by Victor Vassiliev
Dmitry Borisovich served as a kind of Topology Minister of the Arnold seminar,1 responsible for this area of mathematics, and ran the discussion when topological questions arose — which happened frequently because of the ubiquity of Topology in Algebraic Geometry, Mechanics, and Dynamical Systems.
In these cases, he usually clarified the situation in a few words, and either completely closed the question (if it was comparatively easy) or characterized its genre and area, the presumed complexity, and methods by which it probably should be attacked. The number of problems solved by seminar participants based on these hints was apparently very large; however Fuchs himself never appeared as a coauthor — probably, considering this work too easy for himself.
Only in two or three cases, when he did almost all of the compound work, Fuchs wrote the corresponding articles (and, I believe, it was Arnold who insisted that he do it within a finite time).
In one of these works [2] he has calculated the mod 2 cohomology ring of the space of unordered configurations of \( d \) distinct points in the plane (and also related structures: Steenrod algebra action, stabilization as \( d \to \infty \), the Hopf algebra structure on the stable homology and cohomology, geometric realization of generators, etc). Arnold came to the consideration of this ring (or, rather, of the similar cohomology group with arbitrary coefficients) from his study of superpositions of algebraic functions; he made some first calculations, but asked Fuchs to continue them to the case of arbitrary values of \( d \) and dimensions of cohomology groups. Based on Fuchs’ results, Arnold proved some theorems on the absence of decompositions of algebraic functions into superpositions of easier functions. Since then this work (and its various generalizations) has seen numerous other applications: in geometric combinatorics, interpolation theory, complexity theory, etc.
The topological study of configuration spaces has developed since then into a huge and very instrumental area of topology, and the main methods and keywords first used by Fuchs in this pioneering work are now standard for these studies.
A closely related work is the calculation of cohomology rings of ordered configuration spaces of complex numbers, also initiated by Arnold. Accordingly to Arnold’s reminiscences, the first work of this theory [e2] also was written with huge help from Fuchs, although in this case he does not appear as an author.
The other work [1] is a clarification of the work by Maslov and Arnold on the topology of Lagrangian manifolds. In his 1967 work [e1], Arnold proved that the 1-cohomology class of these manifolds, considered previously by Maslov as the obstruction to the quasiclassic quantization, can be deduced by a kind of Gauss map from some universal space, the Lagrangian Grassmannian \( U/O \), thus behaving like a characteristic class of usual manifolds. This class was originally defined as Poincaré dual to the entire set of critical points of the Lagrange projection, but turned out to be an integer cohomology class unlike the first Stiefel–Whitney class defined in the same way for a generic map \( M^n \to {\mathbb R}^n \) of an ordinary manifold. Looking for generalizations of this work, Arnold posed the problem to study the cohomology classes dual to the sets, where the rank of Lagrangian projections drops by at least \( k \), in particular to decide whether these classes also are well-defined as integer cohomology classes. Then Fuchs came to the scene. Firstly, he explained that Lagrangian Grassmannians \( U(n)/O(n) \) are not a new object in topology: their cohomology rings were calculated by A. Borel in the 1950s, and the stable version of them appears as one of eight spaces in the real Bott periodicity. Further, he indeed realized the multiplicative generators of their weak integer cohomology rings in terms of sets of Lagrangian planes in a nongeneral position (thus providing an analogue of the Schubert cell realization of Pontryagin classes in the Lagrangian situation), and showed how the Arnold classes appear as their products.
Topological questions arose in the seminar from three main sources. Firstly, Arnold received many letters and reprints from colleagues from around the world. The new ideas and arguments in them were formulated in the terms standard for other scientific schools, thus demanding a translation and adaptation; sometimes they included some topological notions not standard for the seminar. Secondly, Arnold himself often excavated a new topological problem or approach from his studies in other areas. As a rule, in this case he solved or applied this consideration in the simplest (one-dimensional, say) example, and posed the problem to extend it to more advanced situations, requiring deep and special topological techniques. Finally, the topological issues appeared in the participant’s talks. In all these cases Fuchs provided us with complete, illuminating information. The most important in this service was not only his own top-level literacy in the topic (there are enough highly trained topologists of narrow focus!) but his ability to understand instantly the problem from a different area, and to recognize the topological heart of it.
Here are only two examples concerning myself (in fact there were more of them). Once Arnold asked me to understand and deliver in the seminar a talk on one of S. Smale’s works, in which the cohomology of braid groups was applied to the problems of computational complexity. So, I started to talk proudly, how I managed to improve Smale’s logarithmic lower bound to a linear one (with coefficient 0.5), but Fuchs said immediately: “Oh, this [Smale’s notion introduced for attacking this problem] is the A. S. Schwarz genus!” — and then explained where to read about it. I did it and improved the coefficient to almost 1 (which is also the upper bound), but it hardly would have been possible if I hadn’t also previously read Fuchs’ work [2]. In another case, I started to explain (equally proudly) how I managed to untie the homology groups of discriminant varieties to solve one of Arnold’s problem, and Fuchs reacted: “Oh, this trick is known to the science, it is called the simplicial resolution, and is conveniently realized by a geometric construction rather than this combinatorial one.” And indeed, the combination of the geometric and combinatorial views allowed me then to extend this approach to many other situations.
But the main impact of Fuchs’ work at the seminar was not the solution of concrete problems, nor the good advice. It was the demonstration of what modern topology actually is, its language and philosophy, the standard and most efficient tools, how unexpectedly much can be calculated and proved by these tools, which standard mistakes should be avoided, etc. Actually, the entire Arnold seminar served this purpose with respect to mathematics as a whole, but its topological direction was the service of Dmitry Borisovich.