We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Number Theory

Title: On the signs of Fourier coefficients of Hilbert cusp forms

Authors: Ritwik Pal
Abstract: We prove that given any $\epsilon > 0$ and a primitive adelic Hilbert cusp form $f$ of weight $k=(k_1,k_2,...,k_n) \in (2 \mathbb{Z})^n$ and full level, there exists an integral ideal $\mathfrak{m}$ with $N(\mathfrak{m}) \ll_{\epsilon} Q_{f}^{9/20+ \epsilon} $ such that the $\mathfrak{m}$-th Fourier coefficient of $C_{f} (\mathfrak{m})$ of $f$ is negative. Here $n$ is the degree of the associated number field, $N(\mathfrak{m})$ is the norm of integral ideal $\mathfrak{m}$ and $Q_{f}$ is the analytic conductor of $f$. In the case of arbitrary weights, we show that there is an integral ideal $\mathfrak{m}$ with $N(\mathfrak{m}) \ll_{\epsilon} Q_{f}^{1/2 + \epsilon}$ such that $C_{f}(\mathfrak{m}) <0$. We also prove that when $k=(k_1,k_2,...,k_n) \in (2 \mathbb{Z})^n$, asymptotically half of the Fourier coefficients are positive while half are negative.
Subjects: Number Theory (math.NT)
Journal reference: Ramanujan J (2020)
DOI: 10.1007/s11139-019-00206-4
Cite as: arXiv:2002.00919 [math.NT]
  (or arXiv:2002.00919v1 [math.NT] for this version)

Submission history

From: Ritwik Pal [view email]
[v1] Mon, 3 Feb 2020 17:59:41 GMT (11kb)

Link back to: arXiv, form interface, contact.