<p>This article analyzes the asymptotic behavior of three classes of second-order inertial systems with vanishing damping when applied to non-potential operators. We consider the dynamics <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\text {(V-AVD)}_\alpha \)</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mtext>(V-AVD)</mtext> <mi>α</mi> </msub> </math></EquationSource> </InlineEquation>, <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\text {(V-DIN-AVD)}_{\alpha ,\beta }\)</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mtext>(V-DIN-AVD)</mtext> <mrow> <mi>α</mi> <mo>,</mo> <mi>β</mi> </mrow> </msub> </math></EquationSource> </InlineEquation>, and <InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(\text {(V-ISIHD)}_{\alpha ,\beta }\)</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mtext>(V-ISIHD)</mtext> <mrow> <mi>α</mi> <mo>,</mo> <mi>β</mi> </mrow> </msub> </math></EquationSource> </InlineEquation>, where <InlineEquation ID="IEq4"> <EquationSource Format="TEX">\(V: \mathcal {H} \rightarrow \mathcal {H}\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>V</mi> <mo>:</mo> <mi mathvariant="script">H</mi> <mo stretchy="false">→</mo> <mi mathvariant="script">H</mi> </mrow> </math></EquationSource> </InlineEquation> is a <InlineEquation ID="IEq5"> <EquationSource Format="TEX">\(\lambda \)</EquationSource> <EquationSource Format="MATHML"><math> <mi>λ</mi> </math></EquationSource> </InlineEquation>-cocoercive operator. By using a time-scale change <InlineEquation ID="IEq6"> <EquationSource Format="TEX">\(t = \sqrt{2(\alpha +1)s}\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>t</mi> <mo>=</mo> <msqrt> <mrow> <mn>2</mn> <mo stretchy="false">(</mo> <mi>α</mi> <mo>+</mo> <mn>1</mn> <mo stretchy="false">)</mo> <mi>s</mi> </mrow> </msqrt> </mrow> </math></EquationSource> </InlineEquation>, we show that as the damping parameter <InlineEquation ID="IEq7"> <EquationSource Format="TEX">\(\alpha \rightarrow +\infty \)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>α</mi> <mo stretchy="false">→</mo> <mo>+</mo> <mi>∞</mi> </mrow> </math></EquationSource> </InlineEquation>, the rescaled trajectories converge uniformly on compact time intervals to solutions of first-order differential equations. Specifically, we establish convergence to the monotone flow <InlineEquation ID="IEq8"> <EquationSource Format="TEX">\(\dot{y}(s) + V(y(s)) = 0\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mover accent="true"> <mi>y</mi> <mo>˙</mo> </mover> <mrow> <mo stretchy="false">(</mo> <mi>s</mi> <mo stretchy="false">)</mo> </mrow> <mo>+</mo> <mi>V</mi> <mrow> <mo stretchy="false">(</mo> <mi>y</mi> <mrow> <mo stretchy="false">(</mo> <mi>s</mi> <mo stretchy="false">)</mo> </mrow> <mo stretchy="false">)</mo> </mrow> <mo>=</mo> <mn>0</mn> </mrow> </math></EquationSource> </InlineEquation> for (V-AVD), to a Levenberg-Marquardt type equation <InlineEquation ID="IEq9"> <EquationSource Format="TEX">\(\dot{y}(s) + V(y(s)) + \beta _0 \frac{d}{ds}V(y(s)) = 0\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mover accent="true"> <mi>y</mi> <mo>˙</mo> </mover> <mrow> <mo stretchy="false">(</mo> <mi>s</mi> <mo stretchy="false">)</mo> </mrow> <mo>+</mo> <mi>V</mi> <mrow> <mo stretchy="false">(</mo> <mi>y</mi> <mrow> <mo stretchy="false">(</mo> <mi>s</mi> <mo stretchy="false">)</mo> </mrow> <mo stretchy="false">)</mo> </mrow> <mo>+</mo> <msub> <mi>β</mi> <mn>0</mn> </msub> <mfrac> <mi>d</mi> <mrow> <mi mathvariant="italic">ds</mi> </mrow> </mfrac> <mi>V</mi> <mrow> <mo stretchy="false">(</mo> <mi>y</mi> <mrow> <mo stretchy="false">(</mo> <mi>s</mi> <mo stretchy="false">)</mo> </mrow> <mo stretchy="false">)</mo> </mrow> <mo>=</mo> <mn>0</mn> </mrow> </math></EquationSource> </InlineEquation> for (V-DIN-AVD) in finite dimensions under the condition <InlineEquation ID="IEq10"> <EquationSource Format="TEX">\(\lambda &gt; 2\beta _0\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>λ</mi> <mo>&gt;</mo> <mn>2</mn> <msub> <mi>β</mi> <mn>0</mn> </msub> </mrow> </math></EquationSource> </InlineEquation>, and to an implicit dynamics <InlineEquation ID="IEq11"> <EquationSource Format="TEX">\(\dot{y}(s) + V(y(s) + \beta _0\dot{y}(s)) = 0\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mover accent="true"> <mi>y</mi> <mo>˙</mo> </mover> <mrow> <mo stretchy="false">(</mo> <mi>s</mi> <mo stretchy="false">)</mo> </mrow> <mo>+</mo> <mi>V</mi> <mrow> <mo stretchy="false">(</mo> <mi>y</mi> <mrow> <mo stretchy="false">(</mo> <mi>s</mi> <mo stretchy="false">)</mo> </mrow> <mo>+</mo> <msub> <mi>β</mi> <mn>0</mn> </msub> <mover accent="true"> <mi>y</mi> <mo>˙</mo> </mover> <mrow> <mo stretchy="false">(</mo> <mi>s</mi> <mo stretchy="false">)</mo> </mrow> <mo stretchy="false">)</mo> </mrow> <mo>=</mo> <mn>0</mn> </mrow> </math></EquationSource> </InlineEquation> for (V-ISIHD) when <InlineEquation ID="IEq12"> <EquationSource Format="TEX">\(x_1 = 0\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mi>x</mi> <mn>1</mn> </msub> <mo>=</mo> <mn>0</mn> </mrow> </math></EquationSource> </InlineEquation> and <InlineEquation ID="IEq13"> <EquationSource Format="TEX">\(2\beta _0 &lt; \lambda \)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>2</mn> <msub> <mi>β</mi> <mn>0</mn> </msub> <mo>&lt;</mo> <mi>λ</mi> </mrow> </math></EquationSource> </InlineEquation>. The proofs rely on Lyapunov-type energy estimates and provide a unified framework for understanding the high-damping limit of inertial systems in a non-potential setting. These results show that the non-potential setting cannot be treated as a mere extension of the gradient case <InlineEquation ID="IEq14"> <EquationSource Format="TEX">\(V=\nabla f\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>V</mi> <mo>=</mo> <mi mathvariant="normal">∇</mi> <mi>f</mi> </mrow> </math></EquationSource> </InlineEquation>, but requires fundamentally new Lyapunov constructions and stability arguments beyond those available in the potential framework.</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

Singular Perturbation Analysis of Accelerated Inertial Systems for Non-Potential Operators in the High-Damping Regime

  • Samir Adly,
  • Kouegnon D. Mitchozounnou,
  • Olivier Prot

摘要

This article analyzes the asymptotic behavior of three classes of second-order inertial systems with vanishing damping when applied to non-potential operators. We consider the dynamics \(\text {(V-AVD)}_\alpha \) (V-AVD) α , \(\text {(V-DIN-AVD)}_{\alpha ,\beta }\) (V-DIN-AVD) α , β , and \(\text {(V-ISIHD)}_{\alpha ,\beta }\) (V-ISIHD) α , β , where \(V: \mathcal {H} \rightarrow \mathcal {H}\) V : H H is a \(\lambda \) λ -cocoercive operator. By using a time-scale change \(t = \sqrt{2(\alpha +1)s}\) t = 2 ( α + 1 ) s , we show that as the damping parameter \(\alpha \rightarrow +\infty \) α + , the rescaled trajectories converge uniformly on compact time intervals to solutions of first-order differential equations. Specifically, we establish convergence to the monotone flow \(\dot{y}(s) + V(y(s)) = 0\) y ˙ ( s ) + V ( y ( s ) ) = 0 for (V-AVD), to a Levenberg-Marquardt type equation \(\dot{y}(s) + V(y(s)) + \beta _0 \frac{d}{ds}V(y(s)) = 0\) y ˙ ( s ) + V ( y ( s ) ) + β 0 d ds V ( y ( s ) ) = 0 for (V-DIN-AVD) in finite dimensions under the condition \(\lambda > 2\beta _0\) λ > 2 β 0 , and to an implicit dynamics \(\dot{y}(s) + V(y(s) + \beta _0\dot{y}(s)) = 0\) y ˙ ( s ) + V ( y ( s ) + β 0 y ˙ ( s ) ) = 0 for (V-ISIHD) when \(x_1 = 0\) x 1 = 0 and \(2\beta _0 < \lambda \) 2 β 0 < λ . The proofs rely on Lyapunov-type energy estimates and provide a unified framework for understanding the high-damping limit of inertial systems in a non-potential setting. These results show that the non-potential setting cannot be treated as a mere extension of the gradient case \(V=\nabla f\) V = f , but requires fundamentally new Lyapunov constructions and stability arguments beyond those available in the potential framework.