The unblocked complex of a stabilized grid #
Let G be a grid diagram of size n, let s be a column, and let
G' = G.stabilizeX s.castSucc (G.X s).castSucc s be the stabilization that splits the
X-marking of column s by inserting a new column immediately before s and a new row
immediately below that marking. In G' the new 2 × 2 block is centred at the grid point
c = (s.succ, (G.X s).succ): its two X-markings lie in the squares northwest and southeast
of c, the new O-marking in the square southwest of c, and the square northeast of c is
empty.
The grid states of G' split into those that contain c, denoted I, and the others, denoted
N. This file proves two coefficient-level facts about the unblocked differential ∂⁻ of
G'. They are the ingredients of the presentation of GC⁻(G') as the mapping cone of the
component I → N of ∂⁻ in TauCeti.KnotTheory.Grid.Stabilization.Cone, with the I block
identified with GC⁻(G) over one more variable; this is the form in which Ozsváth, Stipsicz and
Szabó compare the stabilized complex with GC⁻(G).
- The
N-to-Icoefficient block vanishes. A rectangle from a state outsideIto a state inIhascas a corner, so it covers one of the twoX-marked squares of the block. Hence∂⁻has no matrix coefficient fromNtoI(unblockedCoefficient_stabilizeX_eq_zero). - The
Icoefficient block agrees withGC⁻(G)over one more variable. Every state ofIis obtained from a unique state ofGby insertingc(GridState.insertPoint), and insertingcidentifies the rectangles counted by∂⁻between states inIwith those counted by∂⁻onGC⁻(G). The twoX-markings of the block cover exactly the squares that theX-marking of columnscovers inG, and a rectangle avoiding them never coverscor the newO-marking. So the matrix coefficients agree after renaming the variables ofGinto those ofG'; the variable of the newO-marking never occurs (unblockedCoefficient_stabilizeX_insertPoint).
Main results #
TauCeti.GridDiagram.unblockedCoefficient_stabilizeX_insertPoint: the matrix coefficients of∂⁻between states containingcare the renamed coefficients of∂⁻forG.TauCeti.GridDiagram.unblockedCoefficient_stabilizeX_eq_zero:∂⁻has no matrix coefficient from a state not containingcto a state containing it.
References #
These coefficient identities are part of the stabilization-invariance argument in
Ozsváth--Stipsicz--Szabó, Grid Homology for Knots and Links, Section 5.2, where the complex of
the stabilized diagram is identified with a mapping cone built from the complex of G.
A rectangle between states containing the centre c = (s.succ, (G.X s).succ) of the new
block avoids the X-markings of the stabilization exactly when the rectangle it is transported
from avoids the X-markings of G.
If a transported rectangle avoids the split X-square, it covers the O-markings of the old
columns of the stabilization that the original rectangle covers in G, and never the new
O-marking.
If a transported rectangle avoids the split X-square, its weight is the weight of the
original rectangle with the variables renamed into those of the stabilization.
Transporting along the insertion of the centre c = (s.succ, (G.X s).succ) identifies the
rectangles counted by the unblocked differential of G with those counted by the unblocked
differential of the stabilization between states containing c.
The I coefficient block of the stabilized complex. Between grid states containing the
centre c = (s.succ, (G.X s).succ) of the new block, the matrix coefficients of the unblocked
differential of the stabilization are those of G, with the variable of each column renamed to
the variable of the corresponding old column. The variable of the new O-marking does not
occur.
The N-to-I coefficient block vanishes. The unblocked differential of the stabilization
has no matrix coefficient from a grid state not containing the centre
c = (s.succ, (G.X s).succ) of the new block to one containing it: such a rectangle has c as
a corner and covers one of the two X-marked squares of the block.