An alternative approach to the structure theory of PI-rings. (Q533119): Difference between revisions

One of the key theorems in structure theory of PI-rings is the Posner theorem which in its strongest form states that a prime PI-ring has a two-sided classical ring of quotients which is a finite-dimensional central simple algebra and this ring of quotients is actually the algebra of central quotients. Its standard proof is a beautiful illustration of the power and applicability of the classical structure theory of rings. In the paper under review the author suggests another proof of the Posner theorem based on fundamental properties of the extended centroid of a prime ring and in each of its steps avoids representing elements in the rings as matrices or linear operators. Instead, the main ingredients of the standard proof of the Posner theorem are replaced by a single theorem describing one of the basic properties of the extended centroid of a prime ring. This theorem is essentially due to Martindale, and is one of the cornerstones in the theory of generalized polynomial identities as well as in the theory of functional identities. It gives some kind of linear dependence over the extended centroid of the coefficients of a generalized identity in one variable for an ideal of a prime ring. Then the author proves a weaker version (for central simple algebras only) of the Kaplansky theorem that a primitive PI-algebra is finite dimensional over its center and which is an intermediate step to the final goal. Since the proof avoids using the existence of central polynomials on matrix algebras, it also provides a new proof of the existence.