[StuartBenjamin]

Problem. STRIDE declares nzero but never sets it, so it gives no fixed-boundary (internal) ideal stability verdict: "Fixed-boundary mode unstable" is never printed and stride.nc has no nzero. Its free-boundary energy does not catch internal modes either. On a DIII-D shot with q_min ≈ 1.02, STRIDE's total1 stays positive on runs that RDCON finds fixed-boundary unstable at n = 1. Testing only at the interval boundaries would be cheap but inexact, because two crossings inside one interval cancel.

Fix. A new switch, calc_nzero in stride_params (default f), turns the count on. It needs asymp_at_sing = t and no Riccati integration. With it off, STRIDE is bitwise the same as before.

How the count works:

• During the parallel pass: each interval counts the zero crossings of its own principal solution at every ODE step. This uses the inertia of U₂ᴴU₁, from an LDLᴴ factorization; crit is evaluated only on the few steps where the inertia changes.
• After the stitch: ode_zero_crossings adds the crossings of the axis solution beyond those, from each interval's end data alone. These are the negative eigenvalues of dir·X_sᴴP₁₂⁻¹X_e, a Morse-index split that holds for F̄ > 0.
• Next to a singular surface: the intervals there start from the asymptotic solutions, so their propagator is too ill-conditioned for that split. There the axis solution itself is integrated and counted at every step.
• At the singular surfaces: they are crossed as in RDCON's ode_ideal_cross.

nzero is written to stride.nc and drives the existing message. Δ′ and the energies are bitwise unchanged.

Test, with calc_nzero = t (vs develop)	Result
docs/examples DIII-D and Solovev resistive, n = 1–4 (8 runs)	all STRIDE outputs bitwise; nzero 0, as RDCON; wall time +0 to +41 %
DIII-D shot 174956 with q_min ≈ 1.02, 7 equilibria, n = 1–4 (26 runs)	nzero = RDCON on 25 runs; all other STRIDE outputs bitwise; wall time +16 %

The 26th run is a case where RDCON itself is unreliable: it reports 67 to 71 crossings at n = 2, starting exactly on the q = 3/2 surface. There STRIDE gives 0.

🤖 Generated with Claude Code