全部笔记的汇总贴(视频也有传送门):中科大-凸优化
几种解释 P ∗ = d ∗ P^*=d^* P∗=d∗
min f 0 ( x ) s . t . f i ( x ) ≤ 0 , i = 1 , ⋯ , m x ∈ D \min f_0(x)\\s.t.\;f_i(x)\le0,i=1,\cdots,m\\x\in D minf0(x)s.t.fi(x)≤0,i=1,⋯,mx∈D
一、几何解释
min f 0 ( x ) s . t . f i ( x ) ≤ 0 \min f_0(x)\\s.t. f_i(x)\le0 minf0(x)s.t.fi(x)≤0
G = { ( f ( x ) , f 0 ( x ) ) ∣ x ∈ D } P ∗ = inf { t ∣ ( u , t ) ∈ G , u ≤ 0 } g ( λ ) = inf { λ u + t ∣ ( u , t ) ∈ G } ( L ( x , λ ) = λ u + t ) G=\{(f(x),f_0(x))|x\in D\}\\P^*=\inf\{t|(u,t)\in G,u\le0\}\\g(\lambda)=\inf\{\lambda u+t|(u,t)\in G\}{\color{blue}(L(x,\lambda)=\lambda u+t)} G={ (f(x),f0(x))∣x∈D}P∗=inf{ t∣(u,t)∈G,u≤0}g(λ)=inf{ λu+t∣(u,t)∈G}(L(x,λ)=λu+t)
二、鞍点的解释(Saddle Point)
三、多目标优化解释
min { f 0 ( x ) , f 1 ( x ) , ⋯ , f m ( x ) } { λ i } ⇒ min f 0 ( x ) + ∑ i λ i f i ( x ) ( = L ( x , λ ) ) \min\{f_0(x),f_1(x),\cdots,f_m(x)\}\\\{\lambda_i\}\Rightarrow\min f_0(x)+\sum_i\lambda_if_i(x){\color{blue}(=L(x,\lambda))} min{
f0(x),f1(x),⋯,fm(x)}{
λi}⇒minf0(x)+i∑λifi(x)(=L(x,λ))
min f 0 ( x ) s . t . f i ( x ) ≤ 0 , i = 1 , ⋯ , m \min f_0(x)\\s.t.\;f_i(x)\le0,i=1,\cdots,m minf0(x)s.t.fi(x)≤0,i=1,⋯,m
给定 λ \lambda λ,
四、经济学的解释
33、34是两位助教讲附录的知识点,暂时跳过。
下一章传送门:中科大-凸优化 笔记(lec35)-鞍点定理