Topology on coalgebras (Q698519)

This paper defines coprime subcoalgebras and constructs a topology on them. Let \(C\) be a coalgebra over a field \(K\). A subcoalgebra \(D\) of \(C\) is said to be conilpotent if \(\text{corad}(C)\subseteq D\). A nonzero subcoalgebra \(P\) of \(C\) is called coprime if \(P\subseteq X\wedge Y\) implies \(P\subseteq X\) or \(P\subseteq Y\) for any subcoalgebras \(X\) and \(Y\) of \(C\). It is proved that a subcoalgebra \(P\) of \(C\) is coprime if and only if \(P^\perp\) is a prime ideal of the dual algebra \(C^*\). Let \(C\) be a coalgebra and let \(X\) be the set of coprime subcoalgebras of \(C\). For any subcoalgebra \(E\) of \(C\), let \(V(E)=\{P\in X\mid P\subseteq E\}\), \(X_E=X-V(E)\). Let \(\tau=\{X_E\mid E\) is a subcoalgebra of \(C\}\). Then \((X,\tau)\) is a topological space with closed sets \(V(E)\) (or open sets \(X_E=X-V(E)\)). It is shown that the topological space \(X\) is compact (Lindelöf) if \(C\) is irreducible or the number of simple subcoalgebras of \(C\) is finite (countable). If the topological space \(X\) is Hausdorff then every coprime subcoalgebra of \(C\) is simple. If \(P\) is a coprime subcoalgebra of \(C\) then \(Y=V(P)\) is an irreducible subspace of the topological space \(X\). If \(P\) is a coprime subcoalgebra of \(C\) and \(Y=V(P)\) is an irreducible component then \(P\) is a maximal coprime coalgebra of \(C\). This paper also discusses some other properties of the topological space \(X\). There are some gaps in the paper. For any (nonzero) coalgebra \(C\), the coradical \(\text{corad}(C)\) of \(C\) always exists, which is conilpotent. Hence Proposition 2.8 and Corollary 2.4 are absurd. Theorem 2.4(i) is equivalent to that if \(f\colon A\to B\) is an algebra map and \(I\) is a prime ideal of \(B\) then \(f^{-1}(I)\) is a prime ideal of \(A\) (see the proof). However, it is not true. For example, let \(M_n(K)\) be the algebra of all \(n\times n\) matrices over a field \(K\) and let \(T_n(K)\) be the algebra of all upper-triangular \(n\times n\) matrices over \(K\). Let \(n>1\). Then the natural embedding \(f\colon T_n(K)\hookrightarrow M_n(K)\) is an algebra map and \(I=\{0\}\) is a prime ideal of \(M_n(K)\), but \(f^{-1}(I)=\{0\}\) is not a prime ideal of \(T_n(K)\).