Linkage classes of grade 3 perfect ideals (Q2301976)

Let \(R\) be a local ring. The difference between the embedding dimension of \(R\) and the depth of \(R\) is called the embedding codepth \(c\) of \(R\). Rings with \(c =0\) are regular, rings with \(c =1\) are hypersurfaces, and for rings with \(c=2\) there are two possibilities: complete intersection or Golod. For \(c =3\) the field of possibilities widens. Such rings can be Gorenstein and not complete intersection, or they may not even belong to any of the classes mentioned thus far. There is, nevertheless, a classification of local rings of embedding codepth 3 which is based on the multiplication in the homology. This classification is discrete; yet it is subtle enough to facilitate a proof of the rationality of Poincaré series of codepth 3 local rings. For this purpose it was not relevant to know if local rings of every class in the scheme actually exist, and it later turned out that they do not. In 2012 Avramov tightened the classification; that is, he limited the range of possible classes. This was necessary to use the classification to answer the codepth 3 case of a question ascribed to Huneke about growth in the minimal injective resolution of a local ring. In the same paper [J. Pure Appl. Algebra 216, No. 11, 2489--2506 (2012; Zbl 1259.13010)], \textit{L. L. Avramov} formally raised the question about realizability: Which classes of codepth 3 local rings do actually occur? The present paper limits the range of possible classes still further and conjectures that no further limits can be imposed on the range of possible classes. Along the way the present paper proves that every grade three perfect ideal in a regular local ring is in the linkage class of a complete intersection or is in the linkage class of an ideal which defines a Golod ring.