The ideal of relations for the ring of invariants of \(n\) points on the line (Q653287)

This is the main paper in a marvelous series by these four authors presenting a modern take on a problem in classical invariant theory and concluding a story first began by \textit{A. B. Kempe} [Lond. M. S. Proc. XXV. 343--359 (1894; JFM 25.0235.01)] some 115 years earlier. The problem can be stated in a few equivalent ways: 1) describe the ideal of relations among a certain collection of tableau functions, 2) provide equations for the natural projective embedding of the GIT moduli space of \(n\) ordered points on \(\mathbb{P}^1\), 3) give a presentation for a certain \(\mathrm{SL}(2)\)-invariant subring. More precisely, the authors use the generators for the homogeneous coordinate ring of the GIT quotient \[ (\mathbb{P}^1)^n/\!/\mathrm{SL}(2) = \mathrm{Proj} (\bigoplus_{k \geq 0}H^0((\mathbb{P}^1)^n,\mathcal{O}(k,\dots,k)))^{\mathrm{SL}(2)} \] found by Kempe, which are tableau functions obtained as \(2\times 2\) minors of the matrix of homogeneous coordinates for \((\mathbb{P}^1)^n\), and they describe the ideal of relations between these Kempe generators---and hence the equations cutting out this GIT quotient in its natural projective embedding. The main result of the present paper is that for \(6 \neq n\in 2\mathbb{Z}\), this ideal of relations is generated by a collection of easy-to-describe binomial quadrics -- and that up to the action of the symmetric group \(S_n\) only one such is necessary. Moreover, these quadrics are essentially pulled-back from the single case of \(n=8\). This is a truly remarkable result and elegant conclusion to Kempe's centenarian tale. The authors' earlier Duke paper on this subject proved quadric generation of the relations, but not such a clean and simple form as this. That paper also shows how to reduce the case of arbitrary weights/linearization to this democratic setting, so in fact the main result of this paper is even more general than as stated above. The main ingredients in the paper are the following. First of all, they prefer to work with a ``graphical'' (i.e., graph-theoretic) description of the invariants, as opposed to the tradition tableau function perspective. This seems to be only a superficial change of approach, but it actually helps matters significantly as the notation, permutation symmetry, and combinatorial proofs often become more transparent in this way. Second, a key technical workhorse for the authors is a toric degeneration of the ring they are interested in. This allows them to study in a more brute-force but nonetheless tractable fashion various properties of the relations, such as degree bounds, which they then lift nicely to the original setting. The toric degenerations they use essentially come from the tropical geometry of the Grassmannian \(\mathrm{Gr}(2,n)\), and hence of the tropicalized moduli space of points on the line \(M_{0,n}^{\mathrm{trop}}\), as developed by Speyer and Sturmfels' in the former's PhD thesis. Similar degeneration techniques have been applied and generalized (independently of the present work) in various papers of Chris Manon that an interested reader is encouraged to peruse. Finally, used implicitly throughout and quite explicitly once this toric degeneration has bore its fruit is the representation theory of the symmetric group, and its commensurate combinatorics. The one unfortunately confusing aspect of this paper is its precise relation to the other very similarly named papers written by the same authors. In the introduction of the present paper they partially explain the chronology and interplay between the papers, and a few more details (such as arXival merging/replacement) can be discerned from Ravi's webpage if one needs such bibliographic details. I'll also mention in passing that Andrew Snowden has been developing some vast generalizations of certain results/ideas in this paper that an interested reader is encouraged to check out. Also, a natural question closely related to this paper though left untouched thus far is the situation for moduli of points in a projective space of dimension larger than one. Hopefully we won't have to wait another 115 years!