AI & ML
Superspecial curves of genus two over prime fields: Frobenius types and rational (2,2)-kernels
arXiv cs.CR0 views
arXiv:2611.23694v1 Announce Type: cross Abstract: Superspecial curves of genus two over a prime field $\mathbb{F}_p$, and the rational $(2,2)$-isogenies between their Jacobians, are the vertices and edges of the graph on which genus-two isogeny-based cryptography is built. Let $C$ be such a curve, with Jacobian $J$ and Frobenius endomorphism $\pi$. We show that $\pi^2 = p\,\alpha_*$ for a unique automorphism $\alpha$ of $C$ over $\mathbb{F}_p$ whose action on the Weierstrass points is the square of that of Frobenius. For $p \ge 7$ this gives the list of Frobenius types of the Weierstrass points: eight types, each with its Weil polynomial, a congruence condition on $p$, and its number 15, 3, 3, 7, 1, 1, 3 or 1 of $\mathbb{F}_p$-rational $(2,2)$-kernels; when all Weierstrass points are rational, the 2-Tate pairing on $J[2]$ is determined by $p \bmod 8$. We then prove closed formulas, in terms of class numbers of imaginary quadratic orders and of $\zeta_{\mathbb{Q}(\sqrt p)}(-1)$, for the mass of the curves of each type and for the mass of the rational kernels whose quotient is a product of elliptic curves. For $\pi^2 = \pm p$ the total masses are due to Ibukiyama and to Xue and Yu; our proofs use the 2-adic Tate modules, the symmetry of the Richelot correspondence and Eichler's mass formula, and, for the other curves, modules over orders of a quartic CM field. The kernels are counted through gluings of supersingular elliptic curves. As a consequence, the proportion of rational kernels with decomposed quotient is of order $p^{-1/2}(1+L(1,\chi))$, where $\chi$ is the quadratic character of $\mathbb{Q}(\sqrt{-p})$. All formulas agree with an exact enumeration for the 37 primes $7 \le p \le 173$. Finally we describe the graph of $\mathbb{F}_p$-rational Richelot isogenies, which is far from connected in the range computed, and discuss what these results say about its use in isogeny-based cryptography.
Source:https://arxiv.org/abs/2611.23694PreviousA Survey on RLOps for Large Language ModelsNextDetectViT: Test-time Backdoor Detection for Vision Transformers via Inter-Head Attention Discrepancy
Related posts
Razorpay webhook not verified: anyone can fake a payment
0dev.to
From FAQPage Schema to Answerable AI Documentation
0dev.to
Wiring Android's MediaPipe LLM Inference API to a Multi-Turn Conversation Engine
0dev.to
What Is LLM Throughput? Tokens per Second, Concurrency, and Batching
0dev.to
Diese Woche haben mir zwei Tools etwas vorgeschlagen und es direkt abgeschickt.
0dev.to
AI Agent Observability: What to Log Before Production
0dev.to