五千年(敝帚自珍)

主题:【原创】围绕脑科学而发生的若干玄想 -- 鸿乾

共:💬461 🌺824 🌵2
分页树展主题 · 全看首页 上页
/ 31
下页 末页
        • 家园 也许你需要换个思路了

          这是几个小时前刚出炉的, http://www.insidescience.org/content/physicist-proposes-new-way-think-about-intelligence/987

          The researchers suggest that intelligent behavior stems from the impulse to seize control of future events in the environment. This is the exact opposite of the classic science-fiction scenario in which computers or robots become intelligent, then set their sights on taking over the world.

          paper还有paywall现在还拿不到,等晓兵给做个初步解读。

          • 家园 元首一号命令:白軍對tg, 决战千里之外

            that is a good one, thx, 元首

            two disclaimers:

            1)

            决战 here=忽悠, in the foreseeable future, tg=king of china, if not 4ever, and in the mind of majority of Chinese people, tg=china, kind of the language post author's "the Chinese pride is in our blood" type of 6k years Chinese cultural brain formation, with very little 色散 if any, particularly with tg's GFW and SR weaponry, qm hedge and sr hedge, double hedging, smarty tg.

            and for that reason alone, white is going to and must avoid any 正面冲突 with tg/china, so,决战千里之外 is white's long term strategy, 千里 in 相对论四维时空.

            and for the almost same reason, unless totally out of options, tg will not leverage on/invoke Chinese people's tg=china mindset, which is a powerful weapon of last resort type, but basically one time use only, and it could easily backfire, and how do you hedge such a backfire? no hedging? then no fooling around: unlike stupid KMT, tg has been hedging almost from day one, the so-called 统一战线法寶.

            so the fishing island type of smokes are all fake and cheap shots, for 白軍 and tg to play with each other;

            but north korea is a wild variable, what if it falls apart? neither 白軍 nor tg like that if, so it better not fall apart.

            2)

            I 不作 physics research好多年, I hope more folks here in this forum comment and correct me

            that said, 白軍對tg: 决战千里之外 is still the story of the century, with outcome/impact to be seen N years/decades/centuries later.

            now lets go to the core of 白軍 logic

            http://www.ccthere.com/article/3866872

            "90岁彭桓武:对爱因斯坦两段话,至今未能真正理解"

            爱因斯坦's methodology if I may 妄加评论: 自然界统一性 and 逻辑的简单性 as "物理定律",坐标系变换不变, such as SR's 洛伦兹变换, and GR/gauge field theory's相位变换, etc, basically:

            a from logic (of course assisted/comibined with 定向实验) 求实事 model;

            I remember reading somewhere that 爱因斯坦 said that he would not even contemplate a 实验, if such 实验 could not pass his logic test first.

            "I have read some of the posts by

            witten1, changshou, and many others, very good postings on key

            concepts such as 系综, 么正性, gauge, 測度, 度規, etc, the key physics/math aspects of the core of the white logic, I hope we have more of those kind of postings"

            physics is all about energy (can we do anything without energy?), with 最小作用量原理, as one of the "cores of the white logic" to start with.

            1.

            for a system to survive: it has to use 最小作用量, with 最大范围( to get information, often "最大體積" in 洛伦兹流形) 积分, kind of like the plant's animal behavior you guys discussed, and we could call that as plant's 最小作用量原理 or plant's AI, and from plant's AI to 費曼路徑積分, etc

            2.

            if we can manage "1" part, we basically have system's 么正性,因果关系, 逻辑链条 established, in terms of defining 測度 with 機率 distribution in a well defined (L2 可积 mathmatically,or 位能場平方力 physics model wise) hilbert space, 冯-诺伊曼's 泛函, etc;

            and such a system is normally called a "canonical (平衡正則) system", and within the system, we can calculate from 因 to 果, or from 果 to 因, classical 哈密顿系统辛流形, and going further into QFT under 相对论框架 , we could have 虚時間,实時間, 虚粒子,实粒子, off the shell, on the shell, and their 么正性,因果关系, 逻辑链条, etc;

            QFT, gr are all well established and well 实验/觀測(for gr)证明 theory, of course, with 引力场量子化, full interaction partition function =still dreams .

            3.

            so, logic wise, a "canonical (平衡正則) system"=数学完備 ,精确求解, 解析解, more likely 数值解, theoretically, because with with SR, QFT, GR or factored in , 么正性,因果关系, 逻辑链条 can even survive a situation such as "black hole": going through "black hole", system's 信息 not lost, and system can still manage to interact and feed back with the environment, surviving and moving on, I remember witten1 commented about "么正性@blackhole" somewhere.

            4.

            taking the above to the extreme, folks like 霍金 can almost model every system (still not a full interaction model)with certain assumptions about the system.

            "虚时路径积分就是正则统计,统计的样本是一个个热力学系统。一个系统给你,你知道它的Hamiltonian,那么你可以计算 ∫dτ H, 就是那个热力学系统的虚作用量,然后放到路径积分的框架下,就可以算各种平均测值和关联函数了。

            所谓热力学路径积分,统计的是一个个(连接平衡态的)热力学过程。给你一个热力学过程,你可以计算这个过程的熵产生 S,这就是热力学过程的虚作用量,然后同样放到路径积分框架下,就可以开始统计你要的东西"

            热力学还未发展完备? - 豆瓣

            www.douban.com/group/.../7857358/ - 中華人民共和國 - 轉為繁體網頁

            2009年9月1日 – 虚时路径积分就是正则统计,统计的样本是一个个热力学系统。一个系统给你,你知道它的Hamiltonian,那么你可以计算∫dτ H, 就是那个热力学 ...

            5.

            I have not read the paper you quoted yet, but the above physics models'逻辑链条 should be the starting point of their QFT based AI model, and with that model思路

            , physics start penetrating social sciences, a great start

            6.

            6.1

            I want to say it again and again, so far 白軍 logic is the only logic we have at humanity level, in terms of physics/mathematics, and we have to learn and understand it, period.

            human's technological and industrial civilization as we know is the accomplishment of 白軍'from logic 求实事 model of physics/mathematics, period;

            6.2

            now 白軍's model starting getting into social science, big deal, so far economics is based on 白軍's old model of classical physics.

            looks now they are start thinking about an upgrade.

            7.

            for tg to deal with 白軍, tg has to be extremely careful with its current "人海战术实事求是"@ 1 政委统帅下 type of model,

            vs white's from logic (of course assisted/comibined with 定向实验) 求实事 model, with the framework of 市場, 科技, US 4 政委(ws, wh, media, Hollywood) parallel

            can tg get it?

            中国科学院 理论物理所: basically 一代不如一代, the soviet and west educated those few top guns such as 周光召 etc, outstanding, but they are aged now, and I remember reading somewhere that he said very few in china actually really understand SR, not to mention gr, "changshou" post about GR is a pretty good one, I am not sure how many Chinese professors back home can write at his level, and witten1 has written quite a few post about qm, etc.

            I hope more folks could write like that

            -------

            witten1

            http://www.ccthere.com/alist/3830412

            (注:量子力学的Bohm隐参数解释能给出所有正统量子力学的预言,当然出没有给出更多的预言)。确实没有论据可以断言一个能自洽的对量子力学的非定域本性给出决定论式的解释是不存在的。既然Bohm给出了一个,那它就是有可能存在的。(注:个人认为隐参数的量子理论没多大意思,隐参数是怎么?没有人说的清,这就像人们依然没有真正理解量子理论的非定域本性一样。你多了一个自由度--隐参数,显然你可以包含的东西会多得多,这没多大意思。多提一句,到今天为止绝大多数的隐参数量子理论都被实验证否了,这使得那些喜欢玩概念的的就加上越来越“强大”的隐参数,让实验更难区分。。。比如前一阵的Gisin的实验证实任何包含有限速度(可以比光速快)的隐参数理论都是不对的,推论就是,隐参数理论得引进一个在逻辑上更加荒谬的无穷的大的传播速度来拯救自己。)

            changshou:几何直观地介绍广义相对论的时空以及大爆炸模型(0)分页 ...

            www.ccthere.com/topic/3659016/19 - 轉為繁體網頁

            14 篇文章 - 4 位作者 - 2012年6月12日

            changshou 趋加追订 屏蔽 2012-06-12 21:48:02 3738646 复3738572. 你修的是狭义 ... http://www.ccthere.com/article/3193190. “3,《三体》(一) ...


            本帖一共被 2 帖 引用 (帖内工具实现)
            • 家园 晓兵是发散思维的高手,这里有个需要发散的东西,请来谈谈

              Fuhrer提到的那个论文,依赖于提升到更高维空间作熵的计算(包括了各种原因等等)。

              我们知道,从高维空间向低维空间投影,有非常现成的理论。通常的光学投影这类的是最简单的。更复杂一些的是所谓的non-metric multidimensional scaling之类的东西。这些东西多半可以归结到各种熵的公式里面。

              但是,这个论文的作者是反过来的,从低维空间向高维空间嵌入,是否有类似的熵公式来起作用呢?

              这是极端发散的问题,肯定没有答案的。不过,可能产生若干思想火花也未必。

            • 家园 Paper也出来了,可以在这里看

              http://www.alexwg.org/publications/PhysRevLett_110-168702.pdf

              Dr. Alexander Wissner-Gross

              http://www.alexwg.org/

              • 家园 引力, 熵力, 智力:"爱因斯坦场方程的解是个场"

                thanks for the link, and I will read it further when I get some time, and I have a lot of too say, for now:

                their paper started right with Erik Verlinde's theory, which I read a little in the fast

                my first few comments:

                1.

                for a system, 智力 =how to get information economically/mathematically/physics smart/logic test pass first/etc

                1.1 a system has to constantly 消噪 as it gets information

                "个很自然的想法 花1 花大熊 字155 2012-12-31 22:01:32

                ...O Algorithm as music," 消噪" 花1 晓兵 字813 2013-02-15 19:39:25

                1.2

                very often, we human being is a 耗散結構/Dissipative structure, we have to constantly breath in "information", but very often what we actually breath in is "noise"

                so, potentially, this kind of app actually has mental health implication

                2.

                so we have to know 场方程 first, to 决战千里之外

                3. and for complicated system such as 引力场, heat"场" , so far we can only do幾何引力场, 系综统计heat"场",

                4

                we have to have 量子化 methodology, then you can contemplate 定向实验, and eventually do apps if model is developed

                5. I will read the paper further, and come back at a later time

                6.

                引力是熵力吗? | 博客李淼

                limiao.net/1398 - 轉為繁體網頁

                7. it is all about brain living 幾何引力场, 系综统计heat"场",

                and how do we 测信号?

                8.

                how do we handle those 测不到的信号? such 引力

                in a social 引力场, without cold war or brutal killings of physical body, it is all about 信号 processing

                etc

                a great post:

                【原创】数值解爱因斯坦场方程为啥难 [ 测不到的信号

                ----------------

                2010年1月12日 – ppt文件:引力是“熵力”吗? 附注1:. 文章出来之后,Verlinde在自己的主页上加了解释,我想很好。其主要点是将空间(其中一维)当着热力学宏观量, ...

                "引力的熵力假说" [ 晓兵 ] 于:2013-01-18 23:13:41 复:3836575

                but how to quantify partition function would be an issue, among other issues, but intuitively, it makes sense to me, now

                引力的熵力假说- 维基百科,自由的百科全书

                zh.wikipedia.org/zh-hk/引力的熵力假说 - 轉為繁體網頁

                引力的熵力假说是一种关于万有引力本质的理论假说,由荷兰弦理论家埃里克·韦尔兰德(Erik Verlinde)于2009年提出。牛顿的万有引力定律与爱因斯坦的广义相对论 ...

                一个很自然的想法 花1 花大熊 字155 2012-12-31 22:01:32

                ...O Algorithm as music," 消噪" 花1 晓兵 字813 2013-02-15 19:39:25

                ....O "[27] C N Yang and R Mills" 花1 晓兵 字130 2013-02-15 19:51:22

                On the Origin of Gravity and the Laws of Newton

                arxiv.org hep-th

                by EP Verlinde - 2010 - Cited by 350 - Related articles

                Jan 6, 2010 – High Energy Physics - Theory ... Theory (hep-th). Journal reference: JHEP 1104:029,2011 ... From: Erik Verlinde P [view email] [v1] Wed, 6 Jan ...

                http://staff.science.uva.nl/~erikv/page1/page1.html

                Erik Verlinde's theory

                In 2009, Erik Verlinde disclosed a conceptual model that describes gravity as an entropic force.[4] On January 6, 2010 he published a preprint of a 29 page paper titled On the Origin of Gravity and the Laws of Newton.[5] The paper was published in the Journal of High Energy Physics in April 2011.[6] Reversing the logic of over 300 years, it argued that gravity is a consequence of the "information associated with the positions of material bodies". This model combines the thermodynamic approach to gravity with Gerardus 't Hooft's holographic principle. It implies that gravity is not a fundamental interaction, but an emergent phenomenon which arises from the statistical behavior of microscopic degrees of freedom encoded on a holographic screen. The paper drew a variety of responses from the scientific community. Andrew Strominger, a string theorist at Harvard said “Some people have said it can’t be right, others that it’s right and we already knew it — that it’s right and profound, right and trivial."[7]

                -----

                【原创】数值解爱因斯坦场方程为啥难 - 西西河

                www.cchere.com/article/3013365 - 轉為繁體網頁

                2010年7月15日 – 【原创】数值解爱因斯坦场方程为啥难 [ 测不到的信号 ] 于:2010-07-15 03:19:27 复:3012204. 我听说“数值广义相对论”这个小方向从2005年以后变得 ...

                原创】数值解爱因斯坦场方程为啥难 [ 测不到的信号 ] 于:2010-07-15 03:19:27 复:3012204

                我听说“数值广义相对论”这个小方向从2005年以后变得比较热门,是因为你说的这个突破吗?

                没错,就是因为2005-2006年的这个突破。第一个成功的数值模拟是Caltech的一个博士后几乎单枪匹马完成的。近几年来,数值相对论产生了许多重要结果,既包括相对论或引力物理方向的,也包括天体与高能物理方向的。竞争也一度白热化,曾发生几个小组在某会议前三天争相在物理学评论快报上就同一问题灌水的事儿。

                另外还想请教一下,数值解爱因斯坦场方程的难点在什么地方呢?我相信肯定是来源于非线性,但是具体点儿说呢?后来他们怎么突破的?

                这个我也不太懂,随便说说。

                的确,非线性是导致数值解爱因斯坦场方程非常困难的根本原因。数值解非线性偏微分方程本身是个大问题。我的理解是,非线性系统的解对初始条件十分敏感。对于混沌的非线性系统,著名的例子就是“蝴蝶效应”:当初始条件无法严格确定的时候,系统的长期演化是不可预测的。即便对于那些不混沌的非线性系统,当初始条件有偏差时,这个偏差通常也会随时间以指数速度放大,导致初条件失之毫厘而结果谬以千里。

                由于数值计算只能实现有限精度,舍入误差不可避免。如果这些误差不受控制的随非线性系统演化,结果可想而知。

                不过仅就非线性带来的困难而言,爱因斯坦场方程也未见得就比著名的磁流体力学方程更吓人。导致数值解爱因斯坦场方程成为极端难题的是非线性系统的共性与广义相对论的个性的结合。爱因斯坦场方程的解是个场(废话)。这个场描述的不是流体的密度、电磁场的强度之类的普通角色,而是时空的几何结构。在广义相对论中,不但物质与能量的发展变化是统一的,物质能量与时空的演化也是一体的。正如 John Archibald Wheeler 教导我们的:

                Spacetime tells matter how to move; matter tells spacetime how to curve.

                在此,时空不再是物质世界永恒不变的背景,它是物质世界本身。具体到工作在数值相对论领域的物理学家们,空间如何延展,时间如何流逝,都是他们在演化爱因斯坦场方程时随时要澄清的。当这些对经验直观产生挑战的微妙问题与方程本身的非线性特性结合时,产生了一系列(以下的例子均为道听途说和个人理解)。

                一,数学没跟上的代价。

                为了进行数值演化,爱因斯坦场方程要被改写成一组(几十个)一阶偏微分方程,而为了保证解的稳定性这个偏微分方程组必须构成一个双曲的柯西问题。不管后半句是啥意思,总之人们直到2000年左右才搞清楚这一点,并意识到大家几乎白忙了快二十年。

                二,黑洞中心的奇点怎么办?

                所有物理量在奇点都是无穷大,这怎么搞数值模拟呢?根据两种不同数值方法的采用,两种方案产生了。

                第一帮人用的是标准的有限元法,朴实无华,程序相对简单,不过运行速度较慢。他们说,奇点不用担心,只要奇点不恰好运行到坐标格点附近估计就没事儿。如果奇点接近了坐标格点,稍微挪挪格点们就成。信不信由你,这么搞居然成功了,尽管事后几年人们才逐渐理解为什么会成功。

                第二帮人用的是波谱法(pseudo spectral method),程序复杂但运行速度极快。不幸的是,直至目前,他们省下来的运行时间基本上都用来写程序了。他们说,物理量变化越平缓光滑波谱法越牛,甭说奇点本身,就是靠近奇点我们都受不了。我们必须把奇点从计算区域中扣掉,并在扣除的边界上加适当的边界条件。事实证明,加边界条件不难,不过尾随奇点的运动并扣掉适当的区域可费了大劲了。至今波谱法程序还时常受此困扰。

                三,动态的规范(guage)条件。

                这个类似于坐标系的选取。对同一个问题,我们可以自由的选取不同的坐标系,写出不同坐标系下长相不同的方程组,解出不同的数值解,不过有一点可以肯定,我们最终会得到同样的物理过程,同样的因果关系,否则一定有人算错了。你可以把东改成叫“西”,把西改成叫“东”,太阳该从哪边升起还是从哪边升起。当然,恰当的选取坐标系可以极大的简化计算。一个球对称的系统你非要用直角坐标那是找事儿。规范自由度与之类似,有些计算中自由度的选取不影响物理结果本身。在此我以电磁场方程作个类比。如果你对洛伦兹规范这个名称不陌生的话大概也会记得麦克斯韦方程组在这个规范下可以写成一个多么简洁而对称的形式。在解爱因斯坦场方程的过程中,恰当的选取规范条件不但可以简化计算,更是决定成败的关键。

                还是以坐标系的选取为例,对于一个球对称的史瓦西黑洞,我们自然是选取球坐标来描述。可是越接近视界的地方,引力场越强,时空的扭曲也越强。视界的位置上,时空度规变成无穷大,时间与空间发生”反转“,这看似是奇点。其实这只是坐标奇点,是由于不恰当的坐标选取造成的。如果我们选取爱丁顿坐标就会发现,在视界的位置上没有任何物理量是无穷大。可见,一个以平时经验看来自然而然的坐标选择当遇到黑洞这种极端环境时,随时可能产生意想不到的问题。更何况这还是面对一个独自静止的最简单的黑洞。想象一下两个黑洞绕在一起的情形,强烈的时空扭曲不断变化,任何初始选取的坐标系统都会迅速和自身纠缠在一起,产生坐标奇点。必须有办法随着双黑洞系统的演化动态的调整坐标和规范的选择。

                四,艰难的满足约束。

                在此我再以电磁场方程作类比。游MM列出了麦克斯韦方程组(Maxwell’s equations)的微分形式。盯着方程仔细看,第一个和第三个是一类,没有对时间的微分。也就是说,它们描述了电磁场在任何时刻都必须满足的约束,但不描述电磁场如何随时间演化。另两个方程自然就是演化方程了。如果一组初始条件满足了约束方程并依照演化方程演化,则麦克斯韦方程组保证它们在演化过程中的任何时刻都依然满足约束方程(坚持读到这的同学可以自行验证)。这其实也是废话。如果满足某个理论约束的系统依照这个理论的要求演化,演化着演化着就不再满足这个理论自己要求的约束了,这个理论就只有悲剧了,它逻辑上不自洽。

                一样地道理,不过更复杂些,爱因斯坦场方程也有自洽的约束部分和演化部分。不同的是数值误差在麦克斯韦电磁场方程和爱因斯坦引力场方程中的行为。由于舍入误差不可避免,约束在数值演化中不可能严格满足。违反约束的部分起初是与数值精度大小相仿的微小误差。在线性的麦克斯韦电磁场方程中,它们随时间演化但绝对大小几乎不变,而且可以被单独分析。在非线性的爱因斯坦引力场方程中,它们随时间以指数增大并以光速甚至超光速在数值模拟区域内四处乱窜,打在边界上还会到处乱弹,难以被追踪控制。它们爆炸性的增大立即导致计算机的溢出错误,扼杀了刚刚起步的数值模拟。人们试图用各种算法追踪并控制违反约束的部分,但每每发现加了复杂控制系统的程序通常死的更快。误差导致的疯狂扭曲的时空看来总是比人们狡诈一些。

                这实际上是最后被解决的问题,也导致了最终的突破。2005年,上文提到的那位博士后使用了广义和谐规范(generalized harmonic guage,中文听起来很强大)改写他的方程组并取得了成功。广义和谐规范在方程组中引入了一些平时没用的项,因为如果约束被满足,这些项都是零。但当违反约束的部分出现时,这些项会让它们产生自阻尼效果,以致违反约束的部分越大,它们对自己的抑制就越强。由于抑制是违反约束的部分自己产生的,它永远不会像人工控制那样被违反约束的部分欺骗。这也再次表明了恰当的规范选择不仅带来便利,更是成败关键。

                五,边界条件的选取。

                上面说过加边界条件不难,这是指在黑洞位置上被挖去奇点的内边界。由于视界的存在,黑洞内部的边界条件最简单:有出没进,只能从模拟区域流向黑洞中心,不能从黑洞中流进模拟区域。远处的边界条件就复杂些,既要允许符合物理要求的出入条件,又要保证违反约束的部分只出不进还没有反弹。

                我知道的大概就这些。最后说一下,双黑洞系统的演化还相对简单些,因为系统中只有纯时空曲率而无普通意义上的物质。现有的热门问题之一是数值模拟一个中子星(脉冲星)被黑洞吞噬或被黑洞的强大潮汐力撕碎。这很可能是伽马射线爆发的机制,也是很有希望被发现的引力波源。就数值方法来说,这也很有挑战性:爱因斯坦场方程和磁流体力学方程这两大难凑在一起了。

                • 家园 "世界是二维的?关于全息屏和作为熵力的引力"

                  I really like this paper, smarty folks, as I said, thought invoking, and how come we Chinese is always behind and play 華車, 璀喇叭胎教子, writing reviews and follow up papers, as 李淼 said, not sure if 李淼 still even read journals anymore

                  1.

                  爱因斯坦场=gravity, 幾何的, what you can do with 幾何?

                  2.

                  热力学: 熵, temperature are all defined and measured by partition function and stat @macroscopic

                  level, are you a sys admin guy with a daily updated password?

                  none of the above, just a plain individual, then

                  熵@micrscopic state level=0,may be that is why we always feel good and pride, in our blood, and most likely it is 幻覺, fake, baby;

                  penrose has been trying to get a more meaningful 熵@microscopic, going no where;

                  3.

                  Verlinde, after his string theory (can never be tested/measured at current energy capability level) went no where, combined the part 1&2

                  so, now with verlinde's theory, we have some kind of 热力学引力場 model with entropy 动力学 between macroscopic and microscopic with that paper as a start, to get into social

                  引力場 动力学, as an appetizer, kind of, before a really good dinner, white's near term dream

                  4. why doing all these?

                  US 4政委 parallel 制度 can never compete with tg's 1 政委 hierarchy structure (result of that is 愚民, very likely) , in terms of social

                  引力場 information collection and analysis, and tg's 经典物理 power in its top-down 1 政委 hierarchy structure , is unmatched globally.

                  so, white has to go social/market powered AI for social 引力場 , etc, so US 4政委 parallel could work together with market and half 愚民 democracy, to compete with tg's 集权制度, to 改良 social 引力場 測度, 改良生產關係, etc

                  5. as a personal mental health advice for all of us:

                  unless you have tg's minister level 组织文件 or US 4政委's expensive AI model, basically all the information we collect and analyze as an individual @microscopic, are very close to garbage in terms of 機率 density distribution function , but most of the time (it should really be the opposite, upside down kind of mess up) we all feel we see and feel something great, because the system wants us all to do that and feel that way and try our best@microscopic, so system can 費曼 路徑積分 @system and macroscopic level, to come out with a collective level AI, and in the process, any individual ant can be sacrificed with no mercy at all, basically

                  and very often, for an individual ant, garbage in =garbage out@紅外, and once 紅外=4 ever紅外

                  ---------------

                  Verlinde,

                  《宇宙的琴弦》读后感

                  sky.hhu.edu.cn/ec.../read-interactive.do?... - 轉為繁體網頁

                  最近有个搞超弦理论出身的荷兰物理学家Erik Verlinde放弃了超炫理论,另辟蹊径,提出万有引力不是基本作用力,就像皮球内部的压力,只是封闭气体分子的集体 ...

                  ----------

                  http://hi.baidu.com/vgukbflktujoxyr

                  ENGINEER波

                  世界是二维的?关于全息屏和作为熵力的引力

                  Verlinde关于引力起源于熵的论文.先要注意的是论文的逻辑链,作者本人就很重视这一点。论文的逻辑起点在全息屏上。假设我们手头上有全息屏上的有时间变量的微观态物理,而且这些定律是时移不变的,那么就有能量守恒了。而且微观态数不是一,那么我们就有了时间、能量、微观态数作为以下理论的起始点,微观的物理是怎样的现在先不关心。总之靠现有的概念在宏观上就有热力学第一定律。因为宏观上可以有温度、熵这样的emergence(这个词译法好多,不知选哪个)的概念。好,现在还没谈及空间,在这里空间也是emergent出来的,在微观上是没有定义好的。不过这里只是一维是emergence的,另外两维被全息屏所占。好,现在类比下理想气体。理想气体的体积在微观里没有很好的定义对不对?而且有个热力学对偶量压强,那么就可以写下热力学第一定律的方程。类似地,全息屏emergent出一维空间x之后,其热力学对偶量就是力,引力。

                  刚才有了热力学第一定律的方程,现在假设一质量为m的粒子在emergent出来的方向上做个小位移穿过全息屏,屏上的熵变有个熵变的值(公式不打了,只看看逻辑链就算了)。现在有两个公式,再加上安鲁效应,就是温度与加速度正比那个,就可以神奇地把光速,万有引力常数和普朗克常数

                • 家园 引力emerges, ants adapts, 智力,

                  briefly read through, a good paper, thx;

                  1. physics part

                  引力, everywhere, as we interact with each other, and the more we move, the more we contribute to 引力場 from our 自能場 we create as we move, and the stronger 引力場 make us move even more, a full interaction type of evil, non-linear, 无穷大 stuff, kind of challenging issues for GR's gravity quantization, how can you quantize something non-linear?

                  if someday we have a breakthrough in gravity quantization, everybody will double their life spans from 30k day to 60k day, so far, I think we die, because our brain can't handle gravity in today's social 引力場;

                  but physically, we actually live in less troubling 重力場 and our brain often mixes up social 引力場 with the physical 重力場, too simple, too nave?

                  but physics wise, ants, and pretty much all of physical agents do survive and 繁荣昌盛 in such a "junior" and pretty much earth based 引力場 and newton physics has already figured out that one already, and overall, we are still in a kind of 穩定場, 波動 from non equilibrium to equilibrium, or basically around equilibrium as the ultimate great 系综统计 partition function interact/feed back with environment, with its 么正性,因果关系, 逻辑链条 updated and surviving even a black hole, so what is all the fuss about fishing island?

                  the fuss might be about the related folks making noise and getting the "grants" more likely, and to keep social 引力場 going

                  the so-called open system or non-equilibrium system in the paper are not really "open" and "non-equilibrium" so to speak, they are defined as such only relative to their examples.

                  and with a border defined system as they quoted in a few examples, physics part is nothing really new;

                  once 邊界條件, 初始條件 are given, physicists are king, out of those 邊界條件 初始條件, god is king, and the trouble with social 引力場, those 邊界條件 初始條件 changes, all the time

                  2.智力 part

                  physical agents surviving in the above systems, collects and analyzes information, adapts to the emerging entropy force in the systems, and then they coded a software to test those agents' behaviors out, stat, then you have a paper, not "new" stuff in terms of physics, but a good one in terms of physics/AI

                  -------------

                  for a system to survive: it has to use 最小作用量, with 最大范围( to get information, often "最大體積" in 洛伦兹流形) 积分, kind of like the plant's animal behavior you guys discussed, and we could call that as plant's 最小作用量原理 or plant's AI, and from plant's AI to 費曼路徑積分, etc

                  --------------

                  Feynman's Ants - MathPages

                  mathpages.com/home/kmath320/kmath320.htm

                  According to Feynman, the second ant follows the first path, but sometimes .... The lowest-order approximation of the probability of State 5 is given by the integral ...

                  • 家园 智力:keep it simple stupid

                    or the so-called kiss model, if not, we better check the methodology part of our model;

                    otherwise, if your model is smart and robust, you should be very much close to or near @Verlinde全息屏, and theoretically 世界在你脚下, a world of simple, stupid.

                    but, as a hedge/ a cover of your ass, so you don't and never should play naked out there in the cold:

                    but how do you know that Verlinde全息屏 is not a fake one?

                    now , from 弦論 to prof 杨振宁's 規範場論

                    model wise, or if the white logic=science, then 全息屏, or whatever, you have to have a physics/mathematical 精確表示, or it will be a religion/Marxist type of utopia, a fake

                    Rusakov,Witten[2] 等人獲得此理論在一個genus G的二維緊黎曼流型M上的配分函數的精確表示

                    so, the logic chain starts with a field or system@global: where do you do your partition function 配分函數, very often in a 洛伦兹流形, a real world?

                    now, I want to come back to tg's model again, for the sake of comparison to that of white's model: tg's 檔案 system of 1.5 billion people, in a basically a flatty 欧几里德空间 of china mainland formed by 6k years of Chinese culture "pride"brain with its culture/mindset almost no 色散 at all, and to run a 配分函數 of such a system= a piece of cake?

                    thanks again to Fuhrer for this great paper.

                    ------------------

                    [PDF]

                    弦論與規範場論的對應關係

                    psroc.phys.ntu.edu.tw/bimonth/download.php?d=1&cpid=135...

                    檔案類型: PDF/Adobe Acrobat - 快速檢視

                    此文介紹Maldacena 提出的弦論與規範場論的對應關係。 一、引言. 自't Hooft 1974年提出規範場論的大N展開[1]. 後,人們普遍相信弦理論與大N 規範場論間存在 ...

                    bimonth/download.php?d=2&cpid=135&did=6 的 HTML 檔。

                    G o o g l e 在網路漫遊時會自動將檔案轉換成 HTML 網頁。

                    一、引言

                    自‘t Hooft 1974年提出規範場論的大N展開[1]後,人們普遍相信弦理論與大N規範場論間存在對偶等價關係。這個猜想在低維的理論中有較多的研究及較好的結果。特別的,對於二維論純規範場論,先是Rusakov,Witten[2] 等人獲得此理論在一個genus G的二維緊黎曼流型M上的配分函數的精確表示:

                    (1)

                    在此R是規範群的一切不可約表示,dimR 及 C2(R) 為R的維數與二次Casmir,= g2N 為規範場論的’t Hooft 藕合常數,基於此一結果,Gross和Taylor[3]於1994年證明(1)可以重寫成某些具有二維對像空間的弦理論的配分函數的1/N展開,此重要結果提供規範場理論與弦理論對應的第一個嚴格證明。可是不足的說,這並不是一個很實際的關係,眾所皆知二維規範場論是一個拓撲理論,並沒有傳播的自由度。而二維弦理論也是極為簡單的一個模型,舉例說,它沒有包含重力在內,正因兩個理論都如此簡單,所以才有可能嚴格建立一個它們之間的關係。找出四維規範場論的弦表示,一直是理論物理家的夢想,也是最初發展出弦論的一個重要動機,但四維規範場論比二維規範場論複雜多了。一般相信,QCD弦是QCD色監禁的結果,但因為QCD色監禁是一非微擾現象,至今還沒有可以完善處理它的方法。要找出四維QCD理論的弦表示是一個相當難的問題,多年無進展。

                    二、AdS/CFT 對應

                    直至1997年,年輕的Maldacena大膽提出一個關於規範場論與弦理論的假想關係[4],他猜想四維的最大超對稱性規範場論(即所謂N = 4超對稱Yang-Mills理論)是超十維的IIB型超弦理論有一對偶關係(duality relation)。這是一個頗難以置信的關係,記得1997年十二月,筆者在瑞士CERN研究中心參加一個關於弦,D-膜的研究討論會議,Maldacena就他的猜想做了報告。會中不乏在弦理論研究首屈一指的學者(如Susskind, Schwarz, Green, Harvey 等),對於Maldacena的提議皆持懷疑或觀望的態度。甚至Maldacena 自己似乎也並不太樂觀,當Harvey問即使這duality是對,但有什麼用途時,Maldacena開玩笑說道或許它跟BFSS的矩陣模型[5]一樣不太實際,引得哄堂大笑。

                    其實這也難怪,因為Maldacena最初出他的猜想時,並沒有提供一個比較完整的字典,告訴大家怎樣把兩個理論的物理現象對應聯係起來。例如,一個四維的非引力理論,怎樣可以跟一個十維的引力理論等價?十維與四維時空差別如此之大,發生在額外六維的物理現象跑到哪裡去了?

                    在此有必要對Maldacena的猜想做進一步的說明。10維的弦理論有5類型,其中之一是所謂的IIB型,這弦論具有32個時空超對稱荷。在一般的時空背景下,只有一部份超對稱性維持不被破壞。能維持所有32個超對稱性的時空背景,因為對稱性較多的關係,對應的弦論也較為漂亮。在10維時空,人們很早以前便知道只有完全平整的時空R10及AdS5×S5時空可以維持所有的超對稱性。這裡AdS5是5維的Anti-de-Sitter時空,可以想像成在6維時空中的一個超曲面:,它具有如此的度規:

                    (2)

                    AdS5有一個4維的邊界S4,是一個4維球。在Maldacena的構想中,規範場論正是生存在這個邊界時空上。因為S5沒有邊界,所以Maldacena說的是10維的AdS5×S5弦論是跟在它的邊界上的一個4維規範理論成對偶等價關係。而10維理論之所以可以和4維理論有相同的物理,是因為引力的一個奇特的性質—全息原理。

                    全息照片,相信大家在實驗課或博物館中看過,一張二維的全息底片,可以完整記錄三維物體的影像。早在70年代,人們在研究黑洞物理時便發現黑洞的熵遵守面積定理(Hawking, Berenstein),基於此現象’t Hooft[6]首先提出引力理論的全息原理,後來為Susskind[7]發展及深化。根據此原理,引力現象實際像全息現象一樣,可以用一更基本的低維理論去等價描述。這個低維的物理系,是生存在時空的邊界上的。’t Hooft-Susskind的全息原理的提出,標誌著引力探索的一個全新方向。可是儘管這個假說十分吸引,自它提出後一直沒法被證明,甚至連找到一個實際體現這個假說的物理系統也沒有。

                    Maldacena的AdS/CFT對應[CFT:因為N = 4 超對稱Yang-Mills理論]是一共型理論(Conformal field theory)],可以說是一箭雙鵰。10維IIB弦理論作為4維規範場論的一個等價表述,提供了規範場論的弦化的解答,更揭示了可以用弦論去研究色監禁這一個重要的可能性。另一方面,因為規範場論生存在10維時空的邊界上,所以可以把規範場論想像成弦論的一個全息圖,從而成為全息原理的第一個,且最重要的實際體現。

                    在根據Maldacena的AdS/CFT對應,10維時空中的量子圖效應,是跟4維場論中的’t Hooft 1/N展開成一一對應關係。根據’t Hooft的1/N展開,物理量可以依Hooft藕合及1/N做一雙重展開。它們跟弦論中的藕合常數gs及S5的半徑長度R有如下的關係:

                    (3)

                    上面提到,Maldacena提出Ads/CFT對應時,並沒有提供一個“字典”,可供計算及驗證這關係的正確性。所以有好幾個月這個假說大家都對此持比較保守的態度。情況直至98年2月,Witten [8],並Gubser,Klebanov,Polyakov [9] 分別提出一個較為完整的架構,使計算及驗證成為可能。AdS/CFT的研究遂成為弦論的一個主流,自它被提出以後,有不少的進展,有興趣的讀者,可進一步參考[10]。

                    在AdS/CFT對應關係中,時空的引力效應原則上可用非引力的規範場論去描述。一些非常有意義的問題,如時空間的奇異性,黑洞內部的時空性質,都是人們希望可以透過這個對應關係進一步了解的。對AdS/CFT關係更深入的研究,無疑可以加深我們對量子引力的了解

                    可惜,因為AdS弦理論十分複雜,我們對它的了解至今仍只停留在經典的層面上,即是說在討論AdS/CFT對應關係時,多數時間是把弦圈效應忽略,僅考慮經典的引力理論及與之對應的N = ∞ 的超對稱規範場論。這對於研究有限的N的情形(即是說包括弦的量子圈效應)無疑是一極大的限制。

                    三、BMN對應

                    上面提到,長久以來,人們一直以為在十維的時空中,只有兩個時空背景,擁有32個超引力對稱性。直到2001年,情況突然改變。這年秋季,在奧地利維也納的薛丁格研究院有一個關於弦論的研討會,為期幾個月,主持人之一Blau當時是義大利的ICTP的研究員(最近受聘到瑞士Neuchatel大學),他邀請了一批對D-膜,超引力,M-理論等有濃厚興趣的學者參加。在討論過程中,發現了IIB理論中還存在著一個新的,維持最高超對稱性的時空背景[11]。他們並且發現,這個解可以從原來的AdS5×S5時空背景(另一個維持最高超對稱性的時空背景)透過一個特別的極限過程得到。這個極限過程叫Penrose極限;是英國數學家Penrose於1976年研究廣義相對論及時空結構時所提出的一個極為漂亮的構造。Penrose問的問題是,給定任何一個黎曼時空,並考慮當中光所行的路徑,如果我們非常靠近這條路徑,我們會看到怎麼樣的規度呢?這問題跟問在一個n-維流型的某一點附近看到的什麼樣的空間類似 [答案:所看到的切空間為向量空間Rn]。Penrose發現他的問題也有一個普適的答案。他發現在這個光線附近的管狀領域上,度規恆常取同一形式—pp- 波型幾何:

                    (4)

                    而這個極限過程,即考慮某一光線附近的極限幾何,被稱為Penrose極限,需要說明,一般來說,在同一黎曼空間中可已有很多不同種類的光線,不同光線的Penrose limit會給出不同的pp- 波幾何(即不同的AIJ)。可以證明,在AdS5×S5這個黎曼黎曼流型上,只有兩類光線,取它們的Penrose極限,一種給出平直的R10空間,另一種給出一個特別的AIJ =IJ 2. 在此 =常數,為描述這pp- 波空間的曲率參數)。

                    因為對AdS5×S5存在一個對偶的規範場論表述,很自然的會問,pp- 波的弦理論也有對應的對偶表述嗎?這問題於去年二月為Maldacena及合作者Berensteim及Nastase[12]提出並解答。他們發現,如果把pp- 波幾何想像成AdS5×S5的一個極限部分,那麼它的對偶場論也應該是原來的N = 4超對稱YM規範場論中的某個特殊部分 (sector)。

                    他們發現,必須考慮原N = 4 SYM 中的一些特殊的算符(帶有J單位的超對稱U(1)荷),並考慮一個特殊的雙重線度極限:

                    N→∞, J→∞

                    J2/N = 固定, (5)

                    = 固定.

                    這有利於原來的’t Hooft大N極限:

                    N →∞,

                    J2/N →0, (6)

                    N = 固定.

                    這些特殊的算符,被稱為BMN算符。在這極限,它們的anomalous dimension 有限,並與弦論中的各種粒子的質量成對應。BMN並發現在Penrose極限下,AdS5×S5的弦理論得以化簡,極限的pp- 波的弦理論更是完全可積系統,從而它的所有弦態及質量皆可完整獲得。這是人們在規範弦場論對應關係中可以研究弦效應的首次勝利,是原來的AdS/CFT對應關係無法做到的。

                    BMN對應關係比AdS/CFT對應來的實際,因為不管在弦,還是場論方面,人們都可以做微擾計算,從而為研究及驗證這關係打下一個基礎。BMN對應關係亦成為弦論的另一主流方向。在BMN原本的文章中,它們只提出自由的弦論與規範場論的對偶關係。不久,人們意識到引入相互作用的可能性,並提出了弦的三頂角作用與場論中的相關函數的關係 [13]。這猜想獲得一定的支持,但也有一些問題,特別是當推廣到含有費米子的情形,仍待解決。

                    四、結語

                    長久以來,人們期待可以把引力及其他作用力統一描述。Klanza-Klein理論漂亮地把規範場與引力統一到高維的引力理論中。但因愛因斯坦的引力理論存在量子化的問題,所以不能成功。弦論把各種相互作用統一,只需要單一的弦作用,可以包括所有的作用力。不單如此,因為弦論不存在紫外發散問題,所以弦論是一完備的量子引力理論。弦規範場論對應關係的成立,似乎表示或許量子引力及弦論並不是最基本的,或許引力現象只是更基本的規範場的動力學效應的一種展現。其實很久以前,人們己經有這樣的猜想[14]。徐教授最近也再次提出了這個想法[15]。

                    透過弦規範場論對應關係的研究,預期將可加深我們對引力及時空結構的基本了解。或許一場橫時代的物理革命正在醞釀。

                    參考資料:

                    [1] G. 't Hooft, Nucl .Phys .B 72, 461 (1974).

                    [2] B.E. Rusakov, Mod. Phys. Lett.A 5, 693 (1990);

                    E.Witten, Commun. Math. Phys.141, 153 (1991).

                    [3] D. Gross and W. Taylor, Nucl. Phys. B 400, 181 (1993); Nucl. Phys. B 403, 395 (1993).

                    [4] J.M. Maldacena, Int. J. Theor. Phys. 38, 1113 (1999).

                    [5] T.Banks, W. Fischler, S.H. Shenker, and L. Susskind, Phys. Rev. D 55, 5112 (1997).

                    [6] G. 't Hooft ,gr-qc/9310026.

                    [7] L. Susskind, J. Math. Phys. 36, 6377 (1995).

                    [8] E. Witten, Adv. Theor. Math. Phys. 2, 505 (1998).

                    [9] S.S. Gubser, I.R. Klebanov, and A.M. Polyakov,

                    Phys. Lett. B 428, 105-114 (1998).

                    [10] O. Aharony, S.S. Gubser, J.M. Maldacena, H. Ooguri, and Y. Oz, Phys Rept. 323, 183 (2000).

                    [11] M. Blau, J. Figueroa-O'Farrill, C. Hull, and

                    G. Papadopoulos, JHEP 0201, 047 (2002).

                    [12] D. Berenstein, J.M. Maldacena, and H. Nastase,

                    JHEP 0204, 013 (2002).

                    [13] N.R. Constable, D.Z. Freedman, M. Headrick,

                    S. Minwalla, L. Motl, A. Postnikov, and W. Skiba, JHEP 0207, 017 (2002); C.S. Chu, V.V. Khoze and G. Travaglini, JHEP 0206, 011 (2002).

                    [14] H. Kawai, D.C. Lewellen, and S.H.H. Tye,

                    Nucl. Phys .B 269, 1 (1986).

                    [15] A. Zee, hep-th/0309032; Phys. Rev. Lett.

                    55, 2379 (1985).

                    作者簡介

                    朱創新於1991年從香港中文大學物理系畢業,於1996年從美國加州大學柏克萊分校獲得理論物理博士,接著於義大利高等理論研究所(SISSA),瑞士Nechatel大學從事博士後研究。於2000年底到英國Durham大學數學系任教。於2002年秋到台灣國立清華大學物理系任教。專長為量子場論弦理論及數學物理,研究主題包括規範場論,弦理論,非對易幾何,黑洞與量子引力及弦理論。

                    • 家园 Some Critique

                      The authors have missed out some of the developments in the recent decade. There's, for example, Schmidhuber's compressibility gradient, Ay and Der's homeokinesis predictive information maximization, Klyubin et al.'s empowerment method, and the underlying Touchette/Lloyd formalism for describing entropic bounds on open and closed-loop control and many more. Plus, the authors do not seem to mention in their paper Linsker's criticism of Dewar's MEPP derivation. In isolation, it is difficult to put in perspective how the approach sits in the context of the existing work and how much it depends on postulated MEPP-type assumptions. Which are perfectly fine to make, but it'd be nice to see how it fits into other closely related models which are already studied in some detail.

                      • 家园 goog may buy them out

                        1.

                        AI is all about how we individuals as social animals or social agent socially interact with each other more effectively and efficiently, and information collection and analysis is a big part of it;

                        so, as I has been posting all the time, we have a social 场, but with a huge waste of human energy, and social 场 has to be "reformed" not by revolution, but with help of physics/math;

                        and as I have been writing about the white social science "model", white relies on the development of their social human capital across as broad social ranks/areas as possible with a kind of "free and diverse choice/market" style, all the time, it is in their blood, that kind of culture;

                        in contrast, we Chinese are almost like one kind of product, out of Chinese 科技人才 education assemble line, almost clean cookie cutter

                        and we do have quite bit of 技, few of first class level, very little 科 if any at all, the way those Chinese professors are.

                        西南联大 used to be near if not @world advanced level, physics, at least.

                        2.

                        爱因斯坦场=gravity, 幾何的, what you can do with 幾何?

                        热力学: 熵, temperature are all defined and measured by partition function and stat @macroscopic, majority of people are not sysadmins;

                        Verlinde, after his string theory (can never be tested/measured at current energy capability level) went no where, combined the part 1&2, so he basically has a information field theory model established;

                        and with that the paper you quoted and other folks in the field, can start do more app oriented models/softwares, etc, a great start;

                        As discussed about that MIT AI meeting, those AI folks will have to pay attention to the physics side of AI, which none of them know any thing, AI without physics is going no where, goog should buy these two guys outright;

                        3. now I am going to bad mouse on our Chinese culture (including my own mindset the way it is formatted like everybody eles):

                        we have been all brain 光刻 by our dear Chinese culture for 6k years, and possibly as a result, we under perform in many ways if compared to Indians and some European folks here in US with similar background as we do, and I happen to know too many Chinese engineering/physics/math PHDs here in usa: 身在曹營心在漢, with all kind of their own little dreams/tricks for 漢, kind of an open secret in and outside of that community.

                        I often comment to them: you guys watch too much 后宫·甄嬛传, but that is in our blood.

                        and tg as smart as they are know that very well, tg is not going to pay a good price unless you are really of some value, and we all know tg 二代 don't really bother to spend even a few pennies to those famous 两弹一星 元勋二代, no honey, no money;

                        my bad mouse's point is actually more about our selves as an individual: how "too simple too nave" our own brain AI could go in a social 场? and our individual's point of view in a social 场 is very often garbage, in terms of competitiveness, if from physics point of view ;

                        even in physics, 爱因斯坦场, 热力学 can only provide information at macroscopic level about 场力 (the way Verlinde did, still disputed, of course and as always, no 實驗, no money in physics) , you can oly define and measure 爱因斯坦场力, gravity, or 热力, @macroscopic level around a "canonical (平衡正則) system", and how do you define a "canonical (平衡正則) system"? it is already very difficult in physics;

                        so for most of us, instead of reading with your raw eyes and think with your raw brain, you got to have a model, or if you can manage to have some access to tg's internal elite 场, which is very much 等價交換 based like any 场 , and there is no way you can cheat out the big boss;

                        so, that paper in that sense is some kind of breakthrough, in terms of social AI model building, and hopefully from there on, a social 场 will be more competitive based on more "fair" playing information acquisition/process, more like financial market: can tg/or any organization manage/manipulate global international financial market? tg's 金融38軍 has been 內戰 all the time, and then 后宫·甄嬛传

                        • 家园 Well said on the first point

                          Yet be very careful with your "bad mouthing"...haha,

                          Forbes有篇新闻里在描述这个发现是提到了大致与你相同的意思:

                          “We cultivate open minds not because we are liberal or conservative, young or old, but because we understand intuitively that it is a matter of survival—physically, emotionally and intellectually—to maximize access to future possibilities.”

                          可惜的是,这样的intuition在有些人眼里看去只是代表了普世脂油,却不知道这样的文化是内在原因在哪里。

                          不过如果把政治文化放在一边,中国人的学习能力与聪明程度上不差的,相信百度在硅谷开店应该可以第一时间接触最新进展,万一出现了重大突破不至于只能在远处打打酱油。当然这只是科技上,也就是以前被当作“奇技淫巧”的事,而在社会文化上如果能有所领悟,甚至发生变化,还不是一时半会的事,因此中国在文化到科学这条线上的leadership大概不是我们有生之年可以看到的了。

分页树展主题 · 全看首页 上页
/ 31
下页 末页


有趣有益,互惠互利;开阔视野,博采众长。
虚拟的网络,真实的人。天南地北客,相逢皆朋友

Copyright © cchere 西西河