Theorem The exact stalk kernel across the first reflection
✓ theorem proved

The exact stalk kernel across the first reflection

Fix a finite-generator Coxeter system and \(A=\operatorname{Sym}_{\mathbb R}(V)\) for its real extended realization \(V\). Let \(M_{s\mathbf t}=A\otimes_{A^s}M_{\mathbf t}\) be a word module beginning with the simple reflection \(s\), and let \(\alpha_s\) be its root polynomial. For any weight \(x\), put \[ X=M_{s\mathbf t}^x,\qquad Y=M_{s\mathbf t}^{sx}, \qquad f=\pi_x,\quad g=\pi_{sx}. \] Then \[ f(m)=0\quad\Longrightarrow\quad g(m)=\alpha_s y\ \text{for some }y\in Y. \] For the pair-projection upper map \(u:Y\to Q\), this gives the exact formula \[ \ker u=\alpha_s Y. \]

Relation to the OpenAI notes

Word-image construction before the boundary sheaf.

OpenAI notes: The moment-graph sheaf theorem states the quotient description of an edge module and freeness for the normalized indecomposable boundary sheaf, through the cited moment-graph and Soergel constructions. bg:bmp.

Lean (primary): The tensor word first supplies explicit subexpression evaluations and image stalks. Right coefficient characters give label annihilation; a first-factor reflection proves the exact kernel across the initial simple reflection. Reflected pair lattices then describe stalks after prepending a letter. Frobenius:1558, Frobenius:2018, BottSamelson:214, BottSamelson:964, BottSamelson:1130, BottSamelson:1510, Support:426.

Role in the proof: The word-length induction proves stalk freeness and exact kernels together. Its two tail maps, quotient regularity and lattice saturation prove the non-head reflection step; the ascent/descent split sequences prove new stalk freeness. Coefficient transport then gives the polynomial word-sheaf input to the boundary-summand construction.

Proof

Step 1. Relate the reflected tensor to exchanged evaluations

Write \(\tau=\chi_s\), let \(F_s\) reflect the first tensor factor, and let \(\sigma\) exchange the first subexpression bit. We prove \[ E_e(F_s(m))=\tau(E_{\sigma(e)}(m)). \] It is enough to check a pure tensor \(m=a\otimes n\), then extend by additivity.

For \(e=(0,\eta)\), the two sides are \[ \tau(a)E_\eta(n) \quad\text{and}\quad \tau(a\,\tau(E_\eta(n)))=\tau(a)E_\eta(n), \] using \(\tau^2=1\). For \(e=(1,\eta)\), they are \(\tau(a)\tau(E_\eta(n))\) and \(\tau(aE_\eta(n))\), equal by multiplicativity.

Step 2. Prove divisibility of the difference of tensors

For a pure tensor, simple-root divisibility supplies \(b\in A\) with \(a-\tau(a)=\alpha_s b\). Therefore \[ a\otimes n-F_s(a\otimes n) =(a-\tau(a))\otimes n =\alpha_s(b\otimes n). \] A finite sum of pure tensors has difference equal to the sum of these root multiples. Hence every \(m\) satisfies \[ m-F_s(m)=\alpha_s v \] for some \(v\in M_{s\mathbf t}\).

Step 3. Transport vanishing to the opposite weight

Assume \(\pi_x(m)=0\). If \(e\) has weight \(sx\), then \(\sigma(e)\) has weight \(s(sx)=x\). Thus \(E_{\sigma(e)}(m)=0\), and Step 1 gives \(E_e(F_s(m))=0\).

Every coordinate in the \(sx\)-stalk of \(F_s(m)\) is zero, so \(\pi_{sx}(F_s(m))=0\). Project Step 2 to that stalk: \[ \pi_{sx}(m)=\alpha_s\pi_{sx}(v). \] This proves the first assertion with \(y=\pi_{sx}(v)\).

Step 4. Prove both kernel inclusions

If \(z\in\ker u\), the pair-projection kernel criterion gives \(m\) with \(f(m)=0\) and \(g(m)=z\). Step 3 gives \(z\in\alpha_s Y\).

Conversely, the reflection-label annihilation theorem gives \(\alpha_s q=0\) for every \(q\in Q\). Hence for \(y\in Y\), \[ u(\alpha_s y)=\alpha_s u(y)=0. \] Therefore \(\alpha_s Y\subseteq\ker u\), proving equality.

The argument works for every \(x\), and it uses the actual image stalks. In the one-letter example, \(X=Y=A\), \(Q=A/(\alpha_s)\), and the formula is the familiar kernel of the quotient map.

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.TitsSpace.bottSimpleFlip_divisible: Frobenius.lean:1907.

OAI.KLInvariance.TitsSpace.bottEvaluation_flip: Frobenius.lean:1934.

OAI.KLInvariance.TitsSpace.bottStalkProjection_flip_zero: Frobenius.lean:1954.

OAI.KLInvariance.TitsSpace.bottSimpleWall_kernel_projection: Frobenius.lean:1968.

OAI.KLInvariance.TitsSpace.bottSimpleWall_annihilated: Frobenius.lean:2011.

OAI.KLInvariance.TitsSpace.bottSimpleWall_upper_kernel: Frobenius.lean:2018.

OAI.KLInvariance.PairProjection.kernel_upper: Frobenius.lean:1848.

OAI.KLInvariance.PairProjection.upper_kernel_of_divisibility: Frobenius.lean:1853.

Used by