The projectivity step for an upper-coordinate image
Fix a finite-generator Coxeter system with symmetric coefficient algebra \(A\). Suppose a word \(\mathbf t\) has both properties:
- \(L_{\mathbf t}(U)\) is projective for every Bruhat upper set \(U\);
- every compatible stalk tuple on an upper set has a common word lift.
Then, for every simple reflection \(s\), the longer word \(s\mathbf t\) has projective coordinate images on every Bruhat upper set.
Relation to the OpenAI notes
Explicit word-image construction of the extension property.
OpenAI notes: The moment-graph sheaf theorem gives extension of compatible sections on upper sets for the normalized boundary sheaf through the cited Braden–MacPherson and Soergel bridges. bg:bmp.
Lean (primary): The word module gives coordinate images and stable coefficient characters. Reflection geometry makes each prime congruence class a single weight or a reflection pair. Explicit diagonal operators combine class lifts, and prime-denominator removal proves compatible-tuple lifting from image projectivity. The word induction supplies upper-image projectivity using its two kernel sequences. IntegralPairings:470, IntegralPairings:467, IntegralPairings:1299, IntegralPairings:1408, IntegralPairings:2057, IntegralPairings:1620, IntegralPairings:1833, BalancedPairings:292, Support:293.
Role in the proof: Upper-image lifting gives interval flabbiness after extension to the ambient upper closure. Full-interval section realization also has a direct free-image proof. Free stalks, root quotients, flabbiness, generation and the normalized word top then supply the graded boundary summand.
Proof
Step 1. Choose the two-stage maps
Fix an upper set \(U\), and put \(J=U\cup sU\), \(K=U\cap sU\). Both sets are upper and reflection-stable. Write \(L=L_{\mathbf t}\), \(L'=L_{s\mathbf t}\).
Use \[ r:L'(U)\to L'(K),\qquad F_U:L'(U)\to L(U),\qquad k:L(U)\to L(K). \] The restrictions \(r,k\) are surjective. The identity \(kF_U=F_Kr\) gives a map \[ \bar F:\ker r\to\ker k. \]
Step 2. Prove surjectivity on the first kernel
Take \(z\in\ker k\). Choose \(n\in M_{\mathbf t}\) with \(p_U(n)=z\). Its projection to \(K\) is \(k(z)=0\).
Since \(K\) is reflection-stable, seed vanishing gives \(p_K'(1\otimes n)=0\). Therefore \(w=p_U'(1\otimes n)\) lies in \(\ker r\). The first-map seed formula gives \[ \bar F(w)=F_U(p_U'(1\otimes n))=p_U(n)=z. \] Thus \(\bar F\) is surjective.
Step 3. Prove projectivity of the outer target and the first quotient
The first hypothesis gives projectivity of \(L(K)\). Since \(K\) is reflection-stable, stable-image projectivity gives projectivity of \(L'(K)\).
Both \(L(U)\) and \(L(K)\) are projective by the first hypothesis. The surjective restriction \(k\) therefore has projective kernel, by the projective split-sequence theorem. Hence \(\ker k\) is projective.
Step 4. Prove projectivity of the common kernel
Put \[ D=\ker r\cap\ker F_U,\qquad T=\ker(L(J)\to L(U)). \] The first hypothesis makes \(L(J)\) and \(L(U)\) projective. Their restriction is surjective, so \(T\) is projective. Its scalar twist \({}^\tau T\) is projective as well.
The second hypothesis is exactly the compatible-tuple lifting input to the common-kernel theorem. That theorem identifies \(D\simeq{}^\tau T\), proving that \(D\) is projective.
Step 5. Apply the two-stage projectivity theorem
All its inputs have now been checked: \(r\) and \(\bar F\) are surjective, while \(L'(K)\), \(\ker k\), and \(D\) are projective. The theorem applies successively to \[ 0\to D\to\ker r\to\ker k\to0 \] and \[ 0\to\ker r\to L'(U)\to L'(K)\to0. \] It proves projectivity of \(L'(U)\). Since \(U\) was arbitrary, the longer word has upper-image projectivity.
This is the source's bottProjectiveUpper_cons. Its two hypotheses belong to the shorter word; the prime-denominator lifting theorem supplies the second from the first when the full word induction is assembled.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.TitsSpace.bottInter_first_zero: IntegralPairings.lean:1373.
OAI.KLInvariance.TitsSpace.bottInterKernelFirst_surjective: IntegralPairings.lean:1379.
OAI.KLInvariance.TitsSpace.bottCommon_projective: IntegralPairings.lean:1392.
OAI.KLInvariance.TitsSpace.bottProjectiveUpper_cons: IntegralPairings.lean:1408.