Closed

Finite change of calibration

Theorem C.1.9. (Finite change of calibration). Let L/KL/K be finite of degree nn. If YY is good Hausdorff and pure of dimension dd over LL, G\mathcal G is coherent, and β∈AY/L,cd,d(Y)\beta\in\mathcal A^{d,d}_{Y/L,c}(Y), then ∫Y[K]/K,Gβ=n∫Y/L,Gβ.\begin{equation} \int_{Y_{[K]}/K,\mathcal G}\beta =n\int_{Y/L,\mathcal G}\beta . \end{equation} For Y=XLY=X_L and G=FL\mathcal G=\mathcal F_L, this gives (C.1.2). The factors multiply correctly in finite towers.

Proof

Proof. First let Y=ML(B)Y=\mathcal M_L(B) be affinoid with one unit tuple ff, and let NN be the finite BB-module representing G\mathcal G. By Lemma C.1.7, its actual level algebra DwD_w, module P=N⊗BDwP=N\otimes_B D_w on that level, real moment image and tangent lattice are identical in the two field interpretations. Choose a common subdivision avoiding both boundary images at generic top-cell points, as in Proposition C.1.6. There DwD_w and PP are finite canonical over RL,rR_{L,r}. They are also finite canonical over RK,rR_{K,r} by the finite free coefficient decomposition and composed finite presentations. Lemma C.1.8 gives P⊗RK,rHK,r=P⊗RL,rHL,rP\otimes_{R_{K,r}}H_{K,r} =P\otimes_{R_{L,r}}H_{L,r} as the same module with different scalar fields. Its dimension over HK,rH_{K,r} is nn times its dimension over HL,rH_{L,r}. Dividing by the same lattice index proves TfK(N)=nTfL(N)\mathcal T_f^K(N)=n\mathcal T_f^L(N), using the closed finite-module chains. Integrating by the finite-cell functional and then applying the identical finite covers and inclusion–exclusion formula proves (C.1.8) for the ordinary compact integrals.

There is also a direct check of (C.1.2). The basis of L/KL/K gives p∗p∗F≃F⊕np_*p^*\mathcal F\simeq \mathcal F^{\oplus n} on every full affinoid inverse image, compatibly with restriction. The map p:(XL)[K]→Xp:(X_L)_{[K]}\to X is finite with its canonical module norm. The spaces have the required purity and Hausdorffness by the preceding lemmas. Thus the exact finite projection theorem [5.54.1] and coherent additivity give ∫(XL)[K]/K,FLp∗α=∫X/K,p∗FLα=n∫X/K,Fα.\int_{(X_L)_{[K]}/K,\mathcal F_L}p^*\alpha =\int_{X/K,p_*\mathcal F_L}\alpha =n\int_{X/K,\mathcal F}\alpha . This uses the actual coherent direct image [2.35.15], with no local-freeness assumption on F\mathcal F. Finite tower factors multiply because [E:L][L:K]=[E:K][E:L][L:K]=[E:K], while Theorem C.1.1 has factor one in every tower. ◻

Citation

Non-Archimedean Geometry Foundations, Tag NAG-002520, C.1.9.

Proof in chapter · Fixed-edition record