Side-column overlap in two-step rectangle decompositions #
A term in the square of the grid differential is a pair of composable rectangles. The
juxtaposition proof of ∂² = 0 splits according to how the two pairs of vertical sides meet.
This file makes that split exact: the pairs are disjoint, have one common column, or coincide.
The last case occurs exactly when the two rectangle moves return to the source state.
The finite set GridRectangleDecomposition.commonSideColumns is the intersection of the two
two-element side sets. Its cardinality is therefore at most two. Cardinality zero is the predicate
HasDisjointSides; cardinality one is named HasOneCommonSide; and cardinality two is equivalent
to equality of the source and target states. Consequently a nondiagonal coefficient of the
differential square has precisely the two geometric cases of the rectangle pairing: disjoint
rectangles or rectangles sharing exactly one side column.
Main definitions #
TauCeti.GridRectangleDecomposition.commonSideColumns: the side columns used by both rectangles in a decomposition.TauCeti.GridRectangleDecomposition.HasOneCommonSide: the two rectangles share exactly one side column.TauCeti.GridRectangleDecomposition.HasSameSideOverlapandTauCeti.GridRectangleDecomposition.HasMixedSideOverlap: the common column has the same or opposite orientation in the two rectangles.
Main results #
TauCeti.GridRectangleDecomposition.card_commonSideColumns_le_two: at most two side columns are common.TauCeti.GridRectangleDecomposition.card_commonSideColumns_eq_two_iff: both side columns are common exactly for a diagonal two-step path.TauCeti.GridRectangleDecomposition.hasOneCommonSide_iff_sameSideOverlap_or_mixedSideOverlap: the same-side and mixed-side orientations exhaust the one-common-side case.TauCeti.GridRectangleDecomposition.sum_filter_not_hasDisjointSides_eq_sameSide_add_mixedSide: off the diagonal, any weighted family splits into its two overlap orientations.TauCeti.GridRectangleDecomposition.hasDisjointSides_or_hasOneCommonSide_or_eq: every decomposition belongs to one of the three cases.TauCeti.GridRectangleDecomposition.hasDisjointSides_of_disjoint: a decomposition whose target applies two disjoint column transpositions has disjoint side pairs.TauCeti.GridRectangleDecomposition.target_ne_source_iff: a decomposition is nondiagonal exactly when its sides are disjoint or have exactly one column in common.TauCeti.GridRectangleDecomposition.sideColumns_ne_of_hasOneCommonSide: sharing exactly one side column means the two side pairs differ.TauCeti.GridRectangleDecomposition.hasOneCommonSide_of_mem_commonSideColumns: a nondiagonal decomposition with a common side column has exactly one.TauCeti.GridRectangleDecomposition.hasOneCommonSide_of_right_eq_left_of_neandTauCeti.GridRectangleDecomposition.hasOneCommonSide_of_left_eq_right_of_ne: off the diagonal, two rectangles sharing a grid line with opposite orientations share no other side.TauCeti.GridRectangleDecomposition.target_apply_ne_iff_mem_sideColumns_union: when exactly one side is common, the three side columns are precisely the columns moved between the endpoint states.TauCeti.GridRectangleDecomposition.sideColumns_union_eq_of_hasOneCommonSide: two decompositions between the same endpoint states, one of which has one common side, use the same three columns.
References #
The split is the opening bookkeeping of the juxtaposition proof in Ozsváth--Stipsicz--Szabó, Grid Homology for Knots and Links, Chapter 4.6.
The columns which are sides of both rectangles in a two-step decomposition.
Instances For
A column is common to the two rectangles exactly when it is a side of each one.
At most two columns are sides of both rectangles in a decomposition.
Two rectangles in a decomposition share exactly one side column.
Instances For
Two rectangles share exactly one side column, with that column initial for both rectangles or terminal for both rectangles.
Instances For
Two rectangles share exactly one side column, with that column initial for one rectangle and terminal for the other.
Instances For
The defining alternatives for a same-side overlap.
The defining alternatives for a mixed-side overlap.
Having one common side means that there is a unique column which is a side of both rectangles.
Two rectangles sharing their initial side and having distinct terminal sides share exactly one side column.
Two rectangles sharing their terminal side and having distinct initial sides share exactly one side column.
Two rectangles sharing their initial side, the second ending strictly inside the first's column interval, share exactly one side column.
Two rectangles, the first ending where the second starts, whose other sides differ share exactly one side column.
Two rectangles, the first starting where the second ends, whose other sides differ share exactly one side column.
If two rectangles share exactly one side column, that column occurs in one of the four possible ordered-side positions, and the two noncommon sides are distinct.
An overlap of exactly one side is either same-side or mixed-side.
Sharing exactly one side is equivalent to having a same-side or mixed-side overlap.
Same-side and mixed-side overlaps are mutually exclusive.
If two rectangles share exactly one side column, their union of side columns has cardinality three.
When two rectangles share exactly one side column, their three side columns are precisely the columns on which the target state differs from the source state.
The two side pairs are disjoint exactly when their common-side set is empty.
The common-side set has cardinality zero exactly when the side pairs are disjoint.
This is not a separate simp lemma: Finset.card_eq_zero followed by
commonSideColumns_eq_empty_iff already gives the same normal form.
A two-step rectangle path returns to its source exactly when its two rectangles use the same unordered pair of side columns.
Both side columns are common exactly when the two-step rectangle path returns to its source.
Every two-step rectangle decomposition has disjoint sides, exactly one common side, or equal source and target.
Two disjoint column transpositions move four columns, so no two-step rectangle decomposition of the resulting state can share a side column.
This is the entry point that puts a two-step term over such a target in the disjoint-side case of
hasDisjointSides_or_hasOneCommonSide_or_eq.
In a nondiagonal two-step rectangle decomposition, the side pairs are disjoint or share exactly one column. This is the two-case split used by the nondiagonal juxtaposition pairing.
A nondiagonal decomposition whose two rectangles have a common side column has exactly one common side column.
Off the diagonal, when the terminal side of the first rectangle is the initial side of the second, this grid line is their unique common side.
Off the diagonal, when the initial side of the first rectangle is the terminal side of the second, this grid line is their unique common side.
A two-step rectangle decomposition is nondiagonal exactly when its side pairs are disjoint or share exactly one column.
A decomposition with disjoint side pairs is necessarily nondiagonal.
A decomposition with one common side is necessarily nondiagonal.
A decomposition sharing exactly one side column does not use the same unordered pair of side columns twice: coinciding side pairs are the diagonal case.
The disjoint-side and one-common-side cases are mutually exclusive.
Two decompositions between the same endpoint states use the same three side columns when one of them has one common side: the other cannot have disjoint sides, which would move four columns, nor equal source and target.
Off the diagonal, failing to have disjoint side pairs is equivalent to having a same-side or mixed-side overlap.
Every two-step rectangle domain in a diagonal family has nondisjoint side pairs.
For an off-diagonal family of two-step rectangle domains, the sum over nondisjoint side pairs splits into the same-side and mixed-side overlap sums.