Tropical plane geometric constructions: a transfer technique in tropical geometry (Q533385)

Tropical geometry is a shadow of classical algebraic geometry. This is so because there exists a projection map, called tropicalization, from any given \(n\)--dimensional space (affine or projective, over a non--archimedean rank one valued field \(K\)) onto a corresponding \(n\)--dimensional tropical space. Therefore it makes sense to investigate the possibility of transferring results from one geometry to the other. This paper provides conditions (sometimes necessary, sometimes sufficient) for this transfer to happen. Let us think of some classical true statements concerning incidence of plane algebraic curves and points, such as Pappus theorem, Desargues theorem, Pascal hexagon theorem, Fano theorem, Chasles theorem (if two cubic curves intersect in 9 different points and a third cubic passes through eight of those points, then the third cubic also passes through the ninth point), Cayley--Bacharach theorem or others. We wonder if such statements are true on a tropical plane. More precisely, we wonder about the validity of the following method: consider the tropical data (curves and or points), lift them to a classical plane, find the solution in the classical setting and then project the solution to the tropical plane. Is this a solution to the tropical question we started with? If so, are all tropical solutions obtained this way? The author addresses the former questions in this paper, setting up the language (geometric construction, admissible geometric construction, pseudodeterminant, constructible incidence theorem etc.) and ingredients necessary for comparison of incidence configurations int the classical and tropical planes. With these tools, he is able to prove many positive results and provide a bunch of illuminating counterexamples. He discovers that only some versions of some classical theorems can be proved in the tropical setting. Moreover, that these versions must be a careful, step by step process setting up the hypothesis (inputs ---points or lines--- plus the construction of additional points and lines) and ending with the thesis node (point or line), where no point or line is used twice. Another trend of thought is the following. Let us think of a question in classical geometry that we do not know how to solve. Pass this problem to the tropical world via projection. Assume we can solve the tropical problem and lift the solution. Is this a solution to the question we started with? If so, are all solutions obtained this way? This procedure has been successfully applied by Mikhalkin and other authors. The paper is rather long, however worth reading.