关于近期Fano流形上构造Kahler-Einstein度量 |
送交者: 文艺90后 2013年08月31日06:27:19 于 [教育学术] 发送悄悄话 |
http://blog.sciencenet.cn/home.php?mod=space&uid=87484&do=blog&id=721234 关于近期Fano流形上构造Kähler-Einstein度量的工作 最近公布的Fano流形上构造Kähler-Einstein度量的工作,是Kähler几何近年来引人注目的进展,专家们正在验证。若验查无误,将证明丘成桐关于Fano流形的构想与猜测是正确的。Donaldson的稳定性条件是其中的关键步骤,还需在代数几何上把此概念搞清楚,这样丘猜测就为深刻理解Fano流形奠定了基础。由于近期发生了一些混淆不清的事件,我们将相关工作的公开记录做了客观、学术的分析,望有助于澄清事实。本文主要涉及文献的比较,阅读本文无需是专家,数学专业本科高年级学生或研究生可读懂绝大部分。欢迎关于数学上的批评与指正。 本文分三个部分: 1) 陈-Donaldson-孙的报告与文章 2) 田的报告与文章 3) 结论
I. 陈-Donaldson-孙的报告与文章 在最近的一系列文章中,陈秀雄-Donaldson-孙崧(CDS)宣布解决了Kähler几何中悬置多年的问题。 丘成桐猜测:设M为一紧致Kähler流形,其第一陈类为正。此流形上有Kähler-Einstein度量当且仅当流形是K-稳定的。 Kähler-Einstein度量在Kähler几何的研究中和弦理论的研究中起极其重要的作用。陈类为零的情形的Calabi猜测在1976年被丘成桐解决,这类流形称作Calabi-Yau流形,在弦论中是内禀空间的主要候选者。陈类为负的情形被丘成桐和Aubin解决。丘猜测的意义是回答了Kähler流形在陈类为正的时候(也称为Fano流形)何时有Kähler-Einstein度量。 S-T. Yau. On the Ricci curvature of a compact Kähler manifold and the complex Monge-Ampere equation, I∗. Comm. Pure Appl. Math. 31. 339-441, 1978. http://www.maths.ed.ac.uk/~aar/atiyah80.htm (Don-Atiyah, April 22, 2009) http://www2.imperial.ac.uk/~skdona/KENOTES.PDF (Don-KEY, October 19, 2009) http://www.math.northwestern.edu/calendar/complex_geometry_conference.html (Don-NW, October 24-27, 2009) http://xxx.lanl.gov/abs/1007.4220 (Don-Stab, July 23, 2010) http://xxx.lanl.gov/abs/1102.1196 (Don, February 6, 2011) http://xxx.lanl.gov/abs/1104.0270 (Don-Chen1, April 21, 2011) http://xxx.lanl.gov/abs/1104.4331 (Don-Chen2, April 21, 2011) http://xxx.lanl.gov/abs/1112.1594 (Don-Chen3, December 7, 2011) http://xxx.lanl.gov/abs/1206.2609 (Don-Sun, June 12, 2012) http://xxx.lanl.gov/abs/1210.7494 (CDS, October 30, 2012) http://xxx.lanl.gov/abs/1211.4566 (CDSI, November 19, 2012) http://xxx.lanl.gov/abs/1212.4714 (CDSII, December 19, 2012) http://xxx.lanl.gov/abs/1302.0282 (CDSIII, February 1st, 2013) 在陈类为正的情形,人们发现构造Kähler-Einstein度量有障碍。丘成桐意识到这类度量的存在与稳定性有关。这个想法受到他与Donaldson-Uhlenbeck合作的解Hermitian-Yang-Mills方程工作的启发。 S. T. Yau, Open problems in geometry, Problem 65, Proceedings of Symposia in Pure Mathematics, Vol 54 (1993), AMS. “Ten years ago he [Yau] also shared with me his belief that the problem would be related to certain stability properties of the underlying manifolds.” (Tian’s response to receiving Veblen prize in 1996, http://www.ams.org/notices/199603/comm-veblen.pdf).
“The idea that the appropriate condition should be in terms of “algebro-geometric stability” was proposed by Yau about 20 years ago [20]” (CDS, p1)
“The conjecture was refined considerably by work of Tian in the 1990’s [14]…. In Tian’s original definition of K-stability the destabilising objects were projective varieties, smooth or mildly singular, with holomorphic vector fields. In the generalisation of [4] the destabilising objects were allowed to be general schemes with C∗ actions. ” (Don-Stab, p2) “[14] Tian, G. Kahler-Einstein metrics of positive scalar curvature, Inventiones Math. 130 1-57 (1997)” (Don-Stab, p26)
“[4] Donaldson, S. Scalar curvature and stability of toric varieties Jour. Differential Geometry 62 289-349 (2002)” (Don-Stab, p25)
田和Donaldson引入的K-稳定条件,尽管名称一样,有着本质的区别。Donaldson引入的K-稳定条件基于他发现的用Kodaira嵌入计算Futaki不变量,因而是纯代数几何的,也正是丘所期望的。田的K稳定条件则基于一个定义在无穷维空间上的泛函计算Futaki不变量。这样田的条件允许的不稳定对象是一些射影簇,而Donaldson的则允许广得多的对象。 Donaldson在Don-Stab中提出的K-稳定条件,是最后的证明中用到的条件。他在这篇文章里也给出了在自同构群可约的情形,构造不稳定构型的方法。这个步骤也是完成最终证明的一步。假定丘猜测是正确的,从代数几何的角度更好地理解Donaldson引进的K-稳定条件,是深刻了解Fano流形的关键一步。
Donaldson在2009年暑期Atiyah会议上公布了解决此问题的方案(Don-Atiyah)。他建议用带锥的Kähler-Einstein度量来构造Fano流形上的Kähler-Einstein度量,并一直与陈秀雄研究此问题。在丘成桐解决Calabi猜测的文章里引进并研究了这类度量。
“The case we have primarily in mind is when X is a Fano manifold, D is an anticanonical divisor and the metrics are Kähler-Einstein; the motivation being the hope that one can study the existence problem for smooth Kähler-Einstein metrics on X (as a limit when β tends to 1) by deforming the cone angle. … (omitted). In further papers with X-X Chen, we will study more advanced and sophisticated questions.” (Don, p1)
Donaldson-陈秀雄做了一系列工作为致力于解决丘猜测(Don-Atiyah,Don-KEY, Don-NW, Don-Chen I-III)。Donaldson-孙崧在2012年6月份的一篇文章 (Don-Sun) 也是导致解决丘猜测的重要一步。长期以来,人们困扰于一列Kähler-Einstein流形的Gromov-Hausdorff极限改变了复结构。这篇文章对于光滑的Kähler-Einstein流形,在适当的有界条件的限制下,证明存在收敛的子列,其极限是代数的。
2012年10月30日,陈秀雄-Donaldson-孙崧(CDS)公布了他们解决丘成桐猜测的概览。丘成桐猜测在Donaldson的框架下,得到了解决。为解决此问题,需要若干关键的步骤和完成这些步骤的重要细节。这里列举其中的一些关键步骤。
在Donaldson解决此问题的框架下,需要将Don-Sun的工作拓展到带锥的Kähler 流形。在CDS解决丘猜测的概览中,他们给出了关键的步骤。这里关键是引入好的切锥的概念。证明的细节见CDSII。 “We say such a tangent cone C(Y) is “good” if the following holds. For any η > 0 there is a function on Y supported in the neighborhood of the singular set S(Y), equal to 1 on some neighborhood of S(Y) and with the L2 norm of its derivative less than η. One main technical result is that in fact all tangent cones are good. Given this, the arguments of [10] extend almost word-for-word.” (CDS p3-p4, [10] = Don-Sun) 在CDS中作者还给出了另一个关键性的步骤。完成这个步骤的详细证明出现在CDSIII中。 “Case (3), which is the crucial issue, involves two main difficulties. (Case (2b) is covered by the same discussion.), … (omitted) However it is true, by the Luna Slice theorem and the Hilbert-Mumford theorem applied to a slice, if we know that the automorphism group of (W, Δ) is reductive. Thus we need to prove • Aut(W, Δ) is reductive. • The Futaki invariant Futβ∞ vanishes.” (CDS, p4-p5) 完成这些步骤,需要做一些实质性的工作,不能简单地把光滑的情形搬过来。例如证明极限流形的自同构群的可约性,在带锥的度量下,就出现了新的困难,不能做分部积分。过去的方法是证明李代数的可约性,在奇异的情形,人们必须引进新的方法,直接证明自同构群的可约性。新的方法用到丁泛函的凸性和次调和函数的性质。 “This is a variant of the standard Matsushima Theorem, which asserts that the automorphism group of a manifold with a smooth Kähler-Einstein metric is reductive; the new feature being that the proof operates with the Lie groups rather than their Lie algebras.”(CDS, p6)
“One technical tool in the study of weak (conical) Kähler-Einstein metrics is the convexity of a functional defined by Ding [16], as discovered by Berndtsson [6].” (CDSIII) II. 田的报告与文章 Donaldson-Chen的系列工作(Don-Atiyah, Don-KEY, Don-NW, Don-Chen I-III) 为解决丘猜测建立了基本框架。Donaldson-孙菘工作的出现,使人们看到了在Fano流形上解决丘猜测的曙光。田开始加入,田在石溪的一次会议上宣称解决了此问题。在报告结尾石溪数学系LeBrun教授问田的工作与Chen-Donaldson的工作有无联系,田回答说不知道他们的工作。下面我们看到田的文章基本上是按照Donaldson的纲领证明丘的猜测但缺乏关键的细节。田称此问题为Folklore 猜测, 而1996年他在Veblen获奖应答中明言丘先生10年前就告诉他此问题能否有解取决于流形的稳定性。田在KE的主要工作似都是在1997年以前做的。 http://www.math.sunysb.edu/Videos/Cycles2012/video.php?f=14-Tian (Tian-SB, October 25, 2012) CDS在2012年10月30日的一篇文章中公布了他们解决此问题的要点。CDS的全文分为三个部分公开。在CDS丘猜测的证明概要公布三周之后,田在网上公布了自己的文章,也宣称解决了此问题。在CDS第二部分公布后,田公布了修改后的文章。 http://xxx.lanl.gov/abs/1211.4669 (Tian1, first version: Nov. 20, 2012, Tian2, second version: January 18, 2013) 田的文章基于Donaldson提出的方案,但是缺少关键性的步骤和解决这些问题的细节。这几个关键的步骤首次出现在CDS的证明概要文章中,解决这些问题的的细节则出现在CDS后面的三篇文章中。田在第一篇文章中加入了石溪报告中没有的、出现在CDS中的关键性步骤, 在后面的修改中加入了一些首篇文章中没有的、CDS中出现的解决问题的细节。田在爱丁堡的报告中,就一个要点讲了CDS的证明而未给出出处,且承认自己的证明技术上有困难。 http://www.icms.org.uk/workshops/ricci (Tian-EB, July 8th, 2013) 在田的石溪报告中没有“好的切锥”的定义。在CDS的工作中这个性质对于证明一列带锥Kähler-Einstein度量的极限是代数的至为关键。 CDS文章中好的切锥在田的文章中变成引理5.8 (Tian1, 22页)。 田在Tian1中给了半页的证明。在CDS的第二篇文章出现后,Tian2中给出了更长的证明。 在田的石溪报告中没有涉及下面两个带锥Kähler-Einstein流形极限的重要性质: 。全纯自同构群是可约的, 。Futaki不变量Futβ∞为零。 “Lemma 6.3. The Lie algebra η∞ of G∞ is reductive.” (Tian1, p28; Tian2, p35) 在Tian1 和 Tian2 中,田都试图证明李代数是可约的。这个证明有个初级错误。
“Therefore, if we normalize X by multiplication by a complex number such that supM∞ θ∞ = 1, we want to show that the imaginary part of X is Killing. “ (Tian1, p28-29) 因为θ∞是个复值函数,我们无法取上确界。李代数可约性的证明是错误的。田在其2013年7月8日爱丁堡的报告上承认这个困难。在此报告中田讨论了CDS的证明,但未指出其出处。详情见下: “Therefore, it suffices to produce a special degeneration of M to M1. Since M1 is in the closure of the orbit of M [in] CPN under the action of SL(N + 1), we only need to show that the automorphism group of M1 is reductive. This can be deduced from the uniqueness theorem due to Berndtsson and Berman. There is also a more direct proof.”(Tian-EB, part I) 上述两行明显来自于CDS。 “Then the uniqueness result of Berndtsson [4], as extended in [3], can be used to show that the automorphism group is reductive.” (CDS) “Remark: If M1 is smooth, then by standard arguments, one can prove that the group is reductive. But if M1 is singular, one needs to pay attention to a technical problem caused by the singularity.” (Tian-EB, part I) 如CDS所指出的,由于奇点的出现,可约性的证明只能对自同构群做,不能直接对李代数做。在Tian1 和 Tian2 中,田都试图证明李代数是可约的。
在Tian1, 28页田用半页讨论广义Futaki不变量为零。 “In our case, though ω∞ may not be smooth along D∞ even in the conic sense, using the Lipschtz continuity of θ∞, one can still prove the vanishing of fM∞,(1−β)D∞(X) by the same arguments as in the smooth case.” (Tian1, p29; Tian2, p36) 上述一段是我们从Tian1和Tian2中找到的所有的极限流形的Futaki不变量为零的证明。在CDSIII的文章中他们花了很多篇幅证明这一关键的步骤(CDSIII,p31-36, p39-46).
Tian1在关于带锥奇点度量的锥角的计算也犯了初级错误。 “C4. For any x ∈ S2n−2, Cx = C′x × Cn−1, where C′x is a 2-dimensional flat cone of angle 2mπβ∞ for some integer m ≥ 1. (Tian1, p12)
“In our new case, the conic singularity of ωi along D may contribute a term close to 2mπβi in the slicing argument, this is how we can conclude that C′x is a 2-dimensional flat cone of angle 2mπβ∞.” (Tian1, p12)
“Lemma 5.5. For … (omitted), which is biholomorphic to C, of angle 2mπβ∞ < 2π. …(omitted) the Euclidean metric on Cn−1 with a conic metric on C′x of angle 2mπβ∞ < 2π.” (Tian1, p19)
“Remark 5.6. It follows from the volume comparison that 2mπβ∞ ≤ 2π. But x or y is a singularity, so 2mπβ∞ < 2π.” (Tian1, p19)
“(1) There is a tangent cone Cx of the form Cn−1 × C′x for a 2-dimensional flat cone C′x of angle 2mπβ∞, where m ∈ Z;
(2) … (omitted), any tangent cone Cy of Cx at y is of the form Cn−1 × C′y for a 2-dimensional flat cone C′y of angle 2mπβ∞.”(Tian1, p21)
在Tian2 上述角度的错误被称为打字错误,更正为“(1 − “Proposition 13: We have the identity (1 − γ) = μi(1 − β∞).” (CDSII, p16) 在Tian1和Tian2中我们没有看到定理3.2的证明。文章中只有一句话: “Using the same arguments in [CCT95], one can show: Theorem 3.2” (Tian1, p14) 注记5.6从Tian1到Tian2修改如下:
“Remark 5.6. It follows from the volume comparison that 2mπβ∞ ≤ 2π. But x or y is a singularity, so 2mπβ∞ < 2π.” (Tian1, p19)
“Remark 5.6. …(omitted) If β∞ < 1, since (1 − μa) = ma(1 − β∞) for some integer ma, there is a bound on l as well. In fact, one should be able to prove that there is a uniform bound on l depending only on λ.” (Tian2, p22)
从注记5.6我们看到田意识到正确的角度公式在β∞ < 1的情形给出重数的上界。他仍未意识到在β∞ = 1时重数可以是无穷。这两种情形CDS在CDSII和CDSIII中分别作了研究。
“In this paper [CDSII] we consider the case when the limit β∞ is strictly less than 1. In the sequel [CDSIII] we will consider the case when β∞ = 1 and also explain, in more detail than in [8], how our results lead to the main theorem announced there.” (CDSII, p1-2)
数学家可以引用已有的结果,前提是这个结果是正确的。 “A general result of Evans-Krylov type is stated in [13] where two independent proofs are given, but at the time of writing we have had difficulty in following the one of these proofs that we have, so far, studied in detail. Partly for this reason, and partly because it has its own interest, we give a different approach (to achieve what we need) below.” (CDSII, p31) “[13] T. Jeffres, R. Mazzeo, T. Rubinstein. Kähler-Einstein metrics with edge singularities. arXiv: 1105.5216” (CDSII) 田的文章依赖于他们的结果。
“If supM ϕi is uniformly bounded, by the
Harnack-type estimate in Theorem in [JMR11], the C0-norm of ϕi is uniformly bounded. So, by
[JMR11] again, ϕi converge to a solution of (6.1)
for β =
“By our assumption
and the results in [JMR11], ||ϕβ||C0 diverge to ∞ as β tends to III. 结论
比较两组作者的工作时我们发现两个区别。第一,CDS的工作基于Donaldson几年前提出的方案,在他们的系列文章中给出了关键的步骤和详尽的细节,对于中间出现的困难给了充分的处理。他们建立了工作的基本框架,并给出了解决问题的细节,这两者都算是原创性的贡献。
第二,这两组工作公之于众长达数月的时间。在这段时间里,有证据显示,一个组的工作不断地从另一组吸取想法。CDS近年来致力于此问题,建立了解决问题的框架,并在关键性问题上取得了突破。CDS的全文分三部分发表,全部公开用了十周。田的论文第一稿在CDS的工作第一部分三周后公开,加入了许多在石溪报告中声称解决此问题而没有出现的关键步骤。田在CDS第二部分公开后又有修改稿。如前所示,田每次做的修改,其方法和技术都是从前面CDS公开的论文中借用的。
如前所述,伴随着丘猜测证明出现的一些事件使得情况显得混乱。我们从公开记录的分析得出的结论是陈-Donaldson-孙对此问题做出了原创性的、创造性的贡献。
中国科学技术大学教授 普林斯顿大学数学博士 胡森 |
|
|
|
实用资讯 | |
|
|
一周点击热帖 | 更多>> |
|
|
一周回复热帖 |
|
|
历史上的今天:回复热帖 |
2012: | 稿件水平低,没有兴趣,不重要,用不着 | |
2012: | 姚明距离美国绿卡还有多远? | |
2011: | QWE: 科技的档次,如何提高中国科技竞 | |
2011: | 答MC:关于基因决定行为的实验 | |
2010: | 紫才女,你看我的猜想是不是make sense | |
2010: | 光子的3个奇妙特性(旧文重贴) | |
2009: | 基甸恩典:言正轻君,你听我说说来龙去 | |
2009: | 彩虹之约封了“嘁哩喀喳”笔名,上帝很 | |
2008: | 秀才,你如何看蒋介石政府聘请日本战争 | |
2008: | 热帖:德国媒体为什么如此疯狂地攻击中 | |