9.3 Positive Forms

这一节讲的内容在国内的学校教材中常被列为“（半）正定二次型”，从form的角度理解比从矩阵的角度理解要容易一些。$f$是positive与由$f$的矩阵$A$形成的函数$g(X,Y)=Y^{\ast }AX$是positive这二者等价，而Theorem 5说明：$g(X,Y)=Y^{\ast }AX$是positive又等价于存在可逆矩阵$P$使得$A=P^{\ast }P$。这个定理实际用起来并不容易，为了引入对positive的判别，后面介绍了principle minors的定义，也就是$A$的前k行和前k列形成的子矩阵的行列式，记为${\Delta }_k(A)$，而通过引理和Theorem 6，$f$是positive的另一个等价条件是：$A=A^{\ast }$且$A$的所有principle minor都是正的。
在上述这些内容都讨论过后，再定义了positive matrix，即对任何$X\ne 0$都有$X^{t}AX>0$，以及linear operator是positive的定义。这些都可以从form中自然的得到。另外一个有用的结论是：如果$f$是一个positive form，那么$f$实际是一个inner product。
本节的最后，给出了一个关于positive operator，positive forms和positive matrices的等价条件的总结。

Exercises

Exercise 1. Let $V$ be $C^2$, with the standard inner product. For which vectors $\alpha \in V$ is there a positive linear operator $T$ such that $\alpha =T{ϵ}_1$?
Solution: If $T$ is positive, then $T=T^{\ast }$ and we shall have $(\alpha |{ϵ}_1)>0$, so write $\alpha =(a,b)$, we know that $a>0$.

Exercise 2. Let $V$ be $R^2$, with the standard inner product. If $\theta$ is a real number, let $T$ be the linear operator ‘rotation through $\theta$,’
$$T_{\theta }(x_1,x_2)=(x_1\cos \theta -x_2\sin \theta ,x_1\sin \theta +x_2\cos \theta )$$
For which values of $\theta$ is $T_{\theta }$ a positive operator?
Solution: The matrix of $T_{\theta }$ with the standard basis is $A=\left[\begin{array}{cc}\cos \theta & -\sin \theta \\ \sin \theta & \cos \theta \end{array}\right]$. If $T$ is positive, then we require $A=A^{\ast }$ and $\cos \theta >0$, thus we must have $\sin \theta =0,\cos \theta =1$, which gives $\theta =2k\pi ,k\in N$.

Exercise 3. Let $V$ be the space of $n\times 1$ matrices over $C$, with the inner product $(X|Y)=Y^{\ast }GX$ (where $G$ is an $n\times n$ matrix such that this is an inner product). Let $A$ be an $n\times n$ matrix and $T$ the linear operator $T(X)=AX$. Find $T^{\ast }$. If $Y$ is a fixed element of $V$, find the element $Z$ of $V$ which determines the linear functional $X\to Y^{\ast }X$. In other words, find $Z$ such that $Y^{\ast }X=(X|Z)$ for all $X\in V$.
Solution: Since $(X|Y)=Y^{\ast }GX$ is an inner product, we must have $G$ positive, thus invertible.
We have $(T(X)|Y)=Y^{\ast }GAX$, so if we let $U(Y)=(GA{G}^{-1}{)}^{\ast }Y$, we shall have
$$(X|U(Y))=Y^{\ast }GA{G}^{-1}GX=Y^{\ast }GAX=(T(X)|Y)⟹U=T^{\ast }$$
If $Y$ is a fixed element of $V$, then let $Z=(G^{-1}{)}^{\ast }Y$, we have $(X|Z)=Y^{\ast }{G}^{-1}GX=Y^{\ast }X$.

Exercise 4. Let $V$ be a finite-dimensional inner product space. If $T$ and $U$ are positive linear operators on $V$, prove that $(T+U)$ is positive. Give an example which shows that $TU$ need not be positive.
Solution: We directly have $(T+U{)}^{\ast }=T^{\ast }+U^{}$