

  • moduli_max (float) – If not None, reject points with max(|moduli|, |tau|) > moduli_max (runaway check). Defaults to None (no runaway check).

  • verbose (bool) – If True, print which check rejected the point (useful for debugging “why was my candidate filtered?”). Defaults to False.

Returns:

  • bool – True if x is in the physical region by all applicable

  • checks, ``False`` otherwise.

Return type:

bool

Contents