I'm trying to figure out how the sum of two spin 1/2 operators along an arbitrary direction would work.
These operators are of the form \begin{equation} \textbf{S}_j \cdot \hat{\textbf{n}}_j = \frac{\hbar}{2} \begin{pmatrix} \cos (\theta_j) & e^{-i \varphi_j} \sin (\theta_j)\\ e^{i \varphi_j} \sin (\theta_j) & -\cos (\theta_j) \end{pmatrix}_j \end{equation} where $j = 1, 2$, $\textbf{S}_j = \frac{\hbar}{2} \boldsymbol{\sigma}_j$, and $\hat{\textbf{n}}_j = \sin (\theta_j) \cos (\varphi_j) \hat{\textbf{i}} + \sin (\theta_j) \sin (\varphi_j) \hat{\textbf{j}} + \cos (\theta_j) \hat{\textbf{k}}$. The eigenvalues of these operators are $\pm \frac{\hbar}{2}$ which I'll identify with $\pm$ respectively. Then, the eigenvectors of these eigenvalues are \begin{align*} |\textbf{S}_j \cdot \textbf{n}_j, + \rangle &= \cos \Big( \frac{\theta_j}{2} \Big) |S_z, + \rangle^{(j)} + e^{i \varphi_j} \sin \Big( \frac{\theta_j}{2} \Big) |S_z, - \rangle^{(j)}\\ |\textbf{S}_j \cdot \textbf{n}_j, - \rangle &= \sin \Big( \frac{\theta_j}{2} \Big) |S_z, + \rangle^{(j)} - e^{i \varphi_j} \cos \Big( \frac{\theta_j}{2} \Big) |S_z, - \rangle^{(j)} \end{align*} These are also eigenvectors of $S^2$, both of them with eigenvalue $\frac{3 \hbar^2}{4}$. Let $\textbf{S} \cdot \hat{\textbf{n}} = \textbf{S}_1 \cdot \hat{\textbf{n}}_1 + \textbf{S}_2 \cdot \hat{\textbf{n}}_2$. Since $[\textbf{S}_1 \cdot \hat{\textbf{n}}_1, \textbf{S}_2 \cdot \hat{\textbf{n}}_2] = 0$ then \begin{equation*} (\textbf{S} \cdot \hat{\textbf{n}})^2 = (\textbf{S}_1 \cdot \hat{\textbf{n}}_1)^2 + (\textbf{S}_2 \cdot \hat{\textbf{n}}_2)^2 + 2 (\textbf{S}_1 \cdot \hat{\textbf{n}}_1) \cdot (\textbf{S}_2 \cdot \hat{\textbf{n}}_2) \end{equation*} From our first equation, we can see that $(\textbf{S}_j \cdot \hat{\textbf{n}}_j)^2 = \frac{\hbar^2}{4} \mathbb{1}$, so our last equation becomes \begin{equation*} (\textbf{S} \cdot \hat{\textbf{n}})^2 = \frac{\hbar^2}{2} \mathbb{1} + 2 (\textbf{S}_1 \cdot \hat{\textbf{n}}_1) \cdot (\textbf{S}_2 \cdot \hat{\textbf{n}}_2) \end{equation*} This is the part where I get confused, because how would the term $2 (\textbf{S}_1 \cdot \hat{\textbf{n}}_1) \cdot (\textbf{S}_2 \cdot \hat{\textbf{n}}_2)$ operate, for example, on the state $|\textbf{S}_1 \cdot \textbf{n}_1, + \rangle |\textbf{S}_2 \cdot \textbf{n}_2, - \rangle$? In the usual case where we work with $\textbf{S} = \frac{\hbar}{2} (\sigma_x \hat{\textbf{i}} + \sigma_y \hat{\textbf{j}} + \sigma_z \hat{\textbf{k}})$, said term takes the form \begin{equation*} 2 \textbf{S}_1 \cdot \textbf{S}_2 = S_{+_1} S_{-_2} + S_{-_1} S_{+_2} + 2 S_{z_1} S_{z_2} \end{equation*} Which is useful because we know how these operators work with the states $|S_z, \pm \rangle$; but in the more general case, how would the term $2 (\textbf{S}_1 \cdot \hat{\textbf{n}}_1) \cdot (\textbf{S}_2 \cdot \hat{\textbf{n}}_2)$ expand in order to work with the tensor product between the states $|\textbf{S}_1 \cdot \textbf{n}_1, \pm \rangle$ and $|\textbf{S}_2 \cdot \textbf{n}_2, \pm \rangle$?