Superposition formulas for exterior differential systems (Q1028351)

This paper is an advance in the development of explicit methods to integrate systems of partial differential equations, including a wide variety of determined and overdetermined systems. The paper identifies a class of exterior differential system which is explicitly integrable by a generalization of the method of Darboux. The authors apply their method to the Toda lattice, low-dimensional wave maps and some overdetermined examples, achieving explicit formulas for the general solutions. The authors characterize which Pfaffian systems arise by a quotienting process from a direct sum of two Pfaffian systems. (This direct sum is related to the notion of superposition of solutions.) If a Pfaffian system is a quotient in this sense, then the authors recover the two Pfaffian systems from purely local calculations. They can then find the general solution of any such system. Their examples are explicit, clear, and very different from one another, demonstrating the very broad reach of their method. The authors refer (footnote 2, p. 1914) to a constant rank exterior differential system. An exterior differential system \({\mathcal I}\) is said to be of constant rank precisely when it is the collection of all local smooth sections of a sum of subbundles \(I^* \subset \Omega^*\) of the various exterior powers of the cotangent bundle. This reviewer at first glance thought that a constant rank Pfaffian system just meant a vector subbundle of the cotangent bundle, but in fact the authors use the term constant rank of a Pfaffian system to mean that the exterior differential system generated by the local sections of the Pfaffian system must be of constant rank in the sense above. There is an unstated hypothesis (on p. 1912) that every Pfaffian system \(V \subset T^*M\) arising in the construction contains a unique minimal bracket-closed subbundle \(V^{(\infty)}\). This is apparently true for constant rank Pfaffian systems in the sense above. It is not enough that the Pfaffian system have constant rank as a subset of the cotangent bundle, i.e. that the Pfaffian system is a subbundle of the cotangent bundle. The authors slightly misquote Griffiths and Sternberg (p. 1939); the relevant statements on integration of Lie equations require passing to a covering space. The arXiv version of the article looks quite different from the published version. In particular, the figure on p. 57 of the arXiv preprint is not visible. The authors are persistent in making global statements which are not correct strictly speaking, but can presumably be corrected by saying ``at least locally''. They often omit such qualifications. I found it confusing at times to see where to put these qualifications. Their notation allows a expression like \(f : X \to Y\) to mean that \(f\) is a smooth map defined on a nonempty open subset of a manifold \(X\), mapping to a manifold \(Y\). With these minor quibbles in mind, the reader should enjoy the article, which is clearly written and displays a profound new understanding of the Darboux method of integration.