The Dirac bra-ket notation is a concise way to represent objects in a complex vector space \(\mathbb{C}^{n}\).

A ket, denoted by \(\vert \textbf{v} \rangle\), is a vector \(\textbf{v}\). Since a linear operator \(\hat{O}\) maps a vector to another vector, we have \(\hat{O}\vert \boldsymbol{v_{1}}\rangle=\vert \boldsymbol{v_{2}} \rangle\).
A bra, denoted by \(\langle\boldsymbol{u}\vert\) , is often associated with a ket in the form of an inner product, denoted by \(\langle\boldsymbol{u}\vert\boldsymbol{v}\rangle\). If a ket is expressed as a column vector, the corresponding bra is the conjugate transpose of its ket, i.e. \(\langle\boldsymbol{u}\vert=\vert\boldsymbol{u}\rangle^{\dagger}\). The inner product can therefore be written as the following matrix multiplication:
$$\langle\boldsymbol{u}\vert\boldsymbol{v}\rangle=\begin{pmatrix} u_{1}^{*}&u_{2}^{*}& \cdots & u_{N}^{*}\end{pmatrix}\begin{pmatrix} v_{1}\\v_{2} \\ \vdots \\v_{N} \end{pmatrix}=\sum_{i=1}^{N}u_{i}^{*}v_i$$
or in the case of functions:
$$\langle\boldsymbol{u}\vert\boldsymbol{v}\rangle=\int_{-\infty}^{\infty}f_{u}^{*}(x)f_{v}(x)dx$$
Since a linear operator acting on a ket is another ket, we can express an inner product as:
$$\langle\phi_{i}\vert\phi_{k}\rangle=\langle\phi_{i}\vert\hat{O}\vert\phi_{j}\rangle=\int \phi_{i}^{*}\hat{O}\phi_{j}d\tau$$
where \(\hat{O}\vert\phi_{j}\rangle=\vert\phi_{k}\rangle\).
If \(i=j\), then \(\langle\phi\vert\hat{O}\vert\phi\rangle\) is the expectation value (or average value) of the operator \(\hat{O}\).
As mentioned above, bras and kets can be represented by matrices. Therefore, the multiplication of a bra and a ket that involves a linear operator is associative, e.g.:
$$\langle\boldsymbol{u}\vert(\hat{O}\vert\boldsymbol{v}\rangle)=(\langle\boldsymbol{u}\vert\hat{O})\vert\boldsymbol{v}\rangle\equiv\langle\boldsymbol{u}\vert\hat{O}\vert\boldsymbol{v}\rangle$$
$$(\vert\boldsymbol{u}\rangle\langle\boldsymbol{v}\vert)\vert\boldsymbol{w}\rangle=\vert\boldsymbol{u}\rangle(\langle\boldsymbol{v}\vert\boldsymbol{w}\rangle)$$
$$(\hat{O}\vert\boldsymbol{u}\rangle)\langle\boldsymbol{v}\vert=\hat{O}(\vert\boldsymbol{u}\rangle\langle\boldsymbol{v}\vert)\equiv\hat{O}\vert\boldsymbol{u}\rangle\langle\boldsymbol{v}\vert$$
You can verify the above examples using a 2×2 matrix with complex elements to represent the operator acting on a vector in \(\mathbb{C}^{2}\). The three examples reveal that:
-
- \(\hat{O}\vert\boldsymbol{v}\rangle\) produces another ket.
- \(\langle\boldsymbol{u}\vert\hat{O}\) results in another bra. This is because \((\langle\boldsymbol{u}\vert\hat{O})\vert\boldsymbol{v}\rangle=\langle\boldsymbol{u}\vert(\hat{O}\vert\boldsymbol{v}\rangle)=\langle\boldsymbol{u}\vert\boldsymbol{v^{\prime}}\rangle=c\), where \(c\) is a scalar; and if \((\langle\boldsymbol{u}\vert\hat{O})\vert\boldsymbol{v}\rangle=c\), the only possible identity of \((\langle\boldsymbol{u}\vert\hat{O})\) is a bra.
- \(\vert\boldsymbol{u}\rangle\langle\boldsymbol{v}\vert\) , which is called an outer product, is an operator because \((\vert\boldsymbol{u}\rangle\langle\boldsymbol{v}\vert)\vert\boldsymbol{w}\rangle=\vert\boldsymbol{u}\rangle(\langle\boldsymbol{v}\vert\boldsymbol{w}\rangle)=c\vert\boldsymbol{u}\rangle\), i.e. \(\vert\boldsymbol{u}\rangle\langle\boldsymbol{v}\vert\) maps the ket \(\vert\boldsymbol{w}\rangle\) to another ket \(c\vert\boldsymbol{u}\rangle\). In other words, the operator \(\vert\boldsymbol{u}\rangle\langle\boldsymbol{v}\vert\) transforms the vector \(\vert\boldsymbol{w}\rangle\) in the direction of the vector \(\vert\boldsymbol{u}\rangle\), i.e. \(\vert\boldsymbol{u}\rangle\langle\boldsymbol{v}\vert\) projects \(\vert\boldsymbol{w}\rangle\) onto \(\vert\boldsymbol{u}\rangle\).
- The product of two linear operators is another linear operator: \(\hat{O}\hat{O^{\prime}}=\hat{O}(\vert\boldsymbol{u}\rangle\langle\boldsymbol{v}\vert)=(\hat{O}\vert\boldsymbol{u}\rangle)\langle\boldsymbol{v}\vert=\vert\boldsymbol{{u^{\prime}}\rangle\langle\boldsymbol{v}\vert=\hat{O^{\prime}^{\prime}}\).
