全部笔记的汇总贴(视频也有传送门):中科大-凸优化
一、函数的共轭
f ∗ f^* f∗是 f : R n → R f:\R^n\rightarrow\R f:Rn→R的共轭,若 f ∗ ( y ) = sup x ∈ d o m f ( y T x − f ( x ) ) f^*(y)=\sup_{x\in dom\;f}(y^Tx-f(x)) f∗(y)=supx∈domf(yTx−f(x))
例:
min f 0 ( x ) s . t . x = 0 ⇒ L ( x , v ) = f ( x ) + V T x , d o m L = d o m f ∗ R n ⇒ g ( v ) = inf x ∈ d o m f ( f ( x ) + V T x ) = − sup x ∈ d o m f ( − V T x − f ( x ) ) = − f ∗ ( − v ) \min\;f_0(x)\\s.t.\;x=0\\\Rightarrow L(x,v)=f(x)+V^Tx,dom\;L=dom\;f*\R^n\\\Rightarrow g(v)=\inf_{x\in dom\;f}(f(x)+V^Tx)\\=-\sup_{x\in dom\;f}(-V^Tx-f(x))\\=-f^*(-v) minf0(x)s.t.x=0⇒L(x,v)=f(x)+VTx,domL=domf∗Rn⇒g(v)=x∈domfinf(f(x)+VTx