INVERTIBLE IDEALS AND NON-COMMUTATIVE ARITHMETICS (Q2777960)

An ideal \(I\) of a ring \(R\) is said to be full if it contains an invertible ideal of \(R\). The author works with the quotient ring \(S\) of \(R\) which is, by definition, the natural minimal overring of \(R\) which contains inverses for all the invertible ideals of \(R\). It is important to note that this quotient ring of \(R\) always exists, even if \(R\) does not have a full classical quotient ring. If \(R\) is Noetherian then every invertible ideal of \(R\) contains a regular element, and so in this case \(S\) would coincide with the full classical quotient ring of \(R\) if the latter exists. The arithmetic \(A(R)\) of \(R\) is the multiplicative semigroup of all \(R\)-\(R\)-subbimodules \(I\) of \(S\) such that \(P\subseteq I\subseteq P^{-1}\) for some invertible ideal \(P\) of \(R\). Various properties of \(R\) can be defined in terms of \(A(R)\). In particular \(R\) is said to be an Asano ring if every full ideal of \(R\) is invertible, or equivalently if \(A(R)\) is a group. Also \(R\) is said to be maximal if \(R\) has no proper overring which belongs to \(A(R)\). Then \(R\) is an Asano ring if and only if \(R\) is maximal and every element of A(R) is left and right \(R\)-projective. Another characterisation is that \(R\) is an Asano ring if and only if \(R\) is maximal, satisfies the ascending chain condition for full ideals, and every full prime ideal of \(R\) is maximal. A decomposition theorem is given for Noetherian rings in which every maximal ideal is invertible, where each direct summand in the decomposition either has no proper invertible ideals or is a prime Asano order. In terms of localisation, it is shown that a Noetherian ring \(R\) is Asano if and only if every full maximal ideal \(M\) of \(R\) is localisable and the corresponding local ring \(R_M\) is hereditary. This paper extends many known results about Asano orders from the prime case to general rings \(R\) without assuming that \(R\) has a full classical quotient ring and, unlike in the work of Lissaman, without assuming that \(R\) is fully bounded.