Gronwall's conjecture for 3-webs with two pencils of lines (Q6123249)

The Gronwall hypothesis (1912) is considered for linearized flat three-webs of a special type, in which two families of lines are pencils of straight lines. That is, the author proves that the mapping of a web of the specified type into a rectilinear three-web up to a projective transformation is the only one. The proof uses a complex construction. A family of torsion-free projective connections is attached to the web, with respect to which the web lines are geodesic. Using the Darboux derivative, a system of differential equations for web invariants is constructed. In the 9-dimensional space of second-order projective differential invariants, 3-web invariants parametrize some 2-dimensional surface, which the author calls a set of web signatures. The linearization condition for non-hexagonal webs of the specified type is formulated in terms of invariants. The analysis of differential equations for maps preserving the linearity of webs of the type under consideration leads to the main result of the article. Maple is essentially used for calculations. Reviewer's comment: The author notes the contradiction in articles [\textit{J. Grifone} et al., Nonlinear Anal., Theory Methods Appl., Ser. A, Theory Methods 47, No. 4, 2643--2654 (2001; Zbl 1042.93526)] and [\textit{V. V. Goldberg} and \textit{V. V. Lychagin}, J. Geom. Anal. 16, No. 1, 69--115 (2006; Zbl 1104.53011)]: ``the authors of [\textit{V. V. Goldberg} and \textit{V. V. Lychagin}, J. Geom. Anal. 16, No. 1, 69--115 (2006; Zbl 1104.53011)] claim that the main example of [\textit{J. Grifone} et al., Nonlinear Anal., Theory Methods Appl., Ser. A, Theory Methods 47, No. 4, 2643--2654 (2001; Zbl 1042.93526)] is not linearizable, whereas this example possesses an explicit linearization in elementary functions!'' There is a short proof (due to J.-P. Dufour) that the three-web \(W_0\) formed by the foliations \(x=const, y=\mathrm{const}\) and \(z=(x+y)e^{-x}=\mathrm{const} \) is equivalent to a rectilinear three-web, and the corresponding diffeomorphism is given by the equations \(\tilde x=z, \tilde y = y.\) On the other hand, in 2006 V. Goldberg and V. Lychagin established that the web \(W_0\) is not linearizable. The most interesting thing is that both conclusions are correct. The authors of these works used different equivalence relations. The classical (local) web theory, starting with W. Blaschke, uses only isotopic transformations, that is, transformations of the form \(x\rightarrow\tilde x, y\rightarrow\tilde y, z\rightarrow\tilde z\), which do not change the order of web foliations. The authors of the work [\textit{V. V. Goldberg} and \textit{V. V. Lychagin}, J. Geom. Anal. 16, No. 1, 69--115 (2006; Zbl 1104.53011)] used this classical equivalence relation, whereas Dufour's proof uses the substitution \(\tilde x=z\), that is, the so-called parastrophy. This transformation swaps the foliations of the web. Thus, Dufour considers the web \(W_0\) with respect to a broader equivalence relation.