Unique comaximal factorization (Q1883054)

This interesting paper introduces a new factorization theory, linked to comaximality. The authors give the following definitions: A nonzero non-unit element \(b\) of an integral domain \(R\) is called pseudo-irreducible if it is impossible to factor \(b\) as \(b=cd\) with \(c\) and \(d\) comaximal non-units. Now \(b\) is termed pseudo-prime if for every comaximal pair of elements \(c\) and \(d\) in \(R\), if \(b\) divides \(cd\), then either \(b\) divides \(c\) or \(b\) divides \(d\). The ring \(R\) is called a comaximal factorization domain (CFD) if any nonzero non-unit element \(b\) of \(R\) can be written as \(b=b_1\cdots b_m\), where the \(b_i\) are pairwise comaximal pseudo-irreducible elements of \(R\), and a unique comaximal factorization domain (UCFD) if \(R\) is a CFD where such factorizations are unique (up to order and units). These notions are the analogues of irreducible and prime elements, atomic domains and UFD's in the theory of usual factorization. In particular, a UFD is a UCFD. These two theories have many similarities. For instance, a CFD is a UCFD if and only if every pseudo-irreducible element is pseudo-prime and also (a slight difference with UFD's) if and only if every 2-generated invertible ideal is principal. But an example shows that there exist UCFD's for which not all invertible ideals are principal. Another similar property is that a polynomial domain \(R[X_1,\ldots,X_n]\) is a UCFD if and only if \(R\) is a seminormal UCFD. Contrary to UFD's, a UCFD is not necessarily integrally closed. Pullbacks can be used to produce CFD's or UCFD's by the following theorem: Let \(P\) be a prime ideal of a domain \(R\), comparable to every ideal of \(R\), then \(R\) is a CFD (respectively UCFD) if and only if \(R/P\) is a CFD (respectively UCFD). In the last section, the authors introduce the ideal-theoretic version of comaximal factorization, which is better behaved than comaximal factorization of elements.