Murthy's conjecture on 0-cycles (Q2315189)

During the early 60's, two basic results on finitely generated projective modules over Notherian rings showed the researcher a clear way to proceed towards the proof of J.-P. Serre's problem on projective modules; viz. Serre's splitting theorem and Bass's cancellation theorem. In this paper, the author concentrated on the problems which arose from the first result. Serre's theorem says that a finitely generated projective module $P$ over a commutative Noetherian ring $R$ always contains a unimodular element if $\operatorname{rank}P = r > \mathrm{Krull} \dim(R) = d$, i.e., there exists an element $m\in P$ such that $R.m$ is a free summand of $P$ with basis $\{m\}$. The case $r \le d$ was studied by M .P. Murthy for affine algebras over algebraically closed field. Subsequently related problems were studied by S. M. Bhatwadekar, R. Sridharan, M. K. Das, et al. However, question is still open for the general case. Let $A$ be a reduced affine algebra of dimension $d\ge 1$ over an algebraically closed field $k$. Let $\mathrm{K}_0(A)$ denote the Grothendieck group of finitely generated projective $A$-modules. Recall that in this case every maximal ideal in $A$ of height $d$ whose localization is regular has a class in $\mathrm{K}_0(A)$. Let $\mathrm{F}^d\mathrm{K}_0(A)$ be the subgroup of $\mathrm{K}_0(A)$ generated by these classes. Also, recall that a projective $A$-module of rank $d$ admits a Chern class $c_d(P)$, and if $A$ is smooth, $c_d(P)$ maps to the top Chern class of $P$ in the Chow group $\mathrm{CH}^d(A)$ via the Chern class map $c_d: \mathrm{K}_0(A)\to \mathrm{CH}^d(A)$. Given a quasi-projective scheme $X$ of dimension $d$ over $k$, let $\mathrm{CH}^d(X)$ denote the Levine-Weibel Chow group of $0$-cycles on $X$. We write $\mathrm{CH}^d(X)=\mathrm{CH}^d(A)$, when $X=\mathrm{Spec}(A)$. In his seminal work, Murthy asked if $\mathrm{F}^d\mathrm{K}_0(A)$ is torsion free for $d\ge 2$. And, assuming an affirmative answer he showed that $P$ splits off a free summand of positive rank if and only if $c_d(P)=0$ in $\mathrm{K}_0(A)$. In this paper the author answered Murthy's question affirmatively. In fact, he has deduced a more general result. He has solved a closely related problem, \textit{viz. affine Roitman torsion problem} which asks if $\mathrm{CH}^d(X)$ is uniquely divisible for $d\ge 2$. Then, he has answered Murthy's question by deducing that $\mathrm{Ker}(\mathrm{cyc}_A)$ is a torsion group of exponent $(d-1)!$, where $(\mathrm{cyc}_A): \mathrm{CH}^d(A) \to \mathrm{K}_0(A)$ is a cycle class map. In Levine's proof he repeatedly used the fact that $\mathrm{Pic}(C)(k)$ is an $n$-divisible group for any reduced curve $C$ over $k$. But this idea is not applicable if $\operatorname{Ch}k =p >0$. The proof in this paper is based on the induction on $d$, and the torsion property of $\mathrm{Pic}(C)$ and its bounded component. For the $d=2$ case, key ingredients are $(i)$ results of Geisser and Hesselholt, $(ii)$ divisibility property of $\mathrm{SK}_1$ of certain normal surfaces and some other useful facts, $(iii)$ Mayer-Vertories sequence. Generalization for the higher dimension comes mainly through the reduction step which is similar as Levine's argument. Here the author has used Block's trick of blowing up $X$, and then Bertini theorems on complete intersection. Following results are obtained as consequences of the above results. \begin{itemize} \item[1.] For a reduced affine algebra $A$ of dimension $d\ge 2$ over over an algebraically closed field $k$, the Euler class group $E(A)$ is isomorphic to $\mathrm{CH}^d(A)$. \item[2.] Partial answer of a conjecture by Beilinson-Bloch on complete intersection. \item[3.] A stronger version of Bloch-Srinivas conjecture on Cohen-Macaulay singularities in positive characteristic: Let $X$ = affine or a projective variety of dimension $d\ge 2$ over an algebraically closed field $k$, with $\operatorname{Ch}k=p >0$. Suppose $X$ has only isolated Cohen-Macaulay singularities. If there is a resolution of singularities $\pi: \tilde{X}\to X$ with reduced exceptional divisor $E\subset \widetilde{X}$, then one gets \[ \mathrm{CH}^d(X)\cong \mathrm{F}^d\mathrm{K}_0(X)\cong \mathrm{F}^d\mathrm{K}_0(\tilde{X}, E). \] \item[4.] Under certain assumption the restriction map $\mathrm{CH}^d(\widetilde{X}, mE)\to \mathrm{CH}^d(\widetilde{X},E)$ is an isomorphism for all $m\ge 1$. In particular, the pro-abelian group $\{\mathrm{CH}^d(\widetilde{X}, mE)\}_{m\ge 1}$ is constant. \end{itemize}