SearcharxivSearch

arXiv subjects

Jennifer Johnson-Leung

Publications and source records attributed to Jennifer Johnson-Leung.

13 recordsLinked to original sources

The quaternionic Maass Spezialschar on split $\mathrm{SO}(8)$

The classical Maass Spezialschar is a Hecke-stable subspace of the space of holomorphic Siegel modular forms of genus two and level one cut out by certain linear relations among Fourier coefficients. We define an analogous quaternionic Maass Spezialschar, which consists of the quaternionic modular forms of level one on split $\mathrm{SO}(8)$ whose Fourier coefficients satisfy certain linear relations. We characterize this space in terms of a theta lift from the space of holomorphic Siegel modular forms on $\mathrm{Sp}(4)$, and in terms of periods. We also give a conjecture for the Dirichlet series of the standard $L$-function of quaternionic modular eigenforms on $\mathrm{SO}(8)$ and verify our conjecture on the quaternionic Maass Spezialschar.

math.NT

Post-Hoc Understanding of Metaphor Processing in Decoder-Only Language Models via Conditional Scale Entropy

Metaphor requires a language model to resolve a token whose contextual meaning diverges from its basic literal sense. Understanding how transformer models organize this reinterpretation across depth remains an open problem in mechanistic interpretability. We introduce conditional scale entropy (CSE), a wavelet-derived measure of how broadly transformer computation engages across frequency scales at each layer position. Two theorems establish that CSE is invariant to update magnitude, isolating the structural pattern of updates from their intensity. Using CSE, we find that metaphorical tokens produce significantly higher spectral breadth than literal tokens at contiguous layer positions on every decoder-only architecture tested, from 124M to 20B parameters (GPT-2 family, LLaMA-2 7B, GPT-oss 20B). The effect survives cluster-based permutation correction, recurs in the early-to-mid relative depth range across models, and converges with an independent analysis of 200 naturalistic VUA pairs. Specificity controls further show that the effect is not explained by semantic complexity or by matched propositional content. These results identify multi-scale coordination as a consistent signature of metaphorical language processing in the decoder-only architectures examined, and establish CSE as a principled tool for characterizing cross-depth structure in transformers.

cs.CL

An Explicit Theta Lift to Siegel Paramodular Forms

Let $E/L$ be a real quadratic extension of number fields. We construct an explicit map from an irreducible cuspidal automorphic representation of $\mathrm{GL}(2,E)$ which contains a Hilbert modular form with $Γ_0$ level to an irreducible automorphic representation of $\mathrm{GSp}(4,L)$ which contains a Siegel paramodular form and exhibit local data which produces a paramodular invariant vector for the local theta lift at every finite place, except when the local extension has wild ramification.

math.NT

The paramodular Hecke algebra

We give a presentation via generators and relations of the local graded paramodular Hecke algebra of prime level. In particular, we prove that the paramodular Hecke algebra is isomorphic to the quotient of the free $\mathbb{Z}$-algebra generated by four non-commuting variables by an ideal generated by seven relations. Using this description, we derive rationality results at the level of characters and give a characterization of the center of the Hecke algebra. Underlying our results are explicit formulas for the product of any generator with any double coset.

math.NT

Stable Klingen Vectors and Paramodular Newforms

We introduce the family of stable Klingen congruence subgroups of GSp(4). We use these subgroups to study both local paramodular vectors and Siegel modular forms of degree $2$ with paramodular level. In the first part, when $F$ is a nonarchimedean local field of characteristic zero and $(π,V)$ is an irreducible, admissible representation of GSp(4,F) with trivial central character, we establish a basic connection between the subspaces $V_s(n)$ of $V$ fixed by the stable Klingen congruence subgroups and the spaces of paramodular vectors in $V$ and derive a fundamental partition of the set of paramodular representations into two classes. We determine the spaces $V_s(n)$ for all $(π,V)$ and $n$. We relate the stable Klingen vectors in $V$ to the two paramodular Hecke eigenvalues of $π$ by introducing two stable Klingen Hecke operators and one level lowering operator. In contrast to the paramodular case, these three new operators are given by simple upper block formulas. We prove further results about stable Klingen vectors in $V$ especially when $π$ is generic. In the second part we apply these local results to a Siegel modular newform $F$ of degree $2$ with paramodular level $N$ that is an eigenform of the two paramodular Hecke operators at all primes $p$. We present new formulas relating the Hecke eigenvalues of $F$ at $p$ to the Fourier coefficients $a(S)$ of $F$ for $p^2 \mid N$. We verify that these formulas hold for a large family of examples and indicate how to use our formulas to generally compute Hecke eigenvalues at $p$ from Fourier coefficients of $F$ for $p^2 \mid N$. Finally, for $p^2 \mid N$ we express the formal power series in $p^{-s}$ with coefficients given by the radial Fourier coefficients $a(p^t S)$, $t\geq 0$, as an explicit rational function in $p^{-s}$ with denominator $L_p(s,F)^{-1}$, where $L_p(s,F)$ is the spin $L$-factor of $F$ at $p$.

math.NT

Twisting of Siegel paramodular forms

Let $S_k(Γ^{\mathrm{para}}(N))$ be the space of Siegel paramodular forms of level $N$ and weight $k$. Let $p\nmid N$ and let $χ$ be a nontrivial quadratic Dirichlet character mod $p$. Based on our previous work, we define a linear twisting map $\mathcal{T}_χ:S_k(Γ^{\mathrm{para}}(N))\rightarrow S_k(Γ^{\mathrm{para}}(Np^4))$. We calculate an explicit expression for this twist and give the commutation relations of this map with the Hecke operators and Atkin-Lehner involution for primes $\ell\neq p$.

math.NT

Fourier coefficients for twists of Siegel paramodular forms (expanded version)

In this paper, we calculate the Fourier coefficients of the paramodular twist of a Siegel modular form of paramodular level $N$ by a nontrivial quadratic Dirichlet character mod $p$ for $p$ a prime not dividing $N$. As an application, these formulas can be used to verify the nonvanishing of the twist for particular examples. We also deduce that the twists of Maass forms are identically zero.

math.NT

Twisting of paramodular vectors

Let $F$ be a non-archimedean local field of characteristic zero, let $(π,V)$ be an irreducible, admissible representation of $\GSp(4,F)$ with trivial central character, and let $χ$ be a quadratic character of $F^\times$ with conductor $c(χ)>1$. We define a twisting operator $T_χ$ from paramodular vectors for $π$ of level $n$ to paramodular vectors for $χ\otimes π$ of level $\max(n+2c(χ),4c(χ))$, and prove that this operator has properties analogous to the well-known $\GL(2)$ twisting operator.

math.NT

Siegel modular forms of degree two attached to Hilbert modular forms

Let E/Q be a real quadratic field and pi_0 a cuspidal, irreducible, automorphic representation of GL(2,A_E) with trivial central character and infinity type (2,2n+2) for some non-negative integer n. We show that there exists a non-zero Siegel paramodular newform F with weight, level, Hecke eigenvalues, epsilon factor and L-function determined explicitly by pi_0. We tabulate these invariants in terms of those of pi_0 for every rational prime p.

math.NT

Igusa class polynomials, embeddings of quartic CM fields, and arithmetic intersection theory

Bruinier and Yang conjectured a formula for an intersection number on the arithmetic Hilbert modular surface, CM(K).T_m, where CM(K) is the zero-cycle of points corresponding to abelian surfaces with CM by a primitive quartic CM field K, and T_m is the Hirzebruch-Zagier divisors parameterizing products of elliptic curves with an m-isogeny between them. In this paper, we examine fields not covered by Yang's proof of the conjecture. We give numerical evidence to support the conjecture and point to some interesting anomalies. We compare the conjecture to both the denominators of Igusa class polynomials and the number of solutions to the embedding problem stated by Goren and Lauter.

math.NT

On the equivariant and the non-equivariant main conjecture for imaginary quadratic fields

The Iwasawa main conjecture fields has been an important tool to study the arithmetic of special values of $L$-functions of Hecke characters of imaginary quadratic fields. To obtain the finest possible invariants it is important to know the main conjecture for all prime numbers $p$ and also to have an equivariant version at disposal. In this paper we first prove the main conjecture for imaginary quadratic fields for all prime numbers $p$, improving earlier results by Rubin. From this we deduce the equivariant main conjecture in the case that a certain $μ$-invariant vanishes. For prime numbers $p\nmid 6$ which split in $K$, this is a theorem by a result of Gillard.

math.NT