Notes about math, research, and more.

Suppose that \(A,B \in M_{n}(\mathbb{C})\) are commuting matrices. Let \(V_{\lambda}\) be an eigenspace of \(A\nothing\). Show that \(V_{\lambda}\) is \(B\text{-invariant}\).

Let \(v \in V_{\lambda}\). Since \(A,B\) commute, \(A(Bv) = B(Av)\). Since \(v \in V_{\lambda}\), \[ Av = \lambda v \implies B(Av) = B(\lambda v) = \lambda (Bv) \implies Bv \in V_{\lambda}. \] Hence, \(V_{\lambda}\) is \(B-\text{invariant.}\)

Let \(V\) be an inner product space and let \(W \leq V\) be a subspace. Let \(v \in V\) and define \(\hat{v} \in W\) as in the proof of Proposition 2.2.3. Prove that if \(w \in W\) with \(w\neq \hat{v}\), then \(\left\| v-\hat{v} \right\| < \left\| v-w \right\|\). Deduce that \(\hat{v}\) is independent of the choice of orthonormal basis for \(W\). It is called the orthogonal projection of \(v\) onto \(W\).

Hint: Use the notation of the proof of Proposition 2.2.3. Let \(\{ e_{m+1},\ldots,e_{n} \}\) be an orthonormal basis for \(W^\perp\). Then \(\{ e_{1},\ldots,e_{n} \}\) is an orthonormal basis for \(V\). Say \(v = \sum_{i=1}^n a_i e_{i}\) and \(w = \sum_{i=1}^m b_{i}e_{i}\). To compute the norms of \(v - \hat{v}\) and \(v-w\), express both vectors as linear combinations of \(e_{1},\ldots,e_{n}\).

Let \(B = \{ e_{1},\ldots,e_{n} \}\) be an orthonormal basis for \(V\) and \(W\) the subspace of \(V\) spanned by \(\{ e_{m+1},\ldots,e_{n} \}\). Let
  1. \(v = \sum_{i=1}^n a_{i}e_{i}\),
  2. \(\hat{v} = \sum_{i=m+1}^n a_{i} e_{i}\), and
  3. \(W \ni w =\sum_{i=m+1}^{n} b_{i}e_{i} \neq \hat{v}\).
We calculate \(\left\| v-\hat{v}\right\|\) and \(\left\| v-w \right\|\): \begin{align*} \left\| v-\hat{v} \right\| = \left\|\sum_{i=1}^{n} a_{i} e_{i} - \sum_{i=m+1}^{n} a_{i} e_{i} \right\| = \left\|\sum_{i=1}^{m} a_{i} e_{i} \right \| \end{align*} and \begin{align*} \left\| v-w \right\|^2 &= \left\| \sum_{i=1}^n a_{i}e_{i} - \sum_{i=m+1}^{n} b_{i}e_{i} \right\|^2 \\ &= \left\| \sum_{i=1}^m a_{i}e_{i} + \sum_{i=m+1}^{n} (a_{i}-b_{i})e_{i} \right\|^2 \\ &= \left\| \sum_{i=1}^m a_{i}e_{i}\right\|^2 + \left\|\sum_{i=m+1}^{n} (a_{i}-b_{i})e_{i} \right\|^2 \\ \left\| v-w \right\| &= \sqrt{\left\| v-\hat{v} \right\|^2 + \left\|\sum_{i=m+1}^{n} (a_{i}-b_{i})e_{i} \right\|^2} \\ \end{align*} Since \(w \neq \hat{v}\), there is some \(i \in \{ m+1,\ldots,n \}\) such that \(a_{i}\neq b_{i} \implies\left\|\sum_{i=m+1}^{n} (a_{i}-b_{i})e_{i}\right\|^2 > 0\). Hence, \(\left\| v-w \right\| > \left\| v-\hat{v} \right\|\). Further, the norm of an element does not depend on the choice of orthonormal basis the element is expressed in, so \(\left\| v-\hat{v} \right\| < \left\| v-w \right\|\) holds independent of the choice of orthonormal basis for \(W\).

If \(B\) is an orthonormal basis of an inner product space \(V\), and if \(T \in \text{End}(V)\), then \[ [T^*]_{B} = ([T]_{B})^*. \] Hint: Say \(B = \{ v_{1},\ldots,v_{n} \}\). Then

\[ ([T^*]_{B})_{ij} = (i^{th} \text{ entry of } [T^*v_{j}]_{B}) = \langle T^* v_{j}, v_{i} \rangle. \]

We show that \([T^*]_{B} = ([T]_{B})^*\) by showing that for each \(i,j \in \{1,\ldots,n \}\), \(([T^*]_{B})_{ij} = \overline{([T]_{B})_{ji}}\). Let \(i,j \in \{1,\ldots,n \}\). By the hint, \begin{align*} ([T^*]_{B})_{ij} &= \langle T^* v_{j}, v_{i} \rangle\\ &= \langle v_{j}, Tv_{i} \rangle \text{ since }(T^*)^* = T; \\ &= \overline{\langle Tv_{i}, v_{j} \rangle} \\ &= \overline{([T]_{B})_{ji}} \text{ by the hint.} \end{align*} Hence, for each \(i,j \in \{1,\ldots,n \}\), \(([T^*]_{B})_{ij} = \overline{([T]_{B})_{ji}}\) which implies \([T^*]_{B} = \overline{([T]_{B})^T} = ([T]_{B})^*\).

Let \(V\) be an inner product space and \(T \in \text{End}(V)\). Assume that for some orthonormal basis \(B\) of \(V\), the matrix \([T]_{B}\) is self-adjoint. Then \(T\) is a self-adjoint linear operator on \(V\).

To show \(T\) is self-adjoint we must show that for all \(v,w \in V\), \(\langle Tv, w \rangle = \langle v, Tw \rangle\). Denote the elements of \(B\) as \(\{ v_{1},\ldots,v_{n} \}\). For each \(i,j \in \{1,\ldots,n \}\), \begin{align*} \left\langle T v_{i}, v_{j} \right\rangle &= ([T]_{B})_{ji} \text{ by hint in Exercise A;}\\ &= \overline{([T]_{B})_{ij}} \text{ by assumption;} \\ &= \overline{\left\langle Tv_{j}, v_{i} \right\rangle} \\ &= \left\langle v_{i}, Tv_{j} \right\rangle. \end{align*} Let \(v = \sum_{i=1}^{n} a_{i}v_{i}, w = \sum_{i=1}^n b_{i}v_{i} \in V\). From above, \begin{align*} \left\langle Tv, w \right\rangle &= \left\langle T\sum_{i=1}^{n} a_{i}v_{i}, \sum_{j=1}^{n} b_{j}v_{j} \right\rangle \\ &= \sum_{i=1}^{n} \sum_{j=1}^{n} a_{i}\overline{b_{j}} \left\langle Tv_{i}, v_{j} \right\rangle \text{ by linearity of \(T\);}\\ &= \sum_{i=1}^{n} \sum_{j=1}^{n} a_{i}\overline{b_{j}} \left\langle v_{i}, Tv_{j} \right\rangle \text{ from above;}\\ &= \langle v, Tw \rangle. \end{align*} Since \(T^*\) is the unique operator such that \(\langle Tv, w \rangle = \langle v, T^*w \rangle\), we deduce \(T = T^*\).