1
$\begingroup$

Let $V$ be a vector space with a basis $v_1, v_2, \ldots, v_n$. Let $T(V)$ be the tensor algebra of $V$. Let $S(Lie(V))$ be the symmetric algebra of the free Lie algebra of $V$. I think that $T(V)$ is isomorphic to $S(Lie(V))$ as a graded vector space. For example, let $n=2$. We have the degree $2$ component of $T(V)$ is \begin{align} T(V)_2 = Span\{v_1 \otimes v_1, v_1 \otimes v_2, v_2 \otimes v_1, v_2 \otimes v_2\}. \end{align} The degree 2 component of $S(Lie(V))$ is \begin{align} S(Lie(V))_2 = Span\{v_1^2, v_2^2, v_1 v_2, [v_1, v_2]\}. \end{align} A bijection from $T(V)_2$ to $S(Lie(V))_2$ is \begin{align} & v_1 \otimes v_1 \mapsto v_1^2, \\ & v_1 \otimes v_2 \mapsto v_1 v_2, \\ & v_2 \otimes v_1 \mapsto [v_1, v_2], \\ & v_2 \otimes v_2 \mapsto v_2^2. \end{align}

The degree $3$ component of $T(V)$ is \begin{align} & T(V)_3 = Span\{v_1 \otimes v_1 \otimes v_1, v_1 \otimes v_1 \otimes v_2, v_1 \otimes v_2 \otimes v_1, v_2 \otimes v_1 \otimes v_1, \\ & \quad \quad \quad \quad \quad v_1 \otimes v_2 \otimes v_2, v_2 \otimes v_1 \otimes v_2, v_2 \otimes v_2 \otimes v_1, v_2 \otimes v_2 \otimes v_2 \}. \end{align} The degree 2 component of $S(Lie(V))$ is \begin{align} S(Lie(V))_2 = Span\{v_1^3, v_2^3, v_1^2 v_2, v_1 v_2^2, v_1 [v_1, v_2], v_2[v_1,v_2], [[v_1,v_2],v_2], [[v_1,v_2],v_1]\}. \end{align} A bijection from $T(V)_3$ to $S(Lie(V))_3$ is \begin{align} & v_1 \otimes v_1 \otimes v_1 \mapsto v_1^3, \\ & v_2 \otimes v_2 \otimes v_2 \mapsto v_2^3, \\ & v_1 \otimes v_2 \otimes v_2 \mapsto v_1 v_2^2, \\ & v_1 \otimes v_1 \otimes v_2 \mapsto v_1^2 v_2, \\ & v_1 \otimes v_2 \otimes v_1 \mapsto v_1 [v_1, v_2], \\ & v_2 \otimes v_1 \otimes v_2 \mapsto [v_1, v_2]v_2, \\ & v_2 \otimes v_2 \otimes v_1 \mapsto [[v_1,v_2],v_2], \\ & v_2 \otimes v_1 \otimes v_1 \mapsto [[v_1, v_2], v_1]. \end{align} How to write down the general formula $T(V)_n \to S(Lie(V))_n$?

$\endgroup$
  • 2
    $\begingroup$ This is the Poincar\'e--Birkhoff--Witt theorem. Look it up in Jean-Pierre Serre's book on Lie algebras and Lie groups (LNM 1500), or in Christophe Reutenauer's book on Free Lie algebras. $\endgroup$ – Alex Suciu Feb 26 '17 at 16:30
6
$\begingroup$

The natural map is rather from $TV \to U(FreeLie(V))$: consider the forgetful functors $Assoc \to Lie \to Vec$ and compose their left adjoints to get the left adjoint $T$ of the composite. Then, as Alex Suciu comments, your question is a special case of the comparison of $U\mathfrak g$ and $S\mathfrak g$, accomplished by PBW.

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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