The Setting
The graph is the incidence graph of 13 points and 22 faces, each face a triple of points: its vertices are the points and the faces, and each face is joined to its three points. The points of a face fill its slots in increasing order, , and a corner is a face with one of its slots . Two points of a common face form a point pair; there are 33 point pairs, and each lies in exactly two faces.
Put a Lagrangian subspace of a symplectic space over a finite field at each point , and a Lagrangian subspace , the center of , at each face . An activity pattern marks at each face a set of active slots, and the edges of from a face to the points in its active slots form the activity set . With the edge weighted by , for a kernel on pairs of Lagrangian subspaces, the product of the weights over , averaged over all placements, is the homomorphism density . The singular contribution is the part of this average that comes from placements in which the subspaces at the two points of some point pair meet nontrivially. The question is whether it vanishes as the field grows, with one threshold on the dimension for all activity patterns, and whether the same holds in a canonical symplectic space and in a parallel model built from symmetric matrices.
Lagrangian Subspaces and the Singular Tail
Let be a symplectic form on a finite-dimensional vector space over a finite field . A Lagrangian subspace of is a subspace of equal to its own orthogonal , and denotes the set of Lagrangian subspaces of .✦
Fix a Lagrangian subspace and a natural number . The center stratum is the set of Lagrangian subspaces with .✦ For a triple of Lagrangian subspaces and a set of active slots, the active center set is the set of Lagrangian subspaces with for every .✦ The kernel on pairs of Lagrangian subspaces is the function (1) and the active face density of with active slots , centered at at rank , is the real number (2)✦
The actual singular tail of an activity pattern is the real number obtained by averaging the product of the face densities (2) over all placements of Lagrangian subspaces at the 13 points, counting only the placements in which some point pair is not transverse, that is, in which the subspaces at its two points meet nontrivially: (3)✦ the product running over the 22 faces.
The Majorant
A dimension profile in dimension is a pair of a map giving each face a number and a map giving each point pair a number .✦ Slot of a face holds the point pair , slot the pair and slot the pair ✦; with the pair in slot , the profile sizes of are , the difference truncated at . A geometric profile in dimension is a dimension profile with natural values such that every face has for its three point pairs , and .✦
Each face carries a three-bit label, the labels being given by a fixed table with one entry for each face, and the separation count of a point pair is the number of slots at which the labels of its two faces differ. With the triangular function on real numbers, the profile cost of is (4)✦ The face natural cost of a face is the -th summand of the first sum in (4).✦
A tail tuple consists of natural numbers and triples and of natural numbers.✦ The tail sizes , the tail heights , the maximal tail height and the tail offset of are defined by these formulas. The planted gain of real numbers and real triples is (5)✦ the sums running over . The tail gain of in dimension with a bit is for and for .✦
Fix natural numbers and sizes , and put . A full tail choice is a pair of a common parameter, an integer with and , and a base parameter, a quadruple feasible for : , for distinct , , and , and for each .✦ A face tail choice with bit is a full tail choice, and with bit it is the single trivial choice.✦ The face tail tuple of a face tail choice is the planted tail tuple when the bit is and the choice is , and the lower tail tuple when the bit is .✦
Now let be a profile in dimension , a natural number and a bit at each face . A global tail choice picks at each face a face tail choice for , , , the sizes and the bit .✦ The global tail tuple of a global tail choice at a face is the face tail tuple of ✦, and the global tail gain of is .✦ The face tail constant at a face with bit is the lower tail constant , where is the Lagrangian constant; at a face with bit and it is the full tail constant (6) The first factor is the orbit mass constant, the product of and the pair orbit constant , so that ✦; the last factor is the reduced planted constant.✦
The global tail constant of a global tail choice is the product over the faces of the face tail constants of at .✦
The full profile constant is , its last factor being the profile mass constant; the residual direct constant is ✦ and the residual nondirect constant is ✦. For a global tail choice on and a real number , the profile term majorant is (7)✦ The profile tail majorant is , the sum running over the global tail choices✦, and the integrated tail majorant is the real function of that is the sum of the profile tail majorants over all dimension profiles in dimension with for some point pair .
The singular tail threshold is a fixed integer, at least , with the following property: if is even, is a geometric profile in dimension and is a global tail choice for some choice of bits, then either , or at every face, at every point pair and .✦
The Canonical Space and the Matrix Model
For a vector space over with dual space , the canonical symplectic form on takes and to .✦ The vertical Lagrangian is the vertical subspace , the kernel of the projection onto , as a Lagrangian subspace for .✦ A prime is odd when .✦ For an odd prime , let be the field with elements; the canonical singular tail is the actual singular tail computed for with .
For a natural number , the space of symmetric matrices over is the subspace of matrices with .✦ The radical quotient form of a symmetric bilinear form on a finite-dimensional vector space over , with radical , is the form on .✦ The discriminant sign of is , where is the quadratic character of and is the Gram matrix of in a basis; it is when has rank zero.✦ The sign of a matrix is the discriminant sign of the form on .✦
For a natural number and a sign , the signed rank layer is the set of matrices in of rank and sign ✦, and its layer mass is the proportion of it fills✦. The rank kernel is a real function on whose value at depends only on whether lies in . An assignment of a matrix to each point is fully transverse when is invertible for every point pair with .✦
For a set of slots, signs and a triple in , the face local average is the uniform mean of over , the kernels taken in .✦ Given an activity pattern and a sign at each corner, the matrix singular tail is the uniform mean, over all assignments in , of the configuration product, which is when is fully transverse and otherwise the product over the faces of the face local averages at ; over it reads (8)✦
The Claim
Theorem 1 (Vanishing singular contribution). Let be a finite field of odd characteristic, a symplectic space of dimension over , and a natural number with and . For every Lagrangian subspace of and every activity pattern , (9)✦ where gives the bit exactly to the faces with . For all natural numbers and and every choice of a bit at each face, as the real variable tends to , (10)✦ For every natural number with and every activity pattern , as through the odd primes, (11)✦ and, for every choice of signs at the corners, (12)✦
In the paper, the result reads: for every sufficiently large even and every activity set , as , with one threshold for all activity sets. Here is the placement of the centers at the faces, and is the average over all placements and .
What Is Checked
The claims (9) to (12) are stated over a symplectic space of dimension over a finite field of odd characteristic, its Lagrangian subspaces with their center strata and active center sets, the active face density and the actual singular tail , the integrated tail majorant and the threshold , the canonical form with its vertical Lagrangian, and the matrix singular tail with its rank kernels, with limits taken through the odd primes.
About these constructions the library proves the global tail dichotomy: an integer with the property that defines the singular tail threshold exists.✦