Optimization ( IF 1.6 ) Pub Date : 2020-11-18 , DOI: 10.1080/02331934.2020.1822836 Jean-Philippe Chancelier 1 , Michel De Lara 1
The so-called pseudonorm on counts the number of nonzero components of a vector. For exact sparse optimization problems – with the pseudonorm standing either as criterion or in the constraints – the Fenchel conjugacy fails to provide relevant analysis. In this paper, we display a class of conjugacies that are suitable for the pseudonorm. For this purpose, we suppose given a (source) norm on . With this norm, we define, on the one hand, a sequence of so-called coordinate-k norms and, on the other hand, a coupling between and itself, called Capra (constant along primal rays). Then, we provide formulas for the Capra-conjugate and biconjugate, and for the Capra subdifferentials, of functions of the pseudonorm, in terms of the coordinate-k norms. As an application, we provide a new family of lower bounds for the pseudonorm, as a fraction between two norms, the denominator being any norm.
中文翻译:
沿原始射线共轭和 l0 伪范数的常数
所谓的 伪范数计算向量的非零分量的数量。对于精确的稀疏优化问题——使用 伪范式无论是作为标准还是在约束中 - Fenchel 共轭未能提供相关分析。在本文中,我们展示了一类适用于 伪规范。为此,我们假设给定一个(源)规范. 有了这个范数,我们一方面定义了一系列所谓的坐标k范数,另一方面,定义了和它本身,称为 Capra(沿原始射线恒定)。然后,我们提供了 Capra 共轭和双共轭以及 Capra 次微分的公式 伪范数,就坐标k范数而言。作为一个应用程序,我们为 伪范数,作为两个范数之间的分数,分母是任何范数。