<p>In this paper, we consider functionals of the form <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(H_\alpha (u)=F(u)+\alpha G(u)\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mi>H</mi> <mi>α</mi> </msub> <mrow> <mo stretchy="false">(</mo> <mi>u</mi> <mo stretchy="false">)</mo> </mrow> <mo>=</mo> <mi>F</mi> <mrow> <mo stretchy="false">(</mo> <mi>u</mi> <mo stretchy="false">)</mo> </mrow> <mo>+</mo> <mi>α</mi> <mi>G</mi> <mrow> <mo stretchy="false">(</mo> <mi>u</mi> <mo stretchy="false">)</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation> with <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\alpha \in [0,+\infty )\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>α</mi> <mo>∈</mo> <mo stretchy="false">[</mo> <mn>0</mn> <mo>,</mo> <mo>+</mo> <mi>∞</mi> <mo stretchy="false">)</mo> </mrow> </math></EquationSource> </InlineEquation>, where <i>u</i> varies in a set <InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(U\ne \emptyset \)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>U</mi> <mo>≠</mo> <mi mathvariant="normal">∅</mi> </mrow> </math></EquationSource> </InlineEquation> (without further structure). We first show that, excluding at most countably many values of <InlineEquation ID="IEq4"> <EquationSource Format="TEX">\(\alpha \)</EquationSource> <EquationSource Format="MATHML"><math> <mi>α</mi> </math></EquationSource> </InlineEquation>, we have <InlineEquation ID="IEq5"> <EquationSource Format="TEX">\(\inf _{H_\alpha ^\star }G= \sup _{H_\alpha ^\star }G\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mo movablelimits="true">inf</mo> <msubsup> <mi>H</mi> <mi>α</mi> <mo>⋆</mo> </msubsup> </msub> <mi>G</mi> <mo>=</mo> <msub> <mo movablelimits="true">sup</mo> <msubsup> <mi>H</mi> <mi>α</mi> <mo>⋆</mo> </msubsup> </msub> <mi>G</mi> </mrow> </math></EquationSource> </InlineEquation>, where <InlineEquation ID="IEq6"> <EquationSource Format="TEX">\(H_\alpha ^\star {:}{=}\arg \min _UH_\alpha \)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msubsup> <mi>H</mi> <mi>α</mi> <mo>⋆</mo> </msubsup> <mo>:</mo> <mo>=</mo> <mo>arg</mo> <msub> <mo movablelimits="true">min</mo> <mi>U</mi> </msub> <msub> <mi>H</mi> <mi>α</mi> </msub> </mrow> </math></EquationSource> </InlineEquation>, which is assumed to be non-empty. Then, we prove a stronger result that concerns the invariance of the limiting value of the functional <i>G</i> along minimizing sequences for <InlineEquation ID="IEq7"> <EquationSource Format="TEX">\(H_\alpha \)</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>H</mi> <mi>α</mi> </msub> </math></EquationSource> </InlineEquation>, which extends the above Principle to the case <InlineEquation ID="IEq8"> <EquationSource Format="TEX">\(H_\alpha ^\star = \emptyset \)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msubsup> <mi>H</mi> <mi>α</mi> <mo>⋆</mo> </msubsup> <mo>=</mo> <mi mathvariant="normal">∅</mi> </mrow> </math></EquationSource> </InlineEquation>. This fact implies an unexpected consequence for functionals regularized with uniformly convex norms: excluding again at most countably many values of <InlineEquation ID="IEq9"> <EquationSource Format="TEX">\(\alpha \)</EquationSource> <EquationSource Format="MATHML"><math> <mi>α</mi> </math></EquationSource> </InlineEquation>, it turns out that for a minimizing sequence, convergence to a minimizer in the weak or strong sense is equivalent. Finally, we show to what extent these findings generalize to multi-regularized functionals and—in the presence of an underlying differentiable structure—to critical points.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Trade-off Invariance Principle for Minimizers of Regularized Functionals

  • Massimo Fornasier,
  • Jona Klemenc,
  • Alessandro Scagliotti

摘要

In this paper, we consider functionals of the form \(H_\alpha (u)=F(u)+\alpha G(u)\) H α ( u ) = F ( u ) + α G ( u ) with \(\alpha \in [0,+\infty )\) α [ 0 , + ) , where u varies in a set \(U\ne \emptyset \) U (without further structure). We first show that, excluding at most countably many values of \(\alpha \) α , we have \(\inf _{H_\alpha ^\star }G= \sup _{H_\alpha ^\star }G\) inf H α G = sup H α G , where \(H_\alpha ^\star {:}{=}\arg \min _UH_\alpha \) H α : = arg min U H α , which is assumed to be non-empty. Then, we prove a stronger result that concerns the invariance of the limiting value of the functional G along minimizing sequences for \(H_\alpha \) H α , which extends the above Principle to the case \(H_\alpha ^\star = \emptyset \) H α = . This fact implies an unexpected consequence for functionals regularized with uniformly convex norms: excluding again at most countably many values of \(\alpha \) α , it turns out that for a minimizing sequence, convergence to a minimizer in the weak or strong sense is equivalent. Finally, we show to what extent these findings generalize to multi-regularized functionals and—in the presence of an underlying differentiable structure—to critical points.