v*c 发帖数: 42 | 1 【 以下文字转载自 Statistics 讨论区 】
发信人: vic (红尘中的精灵), 信区: Statistics
标 题: 问一个asymptotic distribution的问题
发信站: BBS 未名空间站 (Wed Aug 22 23:12:07 2007)
已知cdf(cumulative distribution funciton)得asymptotic distribution. 能够知道
pdf(probability density function)的asymptotic distribution 是什么么?用delta
method好像这里不行。
That is, we know
sqrt(n)[\hat_{F}(Y)-F(Y)]->N(0,sigma)
what is
sqrt(n)[\hat_{f}(Y)-f(Y)]->?
F is cdf of random variable Y. f is pdf of random variable Y. \hat_{F}(Y) is
the empirical estimator of F(Y). And \ |
|
b****s 发帖数: 1300 | 2 以前画矢量图都是用corel draw,因为不需要画严格的数学形状。现在有了这种需要,
所以请教下tikz 和 asymptote的达人,如果考虑到数学形状,又兼顾一般矢量2D或3D
图形,应该选择哪个作为一劳永逸的工具?毕竟精通其中一个都要花不少时间。
tikz对我来说最大的优点就是和latex集成很好,直接在latex里用很方便,不像
asymptote那样,还得专门用asy的命令搞一下(当然可以用latexmk进行自动化,但毕
竟增加了编译的时间)。 |
|
b****s 发帖数: 1300 | 3 以前画矢量图都是用corel draw,因为不需要画严格的数学形状。现在有了这种需要,
所以请教下tikz 和 asymptote的达人,如果考虑到数学形状,又兼顾一般矢量2D或3D
图形,应该选择哪个作为一劳永逸的工具?毕竟精通其中一个都要花不少时间。
tikz对我来说最大的优点就是和latex集成很好,直接在latex里用很方便,不像
asymptote那样,还得专门用asy的命令搞一下(当然可以用latexmk进行自动化,但毕
竟增加了编译的时间)。 |
|
b****s 发帖数: 1300 | 4 tikz画数学曲线还得再调研别的package,3D曲面估计不好弄,但一般的三维效果可以
通过渐变、阴影,和透视来想表现,估计也不会有啥区别。
asymptote嵌入Latex后编译太花时间了,本来用latexmk实现的pdf实时自动刷新也被弄
得毫无意义了。 |
|
a**e 发帖数: 5794 | 5 我喜欢离线搞好图形再插入TeX文档,这样不需要每次都编译图形。
所以pgf/tikz画图时用preview宏包生成独立文件。
asymptote的语法可能好一点,虽然我没用过。
3D |
|
r****t 发帖数: 10904 | 6 搞阴影,3D 啥的反正 tikz 是不中的,asymptote 应该行。不过这样的阴影不知道怎
么搞。 |
|
|
b****s 发帖数: 1300 | 8 tikz画数学曲线还得再调研别的package,3D曲面估计不好弄,但一般的三维效果可以
通过渐变、阴影,和透视来想表现,估计也不会有啥区别。
asymptote嵌入Latex后编译太花时间了,本来用latexmk实现的pdf实时自动刷新也被弄
得毫无意义了。 |
|
a**e 发帖数: 5794 | 9 我喜欢离线搞好图形再插入TeX文档,这样不需要每次都编译图形。
所以pgf/tikz画图时用preview宏包生成独立文件。
asymptote的语法可能好一点,虽然我没用过。
3D |
|
r****t 发帖数: 10904 | 10 搞阴影,3D 啥的反正 tikz 是不中的,asymptote 应该行。不过这样的阴影不知道怎
么搞。 |
|
|
|
z*t 发帖数: 863 | 13 Beutler得奖,Medzhitov会不会有意见?
http://www.sciencedirect.com/science/article/pii/S1074761309002
Ruslan Medzhitov
This year marks the 20th anniversary of a publication (Janeway, 1989) that
in many ways revolutionized our understanding of the immune system. As part
of the proceedings of the Cold Spring Harbor Symposium on Immune Recognition
, the paper authored by the late Charles A. Janeway, Jr. was not a standard
peer-reviewed publication. In fact, Charlie used to refer to it as “the
best paper I've ... 阅读全帖 |
|
L*******t 发帖数: 2385 | 14 现在literature里面asymptotic approxiamtion的方法此起彼伏,一浪还比一浪高,还
有些专业写手专门干这个的。搞的我都心动了。想问问,业界用这个的多么? |
|
L*******t 发帖数: 2385 | 15 咋没人care这个呢,有了asymptotic expansion,很多没有closed form的东西都有了
semi closed form,calibration,MLE会变得很容易,我现在开发了一套expansion方
法,适用范围很广,而且能证明收敛,准备先解个FBSDE试试看,然后再搞Density
expansion。。 |
|
L*******t 发帖数: 2385 | 16 而且asymptotic expansion的,不管几个参数,上来expand,都适用的啊。 |
|
n******p 发帖数: 283 | 17 Applied asymptotics: case studies in small-sample statistics By Alessandra R
. Brazzale, Anthony Christopher Davison, N. Reid
Thank you so much!!!! |
|
j*********1 发帖数: 15 | 18 Need help on this problem
Test
null: theta=theta.a
alternative theta=theta.b (b
theta is the parameter of exponential distribution
have X1,... Xk+1 (each follow the exponential distribution) at different
time interval t1,...tk (equal) and tk+1=infinite. so X1,... Xk+1 follow a
multinomial distribution with pi determined by pdf theta from ti-1 to ti.
Now consider use likelihood ratio test (used MLE before and want to compare)
Now we have lnLR=sum of [xi*ln(p(theta.a/theta.b)]
how to approxima... 阅读全帖 |
|
g****g 发帖数: 1828 | 19 In probability theory, the normal (or Gaussian) distribution, is a
continuous probability distribution that is often used as a first
approximation to describe real-valued random variables that tend to cluster
around a single mean value. The graph of the associated probability density
function is “bell”-shaped, and is known as the Gaussian function or bell
curve:[nb 1]
f(x) = \tfrac{1}{\sqrt{2\pi\sigma^2}}\; e^{ -\frac{(x-\mu)^2}{2\sigma^2}
},
where parameter μ is the mean (location of the pe... 阅读全帖 |
|
f**d 发帖数: 768 | 20 这是一本计算神经科学的优秀著作,全文拷贝这里(图和公式缺),有兴趣的同学可以
阅读
如需要,我可以分享PDF文件(--仅供个人学习,无商业用途)
From Computer to Brain
William W. Lytton
From Computer to Brain
Foundations of Computational Neuroscience
Springer
William W. Lytton, M.D.
Associate Professor, State University of New York, Downstato, Brooklyn, NY
Visiting Associate Professor, University of Wisconsin, Madison
Visiting Associate Professor, Polytechnic University, Brooklyn, NY
Staff Neurologist., Kings County Hospital, Brooklyn, NY
In From Computer to Brain: ... 阅读全帖 |
|
d******s 发帖数: 180 | 21 http://aimath.org/news/primegaps70m/
第一段就很有料,四月后期投稿,五月中旬便接受,对于一篇55页的paper来说速度惊
人。
Zhang's Theorem on Bounded Gaps Between Primes
by Dan Goldston
In late April 2013 Yitang Zhang of the University of New Hampshire submitted
a paper to the Annals of Mathematics proving that there are infinitely many
pairs of primes that differ by less than 70 million. The proof of this
amazing result was verified with high confidence by several experts in the
field and accepted for publication. A public slightly re... 阅读全帖 |
|
d**e 发帖数: 2420 | 22 【 以下文字转载自 Mathematics 讨论区 】
发信人: dade (leno), 信区: Mathematics
标 题: 见识一下数学文章的一稿多投
发信站: BBS 未名空间站 (Tue Mar 15 15:03:30 2011, 美东)
文章仍然可以下载,打上了RETRACTED标识。第一次见着数学论文一稿多投,三波人,
居然在那么短时间内两次投同一个杂志Journal of Computational and Applied
Mathematics,看来缺少沟通呀,太搞了。
http://xys.org/xys/ebooks/others/science/dajia12/wanglijuan.txt
一鱼三吃、五人共享
作者:卿本佳人
一篇文发到三份期刊刊登,都被撤文了。涉及抄袭的五人是:
北京首都师范大学数学系 及 浙江嘉兴 嘉兴学院 数理与信息工程学院 王立娟
湖南省常德市 湖南文理学院 数学与计算科学学院 彭乐群
浙江嘉兴 嘉兴学院 数理与信息工程学院 王文涛
浙江嘉兴 嘉兴学院 数理与信息工程学院 乐光学
广东广州 广东工业大学 应用数学学院 欧春霞
... 阅读全帖 |
|
d**e 发帖数: 2420 | 23 文章仍然可以下载,打上了RETRACTED标识。第一次见着数学论文一稿多投,三波人,
居然在那么短时间内两次投同一个杂志Journal of Computational and Applied
Mathematics,看来缺少沟通呀,太搞了。
http://xys.org/xys/ebooks/others/science/dajia12/wanglijuan.txt
一鱼三吃、五人共享
作者:卿本佳人
一篇文发到三份期刊刊登,都被撤文了。涉及抄袭的五人是:
北京首都师范大学数学系 及 浙江嘉兴 嘉兴学院 数理与信息工程学院 王立娟
湖南省常德市 湖南文理学院 数学与计算科学学院 彭乐群
浙江嘉兴 嘉兴学院 数理与信息工程学院 王文涛
浙江嘉兴 嘉兴学院 数理与信息工程学院 乐光学
广东广州 广东工业大学 应用数学学院 欧春霞
Asymptotic behavior of solutions to a system of differential equations with
state-dependent delays
Lijuan Wang, Department of Ma... 阅读全帖 |
|
p********a 发帖数: 5352 | 24 ☆─────────────────────────────────────☆
TNEGIETNI (lovewisdom) 于 (Tue Oct 11 20:01:18 2011, 美东) 提到:
朋友告诉我说他要来,还有Harvard的Carl Morris等,$200的Registration fee对我来
说不是一笔小数目。我其实并不欣赏他的bootstrap法,曾公开批评过这个方法的逻辑
错误。如果去,我跟他讲什么呢?版上的恶朋好友可否给点建议?谢谢。
☆─────────────────────────────────────☆
angelsun (安吉笋) 于 (Tue Oct 11 20:38:46 2011, 美东) 提到:
在哪里?什么时候?讲座是什么topic的?
☆─────────────────────────────────────☆
statcompute (statcompute) 于 (Tue Oct 11 23:51:23 2011, 美东) 提到:
$200 for a lecture by Efron? it i... 阅读全帖 |
|
a***r 发帖数: 594 | 25 美国的划分也是这样的。比如:
MIT 物理系的“核物理实验室”涵盖该系的理论物理中心:
http://web.mit.edu/lns/research/index.html
research interests: anti-matter, neutrino physics, dark matter, structure of
proton, structure of high density matter and early universe.
Research on neutrino oscillation (standard and slightly beyong standard
model physics), protron structure (asymptotic and none asymptotic QCD) and
high density matter (heavy ion collider) are loosely considered topics of
modern nuclear physics.
MIT 的核工程系。
http://web.mit.... 阅读全帖 |
|
x****6 发帖数: 4339 | 26 下面是1998年到现在的屋里炸药奖,除中微子外,2008, 2004,1999这几年难道不是
你说的基本粒子或相互作用吗?
2008
Yoichiro Nambu
"for the discovery of the mechanism of spontaneous broken symmetry in
subatomic physics"
Makoto Kobayashi and Toshihide Maskawa
"for the discovery of the origin of the broken symmetry which predicts the
existence of at least three families of quarks in nature"
2004
David J. Gross, H. David Politzer and Frank Wilczek
"for the discovery of asymptotic freedom in the theory of the strong
interaction"
1999
Gerardus 't H... 阅读全帖 |
|
d******e 发帖数: 7844 | 27 Decon的结论其实是有一定道理的。只不过他未必明白更深层的含义,至少我下面说的
本科的统计大都不会提及,美国的硕士统计课程也比较少提及。读PhD的话,修过
Advnaced的Inference课程的话,应该明白我在说什么。
因为是五年前修的课了,所以个别地方可能说的不准确,但是思路应该基本正确。
当一个Bernoulli r.v.的参数(分布均值)非常接近0的时候,想要同时准确估计样本的
参数和样本方差是很难的。现在可以看到p肯定是非常小的,很可能是在1e-5~1e-6的数
量级上。这时想用MLE的Asymptotic normality来做近似就很成问题了,因为对分布的
方差的估计的方差对比分布的方差本身还要大,导致你对方差的估计十分不可靠,最后
testing的结果非常不准,CI也十分不可靠。
解决办法我说两种,感兴趣的可以去看。第一种是去做exact testing,不做近似,但
鉴于样本量大,计算量会非常大。第二种是在推导Asymptotic ditribution时做高阶泰
勒展开,这样可以获得更精确的近似。
当然,也可以Heuristic的去直接拿mean比或者用norm... 阅读全帖 |
|
d******e 发帖数: 7844 | 28 Decon的结论其实是有一定道理的。只不过他未必明白更深层的含义,至少我下面说的
本科的统计大都不会提及,美国的硕士统计课程也比较少提及。读PhD的话,修过
Advnaced的Inference课程的话,应该明白我在说什么。
因为是五年前修的课了,所以个别地方可能说的不准确,但是思路应该基本正确。
当一个Bernoulli r.v.的参数(分布均值)非常接近0的时候,想要同时准确估计样本的
参数和样本方差是很难的。现在可以看到p肯定是非常小的,很可能是在1e-5~1e-6的数
量级上。这时想用MLE的Asymptotic normality来做近似就很成问题了,因为对分布的
方差的估计的方差对比分布的方差本身还要大,导致你对方差的估计十分不可靠,最后
testing的结果非常不准,CI也十分不可靠。
解决办法我说两种,感兴趣的可以去看。第一种是去做exact testing,不做近似,但
鉴于样本量大,计算量会非常大。第二种是在推导Asymptotic ditribution时做高阶泰
勒展开,这样可以获得更精确的近似。
当然,也可以Heuristic的去直接拿mean比或者用norm... 阅读全帖 |
|
l****p 发帖数: 397 | 29 他的意思是我做的optimizing只是在同一个asymptotic order里的小优化,并没有改变
asymptotic order |
|
g********t 发帖数: 39 | 30 贡献刚做的online test,职位是Machine Learning related。
Question 1 / 2 (LaserMaze)
You are standing in a rectangular room and are about to fire a laser toward
the east wall. Inside the room a certain number of prisms have been placed.
They will alter the direction of the laser beam if it hits them. There
are north-facing, east-facing, west-facing, and south-facing prisms. If the
laser beam strikes an east-facing prism, its course will be altered to be
East, regardless of what direction it had been goi... 阅读全帖 |
|
M****g 发帖数: 162 | 31 Question 1 / 2 (LaserMaze)
You are standing in a rectangular room and are about to fire a laser toward
the east wall. Inside the room a certain number of prisms have been placed.
They will alter the direction of the laser beam if it hits them. There
are north-facing, east-facing, west-facing, and south-facing prisms. If the
laser beam strikes an east-facing prism, its course will be altered to be
East, regardless of what direction it had been going in before. If it hits
a south-facing prism... 阅读全帖 |
|
j*******t 发帖数: 223 | 32 也许是我没说清楚,原话是
The results of the previous section imply that even through the
discriminative logistic regression has a lower asymptotic error, the
generative naive bayes classifier may also converge more quickly to its
higher asymptotic error. |
|
m********7 发帖数: 37 | 33 我经常画流程图,如果用Text mode packages,like metapost vs. asymptote,比起
visio 方便吗?
比如移动一个方框,相关连线跟着移动,visio很方便,我觉得像metapost vs.
asymptote恐怕不方便。
希望用过Text mode 和GUI mode 给个比较。 |
|
l***o 发帖数: 390 | 34 用的asymptote。我也不知道尺寸,颜色用的是asymptote预定义的,本来可以用个画图
软件把rgb的值取出来的,不过既然是个大概,就没有深究。
感觉你的比我的轻松,我调试了好一会才得现在这个效果。 |
|
|
r*****z 发帖数: 906 | 36 pstricks/pgf这些和metapost/asymptote是完全不同的两类应用
asymptote已经足够好用了 |
|
k**********g 发帖数: 989 | 37
f =
@(x, c, s) (0.5 * x - 0.5 * sqrt ((x - c) .^ 2 + s) + 0.5 * c)
Input: "x", scalar or vector. Operation will be applied element-wise.
Configuration parameters:
(Param #1)
Corner position - "c" - scalar, positive.
For inputs "x" smaller than "c", asymptotic behavior is y = x.
For inputs "x" larger than "c", asymptotic behavior is y = c.
(Param #2)
Corner smoothness - "s" - scalar, positive.
For smaller values of "s", the curve (x, y) will look like a sharp angle
near the point (c, c).
For lar... 阅读全帖 |
|
e***e 发帖数: 5 | 38 谁能解释解释这两个渐进是什么意思?中文是不是叫做m-估计和n-估计?书上看到的,一点
儿概念都没有.google搜搜也没有找到介绍. |
|
z****t 发帖数: 58 | 39 TikZ,复杂程度和 MetaPost、Asymptote 差不多。要画得精致就是要复杂一些。
PGF/TikZ 比较方便,效果好, 有 node,MetaPost 有 box,都是自动确定连线位置的;
图的数量小,用 TikZ 比较方便;如果图的数量大,通过适当编程,最终可以用
Asymptote 比较容易。
比较专门的工具,包括 Graphviz |
|
n*******l 发帖数: 2911 | 40 我觉得,当你用微扰法的时候,已经假定了这个解以及它的导数们只有O(1)项以及
一些O(eps^n)项。所以,当A=1/eps的时候,尽管通解是f=c1 e^{-A x}+c2,
但是e^{-A x}在x=0附近导致f的导数非常大,不再是O(1)+ \sum_{i>0} O(eps^i),
所以微扰法的基本前提就已经排除了e^{-A x}形态的解。然后毫无疑问的你只能得到
一个f=c2形态的解。 这跟常识一致,因为e^{-A x}在x增大时很快就递减到零。
当然,我认为对微分方程来说,首先应该有一个边界条件,没有边界条件的限制,意义
有限。
If you want to know the behavior of f around x=0, you can rescale
x, say letting x=eps*y, then g(y):=f(x)=f(eps y) and the
original equation becomes g''(y) = -g'(y), involving no eps at all,
and it can be solved.
If you pretend... 阅读全帖 |
|
L*m 发帖数: 235 | 41 近十年JAMS上的华人作者,JAMS上的文章相对更多,搜的没那么仔细,有可能不完全。
大家也看看有没有漏的
2015
Kähler-Einstein metrics on Fano manifolds. I: Approximation of metrics
with cone singularities
Xiuxiong Chen, Simon Donaldson and Song Sun.
Kähler-Einstein metrics on Fano manifolds. II: Limits with cone angle
less than 2π
Xiuxiong Chen, Simon Donaldson and Song Sun.
Kähler-Einstein metrics on Fano manifolds. III: Limits as cone angle
approaches 2π and completion of the main proof
Xiuxiong Chen, Simon Donaldson... 阅读全帖 |
|
|
c****e 发帖数: 2097 | 43 the stupid somebody happens to be einstein, i believe.
well, if the universe is closed (in some way of foliating by spacelike
hypersurfaces), then it simply does not have a boundary.
if it's asymptotically flat or something else, there might be a asymptotic
or conformal boundary, but again, there isn't a outside.
the blackhole has a horizon. in that sense there's a outside. but there's
no singularity at the horizon. coordinates do not invent new spaces or
boundaries. your understanding is not se |
|
L*m 发帖数: 235 | 44 2015诺贝尔物理奖
获奖人:Arthur B McDonald、Takaaki Kajita
获奖领域:实验高能物理
获奖原因:发现了中微子震荡,并因此证明了中微子具有质量
相关文章
Measurement of the Rate of νe+d→p+p+e− Interactions Produced by B8
Solar Neutrinos at the Sudbury Neutrino Observatory
Q. R. Ahmad et al. (SNO Collaboration)
Phys. Rev. Lett. 87, 071301 – Published 25 July 2001
Direct Evidence for Neutrino Flavor Transformation from Neutral-Current
Interactions in the Sudbury Neutrino Observatory
Q. R. Ahmad et al. (SNO Collaboration)
Phys. Rev. Lett. 89, 011301 – ... 阅读全帖 |
|
i**w 发帖数: 71 | 45 背景:fresh physics PhD
还没offer,但年前基本上不会再折腾了。
有重复。基本上都是很标准的题。
简单的题如果人家想问倒你也是很容易的。
面试书recruiter推荐
1) Mark Joshi: "Quant Job Interviews: Questions and Answers". I have
heard very good things about this book.
2) Xinfeng Zhou, "A Practical Guide to Quantitative Finance Interviews"
个人觉得非常有用, 大部分问题都在这两本上。
算法,C++, stochastic calculus 就看比较标准的几本。
- sqrt(i)=?
- You and me roll a dice,first one gets a six wins. You roll first. what
is the probability of you winning?
- A stair of n steps. Each time you st... 阅读全帖 |
|
w**********y 发帖数: 1691 | 46 多谢分享.大概做了做..欢迎补充和指正.
- sqrt(i)=?
e^{\pi/4 i} or - e^{\pi/4 i}
- You and me roll a dice,first one gets a six wins. You roll first. what
is the probability of you winning?
P(I win) = P(Y !win and I win) = 6/11
- A stair of n steps. Each time you step up 1 or 2 steps. How many
different ways are there to reach the top? what is the asymptotic limit?
Fibonacci sequence ..limF(n)/F(n-1)==x for n>2, solve x, and F(n) ~ x^{n-1}
- Moment generating function of standard model.
statistic book…
- Write a si... 阅读全帖 |
|
t*******y 发帖数: 637 | 47 第二题应该是6/11吧
能讲讲这个吗? - X1 and X2 are independent random variable with pdf f and g.
what is what is the pdf of X=X1+X2
Jacobian matrix for X1+X2 and X1-X2..
多谢分享.大概做了做..欢迎补充和指正.
- sqrt(i)=?
e^{\pi/4 i} or - e^{\pi/4 i}
- You and me roll a dice,first one gets a six wins. You roll first. what
is the probability of you winning?
P(I win) = P(Y !win and I win) = 5/6*1/6
- A stair of n steps. Each time you step up 1 or 2 steps. How many
different ways are there to reach the top? what is the asymptotic... 阅读全帖 |
|
p*****k 发帖数: 318 | 48 since there are total of 2^N-1 sums for an N-element set, so 2^(i-1)
optimizes the total sum.
OP's problem (or rather the discussion so far) is related to Erdos's
conjecture:
find a set with such property (i.e., distinctive subset sums) with the
smallest largest element.
(his conjecture is the asymptotic form with N->infty)
the solution was the so-called Conway-Guy sequence, which gives the upper
bound, and they verified up to N=8 that their solution is optimal.
(but the asymptotic form had been... 阅读全帖 |
|
r**a 发帖数: 536 | 49 Hi dude. Calm down. I am just arguing the academic problem with you, nothing
else.
I just dig deeper about your way. I am convinced. Sorry for the previous
confusion.
For the solution, I do have my own method and already mentioned. The random
walk can be mapped to a BM with drift. The theory underlying is similar to
the numerical tree method for the Black-Scholes model. But my way is much
harder than yours. The details are as follows:
Regarding S(n), we have S(n) = S(n-1) + 0.5 + 1.5*Z_{n-1}, wh... 阅读全帖 |
|
r**a 发帖数: 536 | 50
JF
return
这个当然算了,比如SABR model。实际上SABR里面的技术非常universal。只要你不怕
麻烦,你可以用它来算更一般的SV模型。问题在于,并不是什么SV模型都能有closed
form asymptotic expansion的。在SABR里,核心是那个heat kernel on poincare
plane。正是因为这个玩意有closed form,这样才有Hagan的asymptotic expansion。
你的SV模型如果对应的Riemannian Geometry不是poincare geometry。那么you are
probability in trouble.
另外一点是要看你模型如何用。比如如果你要price一个20年的option,你要问一下你
自己你的formula这时候能不能用。效果怎么样。这些都要做test的。
实现起来也 |
|