The largest subsemilattices of the endomorphism monoid of an independence algebra. (Q2250915)

An algebra \((A;F)\) is a matroid algebra if the closure operator `subalgebra generated by' \(\langle-\rangle\) satisfies that for all \(X\subseteq A\) and \(x,y\in A\): if \(x\in\langle X\cup\{y\}\rangle\) and \(x\notin X\) then \(y\in\langle X\cup\{x\}\rangle\). Every matroid algebra has a basis. A matroid algebra \((A;F)\) is an independence algebra if every mapping from a basis of \((A;F)\) can be extended to an endomorphism of \((A;F)\). It is proved that if \((A;F)\) is an independence algebra of a finite dimension \(n\) and \(|A|\geq 2\) then the largest subsemilattice of the endomorphism monoid of \((A;F)\) (a subsemilattice is a subsemigroup consisting of idempotent endomorphisms that commute) has size either \(2^{n-1}\) (if the clone of \((A;F)\) does not contain a constant operation) or \(2^n\) (if the clone of \((A;F)\) contains a constant operation). This result can be applied to linear operators of a finite dimensional vector space, to the monoid of all full transformations of a finite set \(X\), to the monoid of all partial transformations of a finite set \(X\), to the monoid of free \(G\)-sets with a finite set of free generators. Finally several open problems are presented.