1
$\begingroup$

To preface, I'm an applied mathematician trying to parse the meaning of physics notation I've come across in a paper. My goal is to understand the setting in terms of matrices and vectors so that I can test an algorithm I'm studying. Since I don't know the terminology or notation, I haven't been able to figure out how to read more about the topic.

I'm reading about spin systems and I see expressions like: $$ \mathbf{H} = -\sum_{i,j} J_{i,j} \mathbf{s}(i) \cdot \mathbf{s}(j) $$ where $\mathbf{s}(i)$ is the spin operator at site $i$.

My understanding is that $\mathbf{H}$ is can be represented as a matrix of size $(2s+1)^N$ where $N$ is the number of spins and and $s$ is the spin number. I have also seem the spin matrices for specific values of $s$, which are of size $2s+1$ (although I'm not sure if these are the same as the $\mathbf{s}(i)$ since they seem to have $x, y, z, +, -)$. So, the missing piece for me is what is the meaning of $\mathbf{s}(i) \cdot \mathbf{s}(j)$ as well as the meaning of the sum.

$\endgroup$

1 Answer 1

4
$\begingroup$

To start: What is $\mathbf{s}(i)$? This is not an operator (not a matrix) but rather meant to denote a vector of operators. \begin{equation}\mathbf{s}(i)=(s_x(i),s_y(i),s_z(i)),\end{equation} where $s_\sigma(i)$ is an operator, a $(2s+1)^N$-dimensional matrix.

What does the dot product between them mean? It means to imitate the usual dot product, as in \begin{equation} \mathbf{s}(i)\cdot\mathbf{s}(j) = s_x(i)s_x(j)+s_y(i)s_y(j)+s_z(i)s_z(j) \end{equation} where the multiplication $s_\sigma(i)s_\sigma(j)$ is simply matrix-matrix multiplication.

Finally: What are the matrices $s_\sigma(i)$, and how are they $(2s+1)^N$-dimensional? You probably already know the form of the matrix $s_\sigma$ for $\sigma=x,y,z$; these are $(2s+1)$-dimensional matrices. When we describe multiple spins, the total Hilbert space is a tensor product of the Hilbert space of the individual spins, so our total Hilbert space for $N$ particles is $\bigotimes^N\mathbb{C}^{2s+1}$, where the $i$th copy of $\mathbb{C}^{2s+1}$ in the tensor product represents the state of the $i$th particle. This is a $(2s+1)^N$-dimensional space, which is why the overall Hamiltonian is a $(2s+1)^N$-dimensional matrix. When we write $s_\sigma(i)$, we mean the spin operator $s_\sigma$ that acts only on the part of the Hilbert space associated to the $i$th spin. So, strictly speaking, we have \begin{equation} s_\sigma(i) = I\otimes I\otimes\cdots I\otimes \underbrace{s_\sigma}_{i\text{th position}}\otimes I\cdots \otimes I \end{equation} which makes $s_\sigma(i)$ a $(2s+1)^N$ dimensional matrix.

$\endgroup$
3
  • $\begingroup$ Thanks! This is super helpful. So then the sum is just a linear combination of these matrices? Also, do you happen to know a good elementary textbook where I could read more? $\endgroup$ Commented Dec 29, 2021 at 22:49
  • $\begingroup$ That's right, they just add like matrices. I'm afraid I don't know a good reference off the top of my head $\endgroup$ Commented Dec 29, 2021 at 23:38
  • $\begingroup$ No worries, I think this should be enough to get me started! $\endgroup$ Commented Dec 29, 2021 at 23:39

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.