Homology of configuration spaces of hard squares in a rectangle (Q6148482)

The present paper studies configuration spaces of unit squares inside a rectangle. When there are sufficiently few squares relative to the dimensions of the rectangle the homology of these configuration spaces is the same as for the well-studied configuration spaces of points on the plane. However as the density of the squares increases they begin to constrain each other, until there are so many squares that the homology becomes trivial. The authors study here the boundary between trivial and non-trivial homology (this boundary is different in different homology degrees). More formally, they define \(C(n; p, q)\) the configuration space of \(n\) unit squares inside a rectangle of dimensions \(p\times q\) and study necessary conditions for \(H_j[C(n; p, q)]\) to be nonzero. Their first main theorem is a list of three inequalities relating \(j\) to the other 3 variables \(n,p,q\), which each imply that the homology vanishes. Normalizing with the new variables \(x = n/pq\), \(y = j/pq\), these inequalities describe a region of the \(xy\)-plane such that any point outside must give trivial homology. Their second main theorem describes a smaller region of the \(xy\)-plane such that for any rational point \((x,y)\) inside this region there exist corresponding values of \(j,n,p,q\) such that \(H_j[C(n;p,q)]\) does not vanish, and they conjecture that this region is optimal. An important part of the argument is a description of a cubical complex \(X(n;p,q)\) which they prove to be a deformation retract of \(C(n;p,q)\). They can then study this complex via discrete Morse theory to obtain the inequalities of the first theorem. At the end of the paper they give a table of Betti numbers for small values of \(n, p,q\) and explain how they were computed.