Consider a system of two distinguishable spin-1/2 particles with Hamiltonian \begin{align} H &= \frac{\alpha}{4} \vec{\sigma}_1 \cdot\vec{\sigma}_2.\\ \end{align} where $\vec{\sigma}_1 = (\sigma_x\otimes 1, \sigma_y\otimes 1, \sigma_z\otimes 1)$ and $\vec{\sigma}_2 = (1\otimes \sigma_x,1\otimes \sigma_y,1\otimes \sigma_z)$ . In the uncoupled z-basis, we can write the Hamiltonian as \begin{align} H&= \frac{\alpha}{4}\left(\sigma_{x}\otimes\sigma_{x}+\sigma_{y}\otimes\sigma_{y}+\sigma_{z}\otimes\sigma_{z}\right)\\ &= \frac{\alpha}{4}\left(\sigma_{x}+\sigma_{y}+\sigma_{z}\right)\otimes\left(\sigma_{x}+\sigma_{y}+\sigma_{z}\right)\\ &= \frac{\alpha}{4}\begin{pmatrix}1 & 1-i\\ 1+i & -1\end{pmatrix}\otimes \begin{pmatrix}1 & 1-i\\ 1+i & -1\end{pmatrix} \end{align} The matrix $$\begin{pmatrix}1 & 1-i\\ 1+i & -1\end{pmatrix}$$ has eigenvalues $\pm\sqrt{3}$, so in the uncoupled diagonal-basis $$H = \frac{3\alpha}{4}\begin{pmatrix}1 & 0\\ 0 & -1\end{pmatrix}\otimes\begin{pmatrix}1 & 0\\ 0 & -1\end{pmatrix}$$ which has eigenvectors $$\begin{pmatrix}1\\0\end{pmatrix}\otimes\begin{pmatrix}1\\0\end{pmatrix}\hspace{2mm}, \hspace{2mm}\begin{pmatrix}1\\0\end{pmatrix}\otimes\begin{pmatrix}0\\1\end{pmatrix} \\ \begin{pmatrix}0\\1\end{pmatrix}\otimes\begin{pmatrix}1\\0\end{pmatrix}\hspace{2mm}, \hspace{2mm} \begin{pmatrix}0\\1\end{pmatrix}\otimes\begin{pmatrix}0\\1\end{pmatrix}$$ with respective eigenvalues $3\alpha/4, -3\alpha/4, -3\alpha/4, 3\alpha/4$.

We could've rewritten the Hamiltonian as \begin{align} H &= \frac{\alpha}{2}\left[\left(\frac{1}{2}\vec{\sigma}_1+\frac{1}{2}\vec{\sigma}_2\right)^2 - \left(\frac{1}{2}\vec{\sigma}_1\right)^2 - \left(\frac{1}{2}\vec{\sigma}_2\right)^2\right]\\ &=\frac{\alpha}{2}\left[s(s+1) - \frac{1}{2}\left(\frac{1}{2} +1\right) - \frac{1}{2}\left(\frac{1}{2} +1\right)\right]\\ &=\frac{\alpha}{2}\left[s(s+1) - \frac{3}{2}\right] \end{align} where $s$ is the the spin in the coupled basis ($s=0$ or $1$). Therefore the eigenvalues of the Hamiltonian in the coupled basis are $-3\alpha/4$ (with degeneracy 1) and $\alpha/4$ (with degeneracy 3).

The eigenvalues of the Hamiltonian shouldn't depend on your choice of basis, but in the above I get different eigenvalues in the coupled and uncoupled bases. Where am I going wrong?

Solution (thanks to Vadim): In the $|\uparrow\uparrow\rangle, |\uparrow\downarrow\rangle,|\downarrow\uparrow\rangle,|\downarrow\downarrow\rangle$ basis the Hamiltonian takes the form \begin{align} H&= \frac{\alpha}{4}\left(\sigma_{x}\otimes\sigma_{x}+\sigma_{y}\otimes\sigma_{y}+\sigma_{z}\otimes\sigma_{z}\right)\\ &= \frac{\alpha}{4}\begin{pmatrix}1&0&0&0\\0&-1&2&0\\0&2&-1&0\\0&0&0&1\end{pmatrix} \end{align} which has eigenvalues $-3\alpha/4$ and $\alpha/4$. This is not the same as \begin{align} \frac{\alpha}{4}\left(\sigma_{x}+\sigma_{y}+\sigma_{z}\right)\otimes\left(\sigma_{x}+\sigma_{y}+\sigma_{z}\right) = \frac{\alpha}{4}\begin{pmatrix}1&1-i&1-i&-2i\\1+i&-1&2&-1+i\\1+i&2&-1&-1+i\\2i&-1-i&-1-i&1\end{pmatrix} \end{align} which has eigenvalues $\pm3\alpha/4$.

  • $\begingroup$ The most straightforward way is to work in the basis of two-particle states: $|\uparrow, \uparrow\rangle, |\uparrow, \downarrow\rangle, |\downarrow, \uparrow\rangle, |\downarrow, \downarrow\rangle$. The resulting 4-by-4 matrix is actually easily diagonalizable. $\endgroup$
    – Roger V.
    Apr 7, 2020 at 16:31

1 Answer 1


The error is in the first approach: $$\sigma_x\otimes\sigma_x + \sigma_y\otimes\sigma_y + \sigma_z\otimes\sigma_z \neq (\sigma_x + \sigma_y + \sigma_z)\otimes(\sigma_x + \sigma_y + \sigma_z),$$ as it is easy to verify by writing down these matrices in 4-by-4 basis $|\uparrow\uparrow\rangle, |\uparrow\downarrow\rangle, |\downarrow\uparrow\rangle, |\downarrow\downarrow\rangle.$ Working with 4-by-4 matrices may seem dounting at first, but it is actually quite easy, once you get a grasp of how they nest one within the other, e.g. $$ \sigma_x^{(1)}\otimes\sigma_x^{(2)} =\begin{pmatrix} 0&\sigma_x^{(2)}\\\sigma_x^{(2)}&0\end{pmatrix} = \begin{pmatrix} 0&0&0&1\\0&0&1&0\\0&1&0&0\\1&0&0&0 \end{pmatrix}$$ $$ \sigma_x^{(1)}\otimes\sigma_y^{(2)} =\begin{pmatrix} 0&\sigma_y^{(2)}\\\sigma_y^{(2)}&0\end{pmatrix} = \begin{pmatrix} 0&0&0&-i\\0&0&i&0\\0&-i&0&0\\i&0&0&0 \end{pmatrix}$$ Incidentally, it is also helpful when dealing with the $\gamma$-matrices in the Dirac equation.

  • $\begingroup$ Just using distributivity shows one would get $\sigma_x\otimes\sigma_y$ terms which do not appear in the original Hamiltonian. $\endgroup$ Apr 7, 2020 at 19:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.