On the existence of optimizers for nonlinear time-frequency concentration problems: the Born–Jordan distribution

Federico Stra, Erling A. T. Svela, S. Ivan Trapasso · 2026

Read On the existence of optimizers for nonlinear time-frequency concentration problems: the Born–Jordan distribution by Federico Stra, Erling A. T. Svela, S. Ivan Trapasso (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-10535.txt.

Introduction

Federico Stra and two colleagues settle when concentration optimizers exist for the Born-Jordan distribution. They study the Lp focus problem for the Born-Jordan distribution in dimension d above 1, building on a one-dimensional analysis. Their 2026 paper shows whether concentration optimizers exist depends on the exponent p. The key threshold is p-star of d, equal to 2d over d minus 2 for d at least 2.

For exponents below the key value, from 1 up to but not including p-star, the authors prove the top value is finite and is attained. For exponents above the key value, they show the function is unbounded. They also provide the complete solution in the much harder key regime in d equals 2. The Born-Jordan distribution, an average of broad Wigner distributions, is valued for reducing ghost tones in phase-space signal analysis.

Catalog record

CREATORFederico Stra, Erling A. T. Svela, S. Ivan Trapasso
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE70 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.10535

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On the existence of optimizers for nonlinear time-frequency concentration problems: the Born–Jordan distribution Federico Stra, Erling A. T. Svela, S. Ivan Trapasso Abstract We study the L p L^{p} concentration problem for the Born–Jordan distribution in dimension d > 1 d>1 , thus extending the one-dimensional analysis in [Stra–Svela–Trapasso, J. Math. Pures Appl. (2026)]. We show that the existence of concentration optimizers depends on the exponent p p with a critical threshold at p ∗ ​ ( d ) = 2 ​ d d − 2 p_{*}(d)=\frac{2d}{d-2} for d ≥ 2 d\geq 2 (with the understanding that p ∗ ​ ( 2 ) = ∞ p_{*}(2)=\infty ). In particular, for subcritical exponents 1 ≤ p < p ∗ ​ ( d ) 1\leq p<p_{*}(d) we prove that the supremum is finite and is attained, whereas for supercritical exponents p > p ∗ ​ ( d ) p>p_{*}(d) we show that the functional is unbounded. We also provide the complete solution in the (significantly more) challenging critical regime in dimension d = 2 d=2 . MSC: 81S30, 49Q10, 42B10, 94A12. Keywords: time-frequency analysis, Born–Jordan distribution, phase-space concentration, critical exponent. 1 Introduction The Wigner distribution, defined for f\in L^{2}(\mathbb{R}^{d}) by \displaystyle Wf(z) \displaystyle\coloneqq\int_{\mathbb{R}^{d}}e^{-2\pi i\xi\cdot y}f(x+y/2)\overline{f(x-y/2)}\mathop{}\!\mathrm{d}y, \displaystyle z \displaystyle=(x,\xi)\in\mathbb{R}^{2d}, is one of the most popular quadratic phase-space representations for signals in view of its many desirable properties — including being real-valued, continuous and energy preserving. However, as no time-frequency representation may possess every desirable property at once [ 10 ] , even the Wigner transform comes with some features that may represent challenging drawbacks in several applications. For instance, although it may be informally interpreted as a phase-space energy density of the signal f , Wf will in general take both positive and negative values. Moreover, the inherent sesquilinear structure of the Wigner transform produces peculiar ghost interferences . To be more precise, this means that the magnitude of the cross-term 2\Re W(f,g) in the expansion of W(f+g) may be significantly larger than that of the main terms Wf+Wg [ 1 ] . Such scattered interference patterns further obstruct the interpretation of Wf as a genuine energy density, hence several alternative phase space distributions have been introduced in order to tame or circumvent these phenomena. In this connection, the Cohen class [ 4 ] consists of distributions of the form Qf=Wf*\sigma , obtained by smoothing the Wigner transform with a suitable convolution kernel. A distinguished member of this family is the Born–Jordan distribution [ 8 , 9 ] , originally arising in the context of quantization problems in [ 3 ] and popularized as an average of certain generalized Wigner distributions that is effectively able to reduce the magnitude of the ghost frequencies [ 1 , 6 ] . Definition 1.1 . Let f\in L^{2}(\mathbb{R}^{d}) . For \tau\in(0,1) , define the \tau -Wigner distribution \displaystyle W_{\tau}f(z) \displaystyle\coloneqq\int_{\mathbb{R}^{d}}e^{-2\pi i\xi\cdot y}f\bigl(x+\tau y\bigr)\overline{f\bigl(x-(1-\tau)y\bigr)}\mathop{}\!\mathrm{d}y, \displaystyle z \displaystyle=(x,\xi)\in\mathbb{R}^{2d}. (1.1) The Born–Jordan distribution of f is defined by W_{\mathrm{BJ}}f(z)\coloneqq\int_{0}^{1}W_{\tau}f(z)\mathop{}\!\mathrm{d}\tau. (1.2) As a member of the Cohen family, the Born–Jordan distribution has a kernel, which was explicitly determined in [ 5 ] : W_{\mathrm{BJ}}f=Wf*\Theta_{\sigma},\qquad\Theta_{\sigma}(x,\xi)=\begin{cases}-2\mathrm{Ci}(4\pi|x\cdot\xi|)&(d=1)\\ \mathcal{F}(\bm{1}_{\{|s|\geq 2\}}|s|^{d-2})(x\cdot\xi)&(d>1),\end{cases} where \mathrm{Ci}(t)=-\int_{t}^{+\infty}\frac{\cos s}{s}\mathop{}\!\mathrm{d}s is the cosine integral function. This peculiar kernel is the root of the poor behavior of the Born–Jordan distribution regarding regularity and decay, since \Theta_{\sigma} is not globally integrable in any p -norm, and not even locally L^{\infty} [ 5 ] . In fact, many of the properties of the Born–Jordan distribution, such as continuity and boundedness, are only known to hold when d=1 due to the decay of the kernel degrading as d grows. In spite of these considerations, the empirical improvement in terms of interference reduction of the Born–Jordan distribution compared to the Wigner one has been recently confirmed in the context of phase-space concentration problems, a classical topic in time-frequency analysis. To be precise, in this note we study the maximal L^{p} concentration of the Born–Jordan distribution on a prescribed phase space subset \Omega\subset\mathbb{R}^{2d} with positive finite Lebesgue measure, and in particular whether the optimal concentration \sup_{\begin{subarray}{c}f\in L^{2}(\mathbb{R}^{d})\\ \lVert f\rVert_{L^{2}}^{2}=1\end{subarray}}\lVert W_{\mathrm{BJ}}f\rVert_{L^{p}(\Omega)}=\sup_{f\in L^{2}(\mathbb{R}^{d})\setminus\{0\}}\frac{\lVert W_{\mathrm{BJ}}f\rVert_{L^{p}(\Omega)}}{\lVert f\rVert_{L^{2}}^{2}} (1.3) is finite and attained. The existence of local concentration optimizers has been investigated only recently in [ 11 ] for the ambiguity transform and in [ 12 ] by the authors in the case of the Wigner transform in the framework of concentration compactness at the phase space level. As a byproduct, in that work we were able to address the Born–Jordan concentration problem in the 1 -dimensional case, as summarized below. Theorem 1.2 ( d=1 , [ 12 ] ) . Let \Omega\subset\mathbb{R}\times\mathbb{R} be a measurable set with 0<\mathscr{L}^{2}(\Omega)<\infty . Then for every p\in[1,\infty) the functional J_{\mathrm{BJ}}^{p}:L^{2}(\mathbb{R})\to[0,\infty):f\mapsto\lVert W_{\mathrm{BJ}}f\rVert_{L^{p}(\Omega)} is sequentially weakly continuous in L^{2}(\mathbb{R}) . As such, the supremum \sup_{f\in L^{2}(\mathbb{R})\setminus\{0\}}\frac{\lVert W_{\mathrm{BJ}}f\rVert_{L^{p}(\Omega)}}{\lVert f\rVert_{L^{2}(\mathbb{R})}^{2}}<\infty is attained by an optimizer. When p=\infty we have \sup_{f\in L^{2}(\mathbb{R})\setminus\{0\}}\frac{\lVert W_{\mathrm{BJ}}f\rVert_{L^{\infty}(\Omega)}}{\lVert f\rVert_{L^{2}(\mathbb{R})}^{2}}=\pi, but the supremum is not attained. These findings are particularly interesting as they show that, in spite of the challenging nature of the Born–Jordan kernel already discussed, the underlying interference reduction mechanism occurs already at the topological level — it is the first non-trivial example of a time-frequency concentration functional that is proved to be weakly continuous. The goal of the present article is to provide a fairly complete picture of the matter in higher dimensions, solving a problem raised at the end of [ 12 ] . 1.1 Main results We show that the answer to the L^{p} concentration problem for the Born–Jordan distribution changes drastically depending on the exponent p and we must therefore distinguish three regimes, where a crucial role is played by the following critical exponent . Definition 1.3 . For every positive integer d\in\mathbb{N}_{+} let p_{*}(d) be the critical exponent defined as \displaystyle p_{*}(1) \displaystyle=p_{*}(2)=\infty, \displaystyle p_{*}(d) \displaystyle=\left(\frac{1}{2}-\frac{1}{d}\right)^{-1}=\frac{2d}{d-2}\qquad\forall\ d\geq 3. (1.4) We stress that the critical exponent for d>2 coincides with the exponent 2^{*} of the Sobolev embedding W^{1,2}(\mathbb{R}^{d})\to L^{p_{*}(d)}(\mathbb{R}^{d}) , and this occurrence is not coincidental. Indeed, in (symplectic) Fourier variables the Born–Jordan distribution is obtained from the Wigner distribution by multiplication with \Theta(x,\xi)=\operatorname{sinc}(\pi\,x\cdot\xi) , and the asymptotic behavior \lvert\operatorname{sinc}(t)\rvert\sim|t|^{-1} as \lvert t\rvert\to+\infty suggests heuristically that the Cohen smoothing acts like an inverse first-order multiplier, so that W_{\mathrm{BJ}}f gains roughly one derivative compared with Wf . In the natural L^{2} -Sobolev scale this is modeled by the Riesz potential I=(-\Delta)^{-1/2} , which is bounded L^{2}(\mathbb{R}^{d})\to L^{p}(\mathbb{R}^{d}) precisely at the critical exponent p=p_{*}(d) for d>2 . An equivalent viewpoint that will serve as a main motivation for several constructions in our proofs stems from scale invariance: under the L^{2} -preserving scaling F_{\lambda}(x)=\lambda^{d/2}F(\lambda x) one has \|IF_{\lambda}\|_{L^{p}}=\lambda^{d/2-1-d/p}\|IF\|_{L^{p}} , so scale invariance occurs exactly at p_{*}(d) . In dimension d=2 the critical exponent escapes to p_{*}(2)=\infty , which is consistent with the fact that the only borderline case is the endpoint p=\infty . The three following theorems are our main results and they settle the concentration problem ( 1.3 ) in high dimension, with the exception of the critical case in dimension greater than 2 , which is left for future investigation. Theorem 1.4 (Subcritical case in d\geq 2 ) . Let d\geq 2 be an integer and \Omega\subset\mathbb{R}^{d}\times\mathbb{R}^{d} be a measurable set with 0<\mathscr{L}^{2d}(\Omega)<\infty . Then for every p\in\bigl[1,p_{*}(d)\bigr) the functional J_{\mathrm{BJ}}^{p}:L^{2}(\mathbb{R}^{d})\to[0,\infty):f\mapsto\lVert

Read the complete text