Problems about torsors over regular rings (Q2126107)

This paper contains a survey of torsors (principal homogeneous spaces) under reductive groups over regular rings and their properties. The author discuses the various problems and conjectures, including reductions, connections between conjectures, and known results, sometimes in an extended form. Here are some of the conjectures. \begin{itemize} \item[1.] \textbf{Conjectures About Vector Bundles over Regular Rings} \begin{itemize} \item[a)] \textbf{The Bass-Quillen conjecture:} For a regular ring \(R\), every vector bundle on \(\mathbb A^{d}_{R}\) descends to \(R\). The Bass-Quillen conjecture can be reduced to the case that \(d=1\) and the regular ring \(R\) is local. In this case, the conjecture is open for \(R\) ramified, and it is known to be true for \(R\) unramified (A regular local ring \(R\) is unramified if the ring \(R/pR\) is regular, where \(p\) is the residue characteristic of \(R\), and \(R\) is ramified if \(R\) is not unramified). Moreover, the author includes a proof of the conjecture assuming that all the localizations of the regular ring \(R\) at its maximal ideals are unramified. Quoting the author, the Bass-Quillen conjecture ``grew out of Serre's problem solved by \textit{D. Quillen} [Invent. Math. 36, 167--171 (1976; Zbl 0337.13011)] and \textit{A. A. Suslin} [Sov. Math., Dokl. 17, 1160--1164 (1976; Zbl 0354.13010); translation from Dokl. Akad. Nauk SSSR 229, 1063--1066 (1976)]: every vector bundle over an affine space over a field is free.'' \item[b)] \textbf{The Quillen conjecture:} For a regular local ring \((R, \mathfrak m)\) and an \(r \in \mathfrak m\) that is a regular parameter (that is, \(r\notin \mathfrak m^{2}\)), every finite projective \(R[\frac {1} {r}]\)-module is free. Equivalently, every finite projective \(R[\frac {1} {r}]\)-module extends to a finite projective \(R\)-module. If the ring \(R\) is regular local, the Quillen Conjecture holds when \(R\) is of equicharacteristic, and it is open when \(R\) is of mixed characteristic, in particular, if \(R\) is ramified. \item[c)] \textbf{The Lam Conjecture:} For a ring \(R\), every stably free vector bundle on \(\mathbb A_{R}^{d}\) descends to \(R\). Equivalently, for a local ring \(R\), every stably free module over the polynomial ring \(R[t]\) is free. \end{itemize} \item[2.] \textbf{Conjectures on Torsors under Reductive Groups over Regular Rings} \begin{itemize} \item[a)] \textbf{The Grothendieck-Serre Conjecture;} For a regular local ring \(R\) and a reductive \(R\)-group scheme \(G\), no nontrivial \(G\)-torsor over \(R\) trivializes over \(K := \mathrm{Frac}(R)\). Equivalently, \(H^{1}(R, G) \hookrightarrow H^{1}(K, G)\). Quoting the author: ``To sum up, the Grothendieck-Serre conjecture is known in equal characteristic but remains open in mixed characteristic, especially, over ramified regular local rings, for which one may need a substantially different approach. As for the unramified mixed characteristic case beyond quasi-split \(G\), we feel that it may, in principle, be approachable, perhaps by finding some way to improve or to bypass the presentation lemma, but we do not know of a precise way to attack it fruitfully.'' \item[b)] \textbf{The Colliot-Thélène-Sansuc Purity Conjecture:} For a regular local ring \(R\), its fraction field \(K\), and a reductive \(R\)-group scheme \(G\), a \(G\)-torsor over \(K\) that extends to a G-torsor over \(R_{p}\) for every height \(1\) prime \(p \subseteq R\) extends uniquely to a \(G\)-torsor. Equivalently, \[ H^{1}(R, G) = \bigcap_{p\ of\ height\ 1} H^{1}(R_{p}, G)\ inside \ H^{1}(K, G). \] The Colliot-Thélène-Sansuc Purity Conjecture implies he following \item[c)] \textbf{Conjecture:} For a regular local ring \(R\) and its fraction field \(K\), a reductive \( K\)-group extends to a reductive \(R\)-group scheme if and only if it extends to a reductive\( R_{p}\)-group scheme for every prime \(p\subseteq R\) of height \(1\), in which case this extension is unique up to isomorphism. \end{itemize} \item[3.] \textbf{The Grothendieck-Serre Conjecture for Levi Reductions and Parabolic Subgroups} \begin{itemize} \item[a)] \textbf{Conjecture:} For a regular local ring \(R\), its fraction field \(K\), a reductive \(R\)-group scheme \(G\), a parabolic \(R\)-subgroup \(P \subseteq G\), and an \(R\)-Levi \(M \subseteq P\), a \(G\)-torsor \(E\) reduces (necessarily uniquely) to an \(M\)-torsor (equivalently, a \(P\)-torsor) iff \(E_{K}\) reduces to an \(M_{K}\)-torsor (equivalently, a \(P_{K}\)-torsor). \item[b)] \textbf{The Colliot-Théléne-Panin Conjecture} For a regular local ring \(R\), its fraction field \(K\), and a reductive \(R\)-group scheme \(G\), if \(G_{K}\) has a proper parabolic subgroup, then so does \(G\); more precisely, if \(G_{K}\) has a parabolic \(K\)-subgroup of a fixed type, then \(G\) has a parabolic \(R\)-subgroup of the same type. \item[c)] \textbf{The Nisnevich Conjecture:} (as reformulated by the author) For a regular local ring \((R, \mathfrak m)\), a regular parameter \(r \in \mathfrak m\), and a totally isotropic reductive \(R\)-group scheme \(G\), every generically trivial \(G\)-torsor over \(R[\frac {1} {}r]\) is trivial: \(\mathrm{Ker}(H^{1}(R[\frac {1} {r}], G)\to H^{1}(K, G)) = {*}\), where \(K=\mathrm{Frac}(R).\) \end{itemize} \end{itemize} The author discusses the Horrocks Phenomenon for totally isotropic reductive group schemes, the Bass-Quillen conjecture for general reductive groups, the passage to the affine space, etc. He reviews the necessary tools: the Geometric Presentation Theorem, excision and patching techniques, the Popescu Theorem, Lindel's Lemma, etc.