NAG-003025 · Accepted whole
One boundary model for a principal localization
Proper direct image on historical ordinary compact tests
Open in the English reader · PDF page 1885
Statement · exact TeX
\begin{proposition}[One boundary model for a principal localization]\label{prop:samemodel}
For $0\ne h\in A$, retain the model $Y$ of Theorem~\ref{thm:lattice}. Put $V_h=\Spec B_h\subset V$, and let $S_h$ be the codimension-one points of $Y\setminus V_h$. Then $B_h$ is the normalization of $A_h$, $S\subset S_h$ is finite, and the exact lattice $M_{A_h}$ is obtained by the construction of Proposition~\ref{prop:conductor} on this same $Y$, using $S_h$. The conductor of $A_h\subset B_h$ is $I_h$. Thus the inclusion $M_A\to M_{A_h}$ is realized by extending boundary vectors by zero on $S_h\setminus S$, with the respective conductor descent conditions retained.
If $A$ is normal, restriction of orders to the new boundary points gives the additional exact sequence
\[
0\longrightarrow M_A\longrightarrow M_{A_h}
\longrightarrow\Z^{S_h\setminus S},
\qquad [f]\longmapsto(\ord_D f)_{D\in S_h\setminus S}.
\]
There is no assertion of surjectivity onto the final group.
\end{proposition}Full selected proof · exact TeX
\begin{proof}
Integral closure commutes with localization by clearing denominators in a monic equation, so the normalization is $B_h$. The same finiteness argument for the codimension-one boundary of a dense open in $Y$ gives finiteness of $S_h$. Apply Proposition~\ref{prop:conductor} to $V_h$; properness, the function field and the normalized orders are unchanged.
Clearly $I_h$ is contained in the conductor of $A_h\subset B_h$. Conversely, choose finite $A$-module generators $b_1,\ldots,b_q$ of $B$. If $a/h^m$ belongs to that conductor, then each $ab_j$ lies in $A_h$. A common power $h^N$ clears all these denominators, so $h^NaB\subset A$ and $h^Na\in I$. Therefore $a/h^m\in I_h$. This proves the conductor identity. If the localized rings agree, the descent condition is simply vacuous. Units of $A$ have zero orders at the additional points because those points lie in $V$, proving the extension-by-zero description of their boundary vectors.
For the last sequence assume $A=B$. A unit $f$ of $A_h$ with zero orders on $S_h\setminus S$ has zero order at every height-one point of $V$: the remaining points lie in $V_h$, where $f$ is already a unit. By \tagref{031T}(2), both $f$ and $f^{-1}$ then belong to $A$. This proves exactness at $M_{A_h}$, and injectivity at the first arrow was proved in Theorem~\ref{thm:localization}.
\end{proof}Typeset reader excerpt

Typed proof inputs
- NAG-003013 · selected whole (earlier same-packet whole)
- NAG-003015 · selected whole (earlier same-packet whole)
- NAG-003022 · selected whole (earlier same-packet whole)
Reader cross-references
Proof map
Verified map inputs · NAG-003025
AND · joint registered proof inputs
- NAG-003013 · Boundary construction of the actual unit lattice · accepted whole
- Use the actual old proper model, its finite normalization and the conductor test, now on the genuine principal-open chart B_h. Normalized boundary orders and the function field are unchanged.
- NAG-003015 · Exact boundary image and conductor descent · accepted whole
- Use the actual old proper model, its finite normalization and the conductor test, now on the genuine principal-open chart B_h. Normalized boundary orders and the function field are unchanged.
- NAG-003022 · Exact stability, with a normality condition for saturation · accepted whole
- Use the established injection for A→A_h on units modulo the same constants.
NAG-003025 · One boundary model for a principal localization · Chapter argument