0
$\begingroup$

I have 3 operators which are generators of $SO(2,1)$. I don't know the expressions of these operators as differential operators, but I know their action on a basis $\left|l',m\right>$:

$$J_{\mu\nu}\left|l',m\right>=\sum_{\tilde{l'},\tilde{m}}C_{\mu\nu}^{\left(l',m\right),\left(\tilde{l'},\tilde{m}\right)}\left|\tilde{l'},\tilde{m}\right>$$

These relations are quite long, but the shortest one is of the form:

$$J_{01}\left|l',m\right>=\frac{i}{2}\left(l'+m\right)\left|l',m\right>-\frac{il'}{4|l'|}\sqrt{\left(l'+m+\frac{1}{2}\right)\left(l'+m+\frac{3}{2}\right)}\left|l'+1,m+1\right>$$

So, in my notation, $C_{01}^{\left(l',m\right),\left(l',m\right)}=\frac{i}{2}\left(l'+m\right)$. $J_{02}$ and $J_{12}$ split a state labeled by $(l',m)$ amongst $(l'-1,m-1)$, $(l',m)$, and $(l'+1,m+1)$, if that is of any importance. These generators satisfy the commutation relations:

\begin{array}{c} \left[J_{01},J_{02}\right]=iJ_{12} \\\ \left[J_{01},J_{12}\right]=iJ_{02} \\\ \left[J_{02},J_{12}\right]=-iJ_{01} \end{array}

Which satisfy $[J_{\mu \nu },J_{\rho \sigma }]=i(\eta_{\nu \rho }J_{\mu \sigma }+\eta_{\mu \sigma }J_{\nu \rho }-\eta_{\mu \rho }J_{\nu \sigma }-\eta_{\nu \sigma }J_{\mu \rho})$ for $\eta$ with the metric $(-,+,+)$.

Question: How can I find the invariant quadratic form, $-x_0^2+x_1^2+x_2^2$, in terms of the coordinates $x$, $y$, and $z$ and possibly other known quantities like momentum or the Pauli matrices from just this information?

Additionally, from this information, can I write the operators $J_{\mu\nu}$ as differential operators?

Further, I know the Hamiltonian of this system, the closed forms of all the basis states, $\left|l',m\right>$, and the larger symmetry group $SO(3,2)\supset SO(2,1)$.

Follow up: The $e_{ij}$ that I found are: \begin{array}{ccc} e_{00}=\mathbf{1}-J_{12}J_{12} & e_{01}=-J_{02}J_{12} & e_{02}=J_{01}J_{12}\\\ e_{10}=J_{12}J_{02} & e_{11}=\mathbf{1}+J_{02}J_{02} & e_{12}=-J_{01}{J_02}\\\ e_{20}=-J_{12}J_{01} & e_{21}=-J_{02}J_{01} & e_{22}=\mathbf{1}+J_{01}J_{01} \end{array}

And if each of these is $e_{ij}=Q_iP_j$, I get 9 equations of the form:

$$f(x_0,x_1,x_2)-J_{12}J_{12}f(x_0,x_1,x_2)=x_0\frac{\partial}{\partial x_0}f(x_0,x_1,x_2)$$

$\endgroup$

1 Answer 1

4
$\begingroup$

So, first, the Lie algebra $\mathfrak{so}(2,1)$ is isomorphic to $\mathfrak{sl}(2)$. For example $$ J_{01} = \frac{i}{2} \begin{pmatrix} 1&0\\0&-1 \end{pmatrix}\quad J_{02} = \frac{i}{2} \begin{pmatrix} 0&1\\1&0 \end{pmatrix}\quad J_{12} = \frac{i}{2} \begin{pmatrix} 0&1\\-1&0 \end{pmatrix}\quad $$ have the same commutation relations and they span $\mathfrak{sl}(2)$ as a linear space.

Second, for any representation $\pi$ of a Lie algebra, the invariant symmetric bilinear form is $$B(X,Y)=\operatorname{trace}(\pi(X)\pi(Y))$$ where $X, Y$ are in the Lie algebra and the trace is computed on linear operators of the space the Lie algebra is acting on.

Third, if $v_1, \ldots v_k$ is a basis of the space, there is a corresponding dual-basis of linear functionals $v_1^*, \dots, v_k^*$ with the property $v_i^*(v_j) = \delta_{ij}$. Using a basis and dual-basis, the trace is computed as $$\operatorname{trace}(\pi(X)\pi(Y)) = \sum_{i=1}^k v_i^*(\pi(X)\pi(Y)v_i)$$

About the differential operators - You have the associative Weyl algebra which is generated by linearly-independent generators $Q_1,\ldots,Q_l,P_1,\ldots,P_l$ with the relations $$[P_i, Q_j] = \delta_{ij}\mathbf{1}\quad [P_i,P_j]=0\quad [Q_i,Q_j]=0\tag{1}$$ where $\mathbf{1}$ is the unit of the algebra. This associative algebra can realized as operators on functions $f(x_1,\ldots,x_l)$ via $$ \begin{align} Q_j f(x_1,\ldots,x_l)&:= x_j f(x_1,\ldots,x_l)\\ P_j f(x_1,\ldots,x_l)&:= \frac{\partial}{\partial x_j} f(x_1,\ldots,x_l)\\ \mathbf{1}f(x_1,\ldots,x_l)&:=f(x_1,\ldots,x_l) \end{align} \tag{2} $$ Now, from the commutation relations $(1)$ you get $$ \begin{align} [Q_i P_j, Q_k P_l] &= [Q_i P_j, Q_k] P_l + Q_k[Q_i P_j, P_l]\\ &= [Q_i,Q_k]P_j P_l + Q_i[P_j,Q_k]P_l + Q_k[Q_i,P_l]P_j + Q_k Q_i[P_j, P_l]\\ &= \delta_{jk} Q_i P_l - \delta_{il} Q_k P_j\tag{3} \end{align} $$

Turning for a moment to the matrix Lie algebra $\mathfrak{gl}(n)$, it has a basis $e_{ij}$ of matrices with $1$ in the ($i$-th row, $j$-th column) and zero elsewhere that satisfy $$[e_{ij}, e_{kl}] = \delta_{ij} e_{il} - \delta_{il} e_{kj}\tag{4}$$

Comparing $(3)$ and $(4)$ we see that the matrix Lie algebra is embedded inside the Weyl algebra via the mapping $e_{ij}\mapsto Q_i P_j$. Furthermore, the $Q_i P_j$ can be realized as differential operators via $(2)$. So if you have any Lie algebra spanned by matrices whatsoever, you can turn it into a Lie algebra of differential operators by mapping linearly the basis $e_{ij}$ to $Q_i P_j$ to the differential operators. If this sounds strange, when carrying out this process on the Lie algebra $\mathfrak{so}(3)$, the Lie algebra of rotations in $3$-space, you get the differential operators of angular momentum acting on functions (up to scaling by $i\hbar$) - and these in turn have spherical harmonics as eigenfunctions.

Edit: For differential operators try $$ \begin{align} J_{01}&= i \left(x_0\frac{\partial}{\partial x_1} + x_1\frac{\partial}{\partial x_0}\right)\\ J_{02}&= i \left(x_0\frac{\partial}{\partial x_2} + x_2\frac{\partial}{\partial x_0}\right)\\ J_{12}&= i \left(x_1\frac{\partial}{\partial x_2} - x_2\frac{\partial}{\partial x_1}\right) \end{align} $$ These have the same commutation relations as the $J$-s in the original post.

For the invariant symmetric bilinear form, in an irreducible representation it's always going to be $$B(J_{01},J_{01})=B(J_{02},J_{02})=-\alpha\quad B(J_{12},J_{12})=\alpha$$ with other pairs zero for some $\alpha>0$ that depends on the dimension of the irreducible representation. For a $2$-dimensional representation, $\alpha=1/2$. For a $3$-dimensional representation, $\alpha=2$.

I can't tell from the notation in the original post what dimension representation is talked about and whether it's irreducible. It looks like it's a representation that's a tensor product of two irreducible ones decomposed according to Clebsch-Gordan.

$\endgroup$
2
  • $\begingroup$ Thanks for the response. I don't really follow the first half of the solution, but if I understand the second half then I am left with 9 equations (see the edit I made to the question). But if I don't know the $x_i$, how am I supposed to solve these differential equations for $f$? I imagine this is similar to how we relate the $D$-functions to the spherical harmonics in the case of $SO(3)$, but in that case, we have an operator form of the generators. $\endgroup$
    – QPhysl
    Commented Dec 29, 2023 at 21:18
  • 1
    $\begingroup$ @QPhysl: See edit. $\endgroup$
    – Chad K
    Commented Dec 30, 2023 at 9:36

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.