The census of boundary-condition classes with a reflection-coset sector #
Haba, Hosotani and Kawamura classify the boundary conditions of SU(N) gauge theory on
S¹/ℤ₂ by the block sizes [p; q, r; s] of two commuting parity matrices P₀, P₁: p basis
vectors with (P₀, P₁) = (+, +), q with (+, −), r with (−, +), s with (−, −), and
p + q + r + s = N. Boundary conditions related by a gauge transformation that shifts the
Wilson line are equivalent, and the equivalence is generated by the move
[p; q, r; s] ~ [p−1; q+1, r+1; s−1]. The winding element U = P₀ P₁ has
det U = (−1)^{q+r}, so the theory has a reflection-coset sector (where the boundary sign can
act) exactly when q + r is odd. This file counts the classes that have one, for every N.
Main results #
label_move,rank_move,label_eqv,rank_eqv: the label(p+q, q+s)andNare invariant.eqv_of_label: two boundary conditions with the sameNand the same label are joined by a chain of moves, so the label is a complete invariant; withwitness_labelthe classes are the pairs(u, v)withu, v ≤ N, andcard_classesrecovers the count(N+1)².coset_parity:q + randu + v + Nhave the same parity, so having a coset sector is a property of the class.card_cosetClasses: exactly⌊(N+1)²/2⌋of the(N+1)²classes have a coset sector.
References #
- N. Haba, Y. Hosotani, Y. Kawamura, Classification and dynamics of equivalence classes in SU(N) gauge theory on the orbifold S¹/Z₂, arXiv:hep-ph/0309088.
- C. Marín, What the Higgs Potential Cannot See: Bulk Matter That Cannot Help Select a Boundary Condition (Part V), Zenodo, doi:10.5281/zenodo.21727094.
Boundary conditions, the move, and the label #
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
Boundary conditions in the same equivalence class.
Equations
Instances For
The label is a complete invariant #
The direction with content. If x and y have the same N and the same label,
walk x towards y one move at a time. The step is legal because the label
forces the room for it: if x.q < y.q then x.p = y.p + (y.q - x.q) ≥ 1 and
x.s = y.s + (y.q - x.q) ≥ 1.
The label is complete. Same N, same label ⟹ joined by moves.
... and every point of the grid is a label: the explicit representative.
Equations
Instances For
Counting the classes #
The classes of rank N, as their labels: the (N+1) × (N+1) grid.
Equations
- GaugeHiggsUnification.CosetCensus.classes N = Finset.range (N + 1) ×ˢ Finset.range (N + 1)
Instances For
The classes that admit a reflection-coset sector.
Equations
- GaugeHiggsUnification.CosetCensus.cosetClasses N = {c ∈ GaugeHiggsUnification.CosetCensus.classes N | (c.1 + c.2 + N) % 2 = 1}
Instances For
Proposition (Part V), the geometric half of the dichotomy. Of the
(N+1)^2 equivalence classes of boundary conditions, exactly ⌊(N+1)^2/2⌋ admit
a reflection-coset sector -- and in the other ⌈(N+1)^2/2⌉ the boundary sign is
invisible to every multiplet, for reasons that have nothing to do with matter.
Sanity checks against the table for N = 2, …, 10 #
The first two are the control that matters: they do not touch the theorems above.
They enumerate the actual [p;q,r;s] tuples of rank 3, take their labels, and
compare with the grid -- so if witness or label_le were wrong, or the label
missed part of the grid, these would fail and the census would still "prove".
All boundary conditions of rank N, by brute force.
Equations
- One or more equations did not get rendered due to their size.
Instances For
And the controls are not vacuous. A label that looks equally natural --
(p+q, q+r), the two "off-diagonal" sums -- is not invariant under the move,
so label_move is a real constraint and not bookkeeping: