Given a Lie Algebra (such as $su(n), so(n))$ can I always find a set of generators + identity $\{T^a\}\cup \{id\}$ such that there exists a $c$ for any given $a,b$ such that $T^a T^b = C(a,b) T^c $ for a $a,b$-dependent function $C$ into the real/complex numbers?
Asked
Active
Viewed 47 times
1
-
What does $\propto$ stand for. – José Carlos Santos Dec 19 '18 at 13:44
-
1@JoséCarlosSantos: "proportional to". So OP is asking whether there exist constants $k^{ab}_c$ such that $T^aT^b=k^{ab}_c T^c$ (with no Einstein summation, just a single constant per equation) – ziggurism Dec 19 '18 at 13:51
-
are $\mathfrak{su}(n)$ or $\mathfrak{so}(n)$ even closed under multiplication? – ziggurism Dec 19 '18 at 13:56
-
sorry, let's add ${ id }$ to the set of generators – dan-ros Dec 19 '18 at 13:58
-
1Why would such identities be of interest? A representation of the Lie algebra need not preserve such product relations, only commutators. – Jyrki Lahtonen Dec 19 '18 at 17:26
1 Answers
1
The answer is a quick no. See this question. Your left-hand-side is in the universal enveloping algebra, not the Lie algebra itself.
You could have spared yourself the question, if you had considered the (presumed) preamble of your text, that surely reminded you of the spin-1, so, adjoint, representation of su(2), namely 3×3 traceless Hermitian matrices, hence a trivial counterexample: $$ S_z=\operatorname {diag} ~ (1,0,-1), \qquad \Longrightarrow \qquad S_z S_z = \operatorname {diag} (1,0,1). $$ Now, this square cannot be written as a linear combination of $S_z$, the identity, which you allowed, and of course $S_x,S_y$ with zero in their diagonals.
Cosmas Zachos
- 6,028