We study the dyon counting functions of a four-dimensional N=4 string model whose symmetry lies outside the Mathieu group M_24: the CHL model built from the quantum Z_5 symmetry Q (Frame shape 1^-1 5^5, Conway class 5C) of the K3 model that Gaberdiel and Volpato obtain as an asymmetric Z_5 orbifold of Volpato's A_4-torus model, together with its non-cyclic Z_5 × Z_5 twisted-twining lifts. Reading these Siegel-type functions as (twined) quarter-BPS indices, normalised to leading coefficient 1, we run the black-hole consistency tests available at accessible charges, on exact coefficients and under pre-registration. Positivity: the single-centred index is strictly positive at all 237,378 charges tested for this model and for the geometric 5A control, without hair removal; this is large-scale evidence for the positivity conjecture beyond M_24, not a proof. Blind growth tests: on 29,047 fresh charges for the twined function and 19,749 fresh charge classes for the twining function, all predicted before the coefficients were computed, a coherent sum over the dominant zeros, with no fitted continuous parameter, gives the sign of every coefficient and its logarithm to about 0.02; this extends the coherent-saddle method of Banerjee-Jatkar-Sen and Chattopadhyaya-David to twined and twining lifts outside M_24 and, in such a model, makes Sen's discrete-information picture testable; the test confirms it through O(1). The first subleading term, 5/(π√Δ), is Banerjee-Jatkar-Sen's two-loop term rescaled by n_2 = 5 (an application of their result); in a later pre-registered test on fresh charges, with an independent recomputation, the residue series predicts the mean of ln|d| to 10^-5 (per-charge scatter 7·10^-4 for the twined function). Exact formulas: at every Fourier-Jacobi index m ≡ 2, 3 (mod 5) the coefficients are honest (not mock) Jacobi forms with no wall-crossing and no twisted polar part, and the single-centred index is given exactly by an I3/2 Rademacher series whose only inputs are twining-cusp polar coefficients (a theorem given the companion paper); procedurally blind tests reproduced 840 fresh coefficients at m = 2, 3 and 2,430 at the six honest m from 7 to 18, to at least 27.8 and 56 digits; this gives exact single-centred indices against which macroscopic entropy formulas can be compared. Half-BPS structure: we prove that the theta series of the rank-20 coinvariant lattice of the 5C Conway element is η^24 times an explicit combination of the half-BPS functions, and we confirm, on all tested charges, the mock-modular capture of single-centred indices by the finite parts of Fourier-Jacobi coefficients (Bhand-Sen-Singh, after Dabholkar-Murthy-Zagier) for this model and its twined lift. Walls, S-duality, hair and entropy: the non-cyclic lift has exactly the cyclic walls, with jumps that factorise into twined half-BPS indices; the indices are invariant under the expected S-duality shifts and under the Levi image of Γ_0(5); hair-removed positivity holds in both duality frames, unlike the Z_2 toroidal analogue; and the exact coefficients carry the parameter-free Fricke asymmetry predicted by the R^2 coupling, as do those of all 17 unbalanced Frame shapes (six of them in a disclosed, non-blind re-run after a coding error was corrected). All of this is low-charge consistency evidence for the dyon-index reading, not a proof of it; the absolute normalisation of the twisted index is stated as an open question.
No takes yet. Share an insight, caveat, or question.
Jason Bogovich (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: