Higher Lie idempotents (Q1969473)

The Milnor-Moore theorem says that in characteristic zero, any connected graded cocommutative bialgebra \(A\) is canonically isomorphic to the enveloping bialgebra of the Lie algebra of its primitive elements \(\text{Prim}(A)\). There is a weaker form known as the Leray theorem, whose dual statement is that any retract of the vector space inclusion of \(\text{Prim}(A)\) into the augmentation ideal of \(A\) induces an isomorphism between the cofree cocommutative coalgebra over \(\text{Prim}(A)\) and \(A\). When \(A\) is the tensor bialgebra \(T(X)\) on a set \(X\), such a retract is called a Lie idempotent (when it can be computed in the group algebras of the symmetric groups, naturally acting on \(T(X)\)). This paper studies the combinatorics of these Lie idempotents. Using their convolution products, they construct and study families of Lie idempotents, whose properties yield an effective proof of the dual Leray theorem. They define three families of Lie idempotents, one of which describes the combinatorics underlying the dual Leray theorem. Their approach emphasizes the combinatorics of the tensor bialgebra.