几何解释
通过函数值集合理解强弱对偶性
可以通过集合
给出对偶函数的简单几何解释。事实上,此集合是约束函数和目标函数所取得的函数值。利用集合
求解以
如果下确界有界,则不等式
定义了集合

TikZ 代码
\begin{tikzpicture}[line cap=round,line join=round,scale=0.95]
% G set and the supporting line lambda u + t = g(lambda)
\draw[->] (-3.4,0) -- (3.6,0) node[right] {$u$};
\draw[->] (0,-0.8) -- (0,2.6) node[above] {$t$};
\draw[fill=gray!20] plot[smooth cycle,tension=0.85]
coordinates {(-2.7,1.15) (-1.9,1.6) (-0.9,1.95) (0.3,2.0) (1.3,1.75) (2.0,1.3) (2.6,0.5) (1.2,0.6) (0,0.8) (-1.4,0.75)};
\node at (1.0,1.6) {$\mathcal{G}$};
\draw[dashed] (-2.0,0.8) -- (0,0.8);
\fill (0,0.8) circle (1.4pt) node[right] {$p^{\star}$};
\fill (0,0.54) circle (1.4pt) node[right] {$g(\lambda)$};
\draw (-3.2,1.02) -- (3.4,0.03);
\node[anchor=east] at (-3.25,1.12) {$\lambda u + t = g(\lambda)$};
\end{tikzpicture}针对只有一个(不等式)约束的简单问题,对偶函数和下界

TikZ 代码
\begin{tikzpicture}[line cap=round,line join=round,scale=0.95]
% G set and supporting lines for three dual feasible lambdas
\draw[->] (-3.4,0) -- (3.6,0) node[right] {$u$};
\draw[->] (0,-1.4) -- (0,2.6) node[above] {$t$};
\draw[fill=gray!20] plot[smooth cycle,tension=0.85]
coordinates {(-2.7,1.15) (-1.9,1.6) (-0.9,1.95) (0.3,2.0) (1.3,1.75) (2.0,1.3) (2.6,0.5) (1.2,0.6) (0,0.8) (-1.4,0.75)};
\node at (1.0,1.6) {$\mathcal{G}$};
\fill (0,0.8) circle (1.4pt) node[above right] {$p^{\star}$};
\fill (0,0.54) circle (1.4pt) node[right] {$d^{\star}$};
\fill (0,-0.065) circle (1.4pt);
% lambda1: nearly flat, touching the right tail
\draw (-3.2,0.529) -- (3.4,0.496);
\node[anchor=east] at (-3.25,0.62) {$\lambda_1 u + t = g(\lambda_1)$};
% lambda*: touching the dip
\draw (-3.2,1.02) -- (3.4,0.03);
\node[anchor=east] at (-3.25,1.12) {$\lambda^{\star} u + t = g(\lambda^{\star})$};
% lambda2: steep, intercept below zero
\draw (-3.2,1.375) -- (2.4,-1.145);
\node[anchor=east] at (-3.25,1.5) {$\lambda_2 u + t = g(\lambda_2)$};
\end{tikzpicture}如图所示,对偶可行的三个
上镜图变化
通过下面的理解,我们就可以解释为什么对(大部分)凸问题,强对偶性总是成立。定义集合
我们可以认为
可以用
同样的,我们可以通过极小化仿射函数得到关于
如果确定下确界有界,则
定义了
特别地,由于
即弱对偶性成立。强对偶性成立,当且仅当存在某些对偶可行变量

TikZ 代码
\begin{tikzpicture}[line cap=round,line join=round,scale=0.95]
% epigraph variation A = G + R_+ x R_+
% shared bottom boundary of G (from right tail back to left tip)
\fill[gray!20] (-2.7,1.15) -- (-2.7,2.35) -- (3.3,2.35) -- (3.3,0.5) -- (2.6,0.5)
plot[smooth] coordinates {(2.6,0.5) (1.2,0.6) (0,0.8) (-1.4,0.75) (-2.7,1.15)};
% outline of G: smooth top + smooth bottom (same points)
\draw plot[smooth] coordinates {(-2.7,1.15) (-1.9,1.6) (-0.9,1.95) (0.3,2.0) (1.3,1.75) (2.0,1.3) (2.6,0.5)};
\draw plot[smooth] coordinates {(2.6,0.5) (1.2,0.6) (0,0.8) (-1.4,0.75) (-2.7,1.15)};
\node at (2.7,2.0) {$\mathcal{A}$};
\draw[->] (-3.4,0) -- (3.6,0) node[right] {$u$};
\draw[->] (0,-0.8) -- (0,2.6) node[above] {$t$};
\draw[dashed] (-2.0,0.8) -- (0,0.8);
\fill (0,0.8) circle (1.4pt) node[right] {$(0, p^{\star})$};
\fill (0,0.54) circle (1.4pt) node[right] {$(0, g(\lambda))$};
\draw (-3.2,1.02) -- (3.4,0.03);
\node[anchor=east] at (-3.25,1.12) {$\lambda u + t = g(\lambda)$};
\end{tikzpicture}针对具有一个(不等式)约束的问题,对偶函数和下界
在约束规则下强对偶性成立的证明
本文从略
多准则解释
多准则凸优化问题的每个 Pareto 最优解都是给定某个非负权向量
的最小点,我们考虑集合
这和研究 Lagrange 对偶问题时的定义一样,此时所需权向量也是集合在任意一个 Pareto 最优点处的支撑超平面。在多准则优化问题中,权向量的含义是目标函数的相对权重。当我们固定权向量的最后一个分量(和函数