Monadic decompositions and classical Lie theory (Q2254604)

Let \(Bialg_{\mathbb{K}}\) be the category of bialgebras and \(\Omega : Bialg_{\mathbb{K}} \to Vect_{\mathbb{K}}\) be the forgetful functor. Let \(P: Bialg_{\mathbb{K}} \to Vect_{\mathbb{K}}\) be the subfunctor of \(\Omega\) that assigns to a bialgebra \(A\) its space \(PA\) of primitive elements. The paper is devoted to prove that \(P\) has a monadic decomposition of monadic length at most 2. Namely, the main result establishes: { Definition.} Fix \(N \in \mathbb{N}\). It is said that a functor \(R\) has a \textit{monadic decomposition of monadic length} \(N\) whenever there exists a sequence \((R_n)_{n \leq N}\) of functors \(R_n\) such that (1) \(R_0=R\); (2) for \(0 \leq n \leq N\), the functor \(R_n\) has a left adjoint functor \(L_n\); (3) for \(0 \leq n \leq N-1\), the functor \(R_{n+1}\) is the comparison functor induced by the adjunction \((L_n, R_n)\) with respect to its associated monad; (4) \(L_N\) is full and faithful while \(L_n\) is not full and faithful for \(0 \leq n \leq N-1\). { Theorem.} The functor \(P\) has a monadic decomposition of monadic length at most 2. Keep the notations of Definition (so, in particular, \({\mathcal B}_0 = Vect_{\mathbb{K}}, R_0 = P\), and \(L_0=T\)): (1) The functor \(L_1\) is given, for all \((V_0, \mu_0) \in {\mathcal B}_1\), by \[ L_1(V_0, \mu_0) = \frac{L_0 V_0}{\left( \text{Im} \left( \text{Id}_{R_0 L_0 V_0} - \eta_0 V_0 \circ \mu_0 \right) \right)} \] (2) The adjunction \((L_1, R_1)\) is idempotent. (3) For all \(V_2 := ((V_0, \mu_0), \mu_1) \in {\mathcal B}_2\), we have the following cases: \(\bullet\) char\((\mathbb{K}) = 0\). Then, for all \(x, y \in V_0\), we have that \(xy-yx \in R_0L_0V_0\). Define a map \([-,-]: V_0 \otimes V_0 \to V_0\), by setting \([x,y]:= \mu_0(xy-yx)\). Then \((V_0,[-,-])\) is an ordinary Lie algebra and \(L_2 V_2\) is the universal enveloping algebra \[ {\mathfrak U} V_0 := \frac{TV_0}{\left(xy-yx-[x,y] \mid x, y \in V_0 \right)} \] \(\bullet\) char\((\mathbb{K}) = p\), a prime. Then, for all \(x, y \in V_0\) we have that \(xy-yx, x^p \in R_0 L_0 V_0\). Define two maps \([-,-]: V_0 \otimes V_0 \to V_0\) and \(-^{[p]}:V_0 \to V_0\) by setting \([x,y]:= \mu_0(xy-yx)\) and \(x^{[p]}:=\mu_0(x^p)\). Then \((V_0,[-,-], -^{[p]})\) is a restricted Lie algebra and \(L_2 V_2\) is the restricted enveloping algebra \[ {\mathfrak u} V_0 := \frac{TV_0}{\left(xy-yx-[x,y], x^p-x^{[p]} \mid x, y \in V_0 \right)}. \]