Definition The coset interval embedding into an ambient interval
✓ definition formalized

The coset interval embedding into an ambient interval

Let \(W\) be a finite-generator real Coxeter system, \(B\) a plane basis, and \(D\) its induced dihedral subgroup. Choose \(a_0\in W\) satisfying \[ \ell_W(a_0)\le\ell_W(da_0)\qquad(d\in D). \] Fix ambient endpoints \(u,b\in W\) and intrinsic endpoints \(g,k\in D\), with \[ u\le_Wga_0,\qquad ka_0\le_Wb. \] The coset interval embedding is \[ \iota_{a_0}:[g,k]_D\hookrightarrow[u,b]_W,\qquad \iota_{a_0}(z)=za_0. \] In Lean this map is PlaneBasis.cosetIntervalMap. Its construction is defined even if the source interval is empty; the transport theorem separately assumes \(g\le_Dk\).

The endpoint bounds make the map well-defined

For \(z\in[g,k]_D\), the least-coset graph comparison transfers \(g\le_Dz\le_Dk\) to \[ ga_0\le_Wza_0\le_Wka_0. \] Combining this with the displayed ambient bounds gives \(u\le_Wza_0\le_Wb\). Thus the product belongs to the stated ambient interval.

Injectivity and the graph comparison

If \(za_0=wa_0\), right cancellation gives \(z=w\). This proves injectivity.

The coset graph theorem also identifies upward reflection edges: \[ z\to_Dw\quad\Longleftrightarrow\quad za_0\to_Wwa_0. \] A relative Bruhat comparison is a chain of these edges, so the embedding preserves relative order. The source rank is measured by \(\ell_D\), and the codomain's ambient rank by \(\ell_W\).

The example of a relative simple reflection with ambient length three in the relative-interval definition shows how these two ranks can differ. Full interval transport constructs a target relative rank correspondence after composing \(\iota_{a_0}\) with an ambient interval order isomorphism.

Relation to the OpenAI notes

Explicit coordinates, affine lengths and full target range.

OpenAI notes: The proposition describes the image directly as a full relative interval in a target plane coset. Its proof recovers covers from rank-two intervals and exhausts each internal target rank with two image vertices. dih:transport.

Lean (primary): The construction chooses a least target representative and writes F(z) as the ambient image multiplied on the right by its inverse. Rank-two preservation gives consecutive-rank preservation, then an affine length identity by induction. Injectivity and the two-element internal ranks give the exact full range. The theorem returns all coordinate, order, edge and length properties explicitly. DyerPaths:676, DyerPaths:504, DyerPaths:536, DyerPaths:567, DyerPaths:736.

Role in the proof: The Blueprint uses the intrinsic coordinate map F and retains every ambient endpoint bound. The source and target bottom lengths may differ by the constant prescribed by the affine identity.

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.TitsSpace.PlaneBasis.cosetIntervalMap: DyerPaths.lean:676.

Used by