Interpolation of Caldéron and Ovčinnikov pairs (Q1078483)

This is not an easy paper to read so we make a special point of noting that the persistent reader will find a lot of information and interesting ideas in it. Calderón pairs are those pairs of Banach spaces for which the norms of all the interpolation spaces can be obtained by applying suitable lattice norms to the K-functional. One of the major results of the paper states essentially that the dual couple of a Calderón pair is again a Calderón pair. More generally we say that \(\bar A=(A_ 0,A_ 1)\) and \(\bar B=(B_ 0,B_ 1)\) are relative Calderón pairs if all relative interpolation spaces A and B with respect to these couples are characterized by the property: If \(a\in A\) and \(b\in B_ 0+B_ 1\) satisfy \(K(t,b;B_ 0,B_ 1)\leq K(t,a;A_ 0,A_ 1)\) for all \(t>0\), then \(b\in B.\) It is well known that \(\bar L^ 1=(L^ 1_{w_ 0},L^ 1_{w_ 1})\) and \(\bar B\) are relative Calderón pairs for ''almost'' every choice of pair \(\bar B\) and all choices of weight functions \(w_ j\). Similarly \(\bar A\) and \(\bar L^{\infty}=(L^{\infty}_{w_ 0},L^{\infty}_{w_ 1})\) are relative Calderón pairs for every \(\bar A.\) On the other hand \(\bar L^{\infty}\) and \(\bar L^ 1\) (we have written them in the reverse order now) fail in a very strong sense to be relative Calderón pairs and the author has introduced the notion of relative Ovčinnikov pairs for pairs which ''strongly fail'' in this sense to be relative Calderón. One question which still remains open from the author's study of these pairs is whether there are examples of Oc\v{v}innikov pairs which are essentially different from the ''canonical'' example \(\bar L^{\infty}\) and \(\bar L^ 1.\) One of the many results here which particularly caught this reviewer's attention is Proposition 4.8 which shows some interesting properties of the couples \({\mathcal J}\bar L^ 1=({\mathcal J}L^ 1_{\alpha_ 0},{\mathcal J}L^ 1_{\alpha_ 0})\) and \({\mathcal J}\bar L^{\infty}=({\mathcal J}L^{\infty}_{\alpha_ 0},{\mathcal J}L^{\infty}_{\alpha_ 0})\). (Here \({\mathcal J}L^ p_{\alpha}\) denotes the space of sequences \(\{\gamma_ n\}^{\infty}_{n=-\infty}\) for which \(\sum^{\infty}_{n=-\infty}2^{\alpha n}\gamma_ ne^{inx}\in L^ p[0,2\pi].)\) These couples play special roles in \textit{Svante Janson}'s study of the complex method [J. Funct. Anal. 44, 50-73 (1981; Zbl 0492.46059)] and elsewhere. They are an interesting source of several counterexamples. Implicit in Proposition 4.8 is the fact that both \({\mathcal J}\bar L^ 1\) and \({\mathcal J}\bar L^{\infty}\) are not Calderón pairs and furthermore \({\mathcal J}\bar L^ 1\) and \({\mathcal J}\bar L^{\infty}\) taken together do not constitute relative Calderón pairs, all this in complete contrast to the situation for the analogous couples with ''\({\mathcal J}''\).