a********e
发帖数: 508 | 1 a.s. => L2 converge => converge in probability
not the other way around |
|
M****i
发帖数: 58 | 2 Recall that a necessary condition for the series \Sigma_n a_n being
convergent is that a_n converges to 0. |
|
a********e
发帖数: 508 | 3 thank you. It's a smart way to think about the example,
but it might need some improvement
note that such sequence must converge to 0 for the sum series to converge |
|
y****e
发帖数: 28 | 4 Here is an counter example.
Define a_{3k-2}=a_{3k-1}=a_{3k}=(3k)^{-1/3} for all k=1,2,...
Define b={-2,1,1,-2,1,1,-2,1,1,...}
let c_n = a_n * b_n
By using Dirichlet's test http://en.wikipedia.org/wiki/Dirichlet_test,
we have c_n converges, since the series \sum_{n=1}^N b_n is bounded.
It's easy to see that |c_n| and c_n^3 do not converge |
|
z****g
发帖数: 1978 | 5 Try the second result, locally convergence by measure. It is a complete
and sufficient criteria. Or, the question itself is wrong since it is very
rare to prove a.s. convergence for stochastic process.
of |
|
c**********e
发帖数: 2007 | 6 Your idea is good. But your proof is for convergence in probability
while the original problem is to prove convergence a.s..
b) |
|
c**********e
发帖数: 2007 | 7 You are right. As Tn-T1 is the sum of some iid positive difference,
convergence in probability implies convergence a.s..
|
|
c**********e
发帖数: 2007 | 8 Your idea is good. But your proof is for convergence in probability
while the original problem is to prove convergence a.s..
b) |
|
c**********e
发帖数: 2007 | 9 You are right. As Tn-T1 is the sum of some iid positive difference,
convergence in probability implies convergence a.s..
|
|
r**a
发帖数: 536 | 10 Of course, you need to first check the convergence of the numerical scheme
you want to use. But besides the convergence, stability also matters
sometime.
to
What
loss
PDE |
|
G*******s
发帖数: 10605 | 11 测度论里面这一块我并不会问很具体的证明,甚至连具体的定理本身不会完全复述都没
关系(例如Fatou's Lemma要求必须是Non-negative函数，没有这个定理结果不成立)
如果这位同学说自己学过实分析（一般是5年之内学的），我们会问在实分析里面有一
系列定理（Monotone Convergence Theorem, Dominated Convergence Theorem, Fatou
's Lemma),这一系列定理都围绕着一个主题，请问这个主题是什么？能正确回答出这个
问题的candidate通常有比较强的概括思维能力，也就是说学习东西的时候善于从更高
层面看问题，这个是我们很看重的一个素质，这样的素质在把一个领域的技巧应用到另
外一个领域时很重要。
可惜的是很多数学统计Ph.D这个问题都摸不着头脑，有一个Princeton University的本
科毕业生反而可以很轻松的回答，牛校就是有人才。 |
|
s**e
发帖数: 103 | 12
I think it is sufficient if anything is absolute convergent, it is convergent,
but not vice versa. This is my memory. :-) |
|
|
c******d
发帖数: 306 | 14 I am running a "for loop in R. In this loop, there is convergence function.
I need plug in around 200 datasets to see if they are converged or not.
The problem is when the loop goes to some unconverged datasets , the loop
stopped and reported error. Is there any method that I can keep the loop
running even when error was reported?
Thx |
|
t**i
发帖数: 688 | 15 I am not familiar with simplex(), but this error seems to stem from
iteration/convergence criteria. Either obfun> -eps or it<=n.iter gives a NA
and that caused the problem. Try to play with eps or n.iter options.
However, caution is needed because the thus-generated results may not have
been drawn from the convergence status. |
|
m*****e
发帖数: 13 | 16 Deal All,
I want to ask a stupid question about the null model fit statistics in proc
logistic. I want to get the -2LogL of the null model. I thought there are
two ways to do that.
1) Run the model without specifying any independent variable, i.e.
proc logistic data = mydata descending;model youtcome = /firth;run;
In the output, I found:
Model Convergence Status
Convergence criterion (GCONV=1E-8) satisfied.
-2 Log L = 38.858
2) Run the model by specifying independen |
|
H******r
发帖数: 2879 | 17 谁说markov chain algorithm 不对一定diverge？有时候也可以converge的，有时候
converge到了边界就直接运行不下去了。 |
|
D******n
发帖数: 2836 | 18 在看看书吧。。。。
LLN基本是说sample statistic跟population 的parameter 会converge当n-> infinity
CLT means sum of sufficiently large number of independent variables, converg
e to normal
布可以用正态近似，无论样本本身分布如何。
本的 estimate 越接近真值？ |
|
i******a
发帖数: 27 | 19 谢谢！
point converge 和 distribution converge 我可以理解，就是后面的 WLLN, SLLN 还
有那些什么什么从来没看过。。。。。
其实在复习 probability 的时候肯定有看过记过LLN&CLT，后来就淡忘了。后来见到推导 SEM 的时候又想起了 CLT，印象比较深刻一点。
我发现我就只会考试，考完就忘，拿到实际当中不会用，也贯穿不起来。所以要和大家
多讨论，听你们说比我看多少遍书印象都深刻。
大家多包容啊 |
|
q**j
发帖数: 10612 | 20 请问一般如何大概估计一下要sample多少次，头多少次的要扔掉？
看了chib and greenberg的文章，我对一次sample所有参数有了感性认识。但是对bloc
k by block update还是有问题。这样做每次的conditional distribution由于conditi
onal on的参数不同，都是应该是不同的函数。怎么肯定会converge呢？一个函数conve
rge还好理解，怎么理解一个family的函数一起converge？
当然我可能是杞人忧天了，应该有大牛做过理论证明的。纯粹是我的心理不好接受。请
问有没有人给个形象的解释或者例子。 |
|
w********n
发帖数: 753 | 21 大家好，两个问题。
第一个是R的：
我自己写的一个function，然后run了1000个 data file，是用loop run的，最后有个
warning message： Hessian matrix at convergence is not positive definite;
unstable solution.
我想知道的是这1000个data file中是哪个出了问题，哪个没有converge properly啊？
第二个问题是SPSS的：
有一个survey中的variable如下：
mitbbs university
Mitbbs University
MITBBS UNIVERSITY
mIT-BBS uNIVERISTIY
Other： MITBBS uNIVERSITY
.
.
.
.
我希望把这个variable recode成同意格式，例如“Mitbbs University”， syntax该
怎么写啊？ |
|
a****e
发帖数: 150 | 22 那个真不能叫probabilty theory. probabilty theory还是要建立在measure theory
上的,讲的一般是以下内容:The strong laws of large numbers. Conditional
expectations. Discrete parameter martingales: convergence, stopping times,
and optional sampling theorems. Uniform integrability. Weak convergence:
characteristic functions and the central limit theorem. Elements of large
deviations. The ergodic theorem. 你列的课跟我们系master level的theory课是
一个难度的. |
|
w***e
发帖数: 35 | 23 不知道我的理解对不对。
过去做Bayesian Analysis的时候，可以用gibbs sampling, M/H sampling, and slice
sampling等方法针对 posterior distribution 进行采样。对于采样的结果是否符合
要求，通常会对采样的结果做一个Convergence的测试。如果证明Converge了，就认为
采样的结果可以用于进一步的分析。
的， |
|
R*******c
发帖数: 249 | 24 这个只是看是否converge吧，我想知道的是，它是否converge到所需要的分布
slice |
|
p********a
发帖数: 5352 | 25 ☆─────────────────────────────────────☆
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 |
|
m*****z
发帖数: 357 | 26 可能是uniform convergence，也可能是point-wise convergence，取决于你收敛的时
候取的epsilon跟你的x or y是不是有关 |
|
h**********1
发帖数: 155 | 27 what do you mean can't get true value? based on one simu? or 1000? usually
you run 1000 simu and check whether your estimators are unbiased. check
whether each run is converged. use function hist to see the distribution of
your estimators, they should approximately follow normal distribution if you
've large sample (n=1000).
also if you likelihood is very flat, it's hard to converge and sensitive to
the initial value.
try to supply a gradient function which could help a lot.
try different optimi... 阅读全帖 |
|
n**h
发帖数: 22 | 28 The Cramer-Wold theorem states that if every fixed linear combination of d
random variables converges to a normal distribution, then the d variables
jointly converges to a multivariate normal distribution. Does this theorem
hold when the dimension d goes to infinity? Thanks. |
|
c***z
发帖数: 6348 | 29 Some update: clustering didn't work well
I knew that k-mean won't work since Jaccard doesn't follow triangular
inequality - hence convergence of mean doesn't guarantee convergence of
variance.
I tried hierarchical agglomerative and it didn't work well. I believe the
reason is feature selection - I should have used trigrams and such, instead
of words, as trivial words led to mis-clustering.
I am working on trigrams as well as NN, will keep updating here. Thanks a
lot casact and guys! |
|
c***z
发帖数: 6348 | 30 I have been asking the same question to my boss, about the practical use of
this abstract metric...
The reason we can't use k-mean is that these metrics are actually not real
metrics, as they don't follow triangular inequality, and hence the mean
means nothing (convergence of mean doesn't imply convergence of variance).
The only thing I can think of is then hierarchical clustering... |
|
c***z
发帖数: 6348 | 31 I have been asking the same question to my boss, about the practical use of
this abstract metric...
The reason we can't use k-mean is that these metrics are actually not real
metrics, as they don't follow triangular inequality, and hence the mean
means nothing (convergence of mean doesn't imply convergence of variance).
The only thing I can think of is then hierarchical clustering... |
|
|
s*****e
发帖数: 1368 | 33 第四章主要讲了一些技术指标。这些指标主要分为3种：顺势指标、摆荡指标与杂项指标。虽然有些朋
友说这是一本入门书，但是我却觉得这章对我有些难度，或者说有些枯燥的感觉。
顺势指标相对来说简单一些，比较容易懂，这些指标适用于趋势比较明显的行情。但对于横向走势，
它们所提供的讯号并不理想。不过我倒觉得对我这种初学者来说已经够了，我又不靠它吃饭，看多了
反而消化不了。我主要看了三个指标，移动平均、MACD（Moving average convergence-
divergence，移动平均收敛发散指标）、MACD柱状图。其它的看着头晕，呵呵。可能以后再读的时
候会有感觉吧。
移动平均（moving average,简称MA）比较简单，但是它没有考虑权重的问题，而指数移动平均
(exponential moving average，简称EMA)是一种比较理想的顺势工具。它给予最近资料较大
的权数，对于最近价格变动的反应快于简单移动平均。我想今后我在看K线图时会应用EMA的，看10
日均线和30日均线的关系来与预测股价的走势。这是我看书时的感觉。
但是我看到作者介绍MACD时，（移动平均收敛发散指... 阅读全帖 |
|
s*****e
发帖数: 1368 | 34 第四章主要讲了一些技术指标。这些指标主要分为3种：顺势指标、摆荡指标与杂项指标。虽然有些朋
友说这是一本入门书，但是我却觉得这章对我有些难度，或者说有些枯燥的感觉。
顺势指标相对来说简单一些，比较容易懂，这些指标适用于趋势比较明显的行情。但对于横向走势，
它们所提供的讯号并不理想。不过我倒觉得对我这种初学者来说已经够了，我又不靠它吃饭，看多了
反而消化不了。我主要看了三个指标，移动平均、MACD（Moving average convergence-
divergence，移动平均收敛发散指标）、MACD柱状图。其它的看着头晕，呵呵。可能以后再读的时
候会有感觉吧。
移动平均（moving average,简称MA）比较简单，但是它没有考虑权重的问题，而指数移动平均
(exponential moving average，简称EMA)是一种比较理想的顺势工具。它给予最近资料较大
的权数，对于最近价格变动的反应快于简单移动平均。我想今后我在看K线图时会应用EMA的，看10
日均线和30日均线的关系来与预测股价的走势。这是我看书时的感觉。
但是我看到作者介绍MACD时，（移动平均收敛发散指... 阅读全帖 |
|
f**********g
发帖数: 2252 | 35 Solar Winners: First Solar, LDK
By Eric Rosenbaum 01/24/11 - 07:20 PM EST
http://www.thestreet.com/_yahoo/story/10983623/1/solar-winners-
2 CommentsAdd CommentStock quotes in this article:FSLR, LDK, TSL NEW YORK (
TheStreet) -- Shares of Chinese solar company LDK Solar(LDK_) and First
Solar(FSLR_) led a continued rally in solar shares on Monday. LDK Solar was
up 10% and back near a 52-week high level, while First Solar was charting
new territory, surpassing its highest share price in a year ... 阅读全帖 |
|
f**********g
发帖数: 2252 | 36 Solar Winners: First Solar, LDK
By Eric Rosenbaum 01/24/11 - 07:20 PM EST
http://www.thestreet.com/_yahoo/story/10983623/1/solar-winners-
2 CommentsAdd CommentStock quotes in this article:FSLR, LDK, TSL NEW YORK (
TheStreet) -- Shares of Chinese solar company LDK Solar(LDK_) and First
Solar(FSLR_) led a continued rally in solar shares on Monday. LDK Solar was
up 10% and back near a 52-week high level, while First Solar was charting
new territory, surpassing its highest share price in a year ... 阅读全帖 |
|
m*********k
发帖数: 10521 | 37 本次统计截止时间为：2013-08-13 02:00:00 （美东时间）
成功奖励 20 伪币的用户: Saob, oldgeneral, KingOfLunHui, MaruCat, walksing,
shoonshoon, Airui, florababy, waitingforu, dzxy, kickok, dudunangnang,
Regina, hexieNo1, zhangsan, IlIllIlI, ldy, icq9999, convergence, blackmoth,
lczlcz, weijinger, hahafish, moonflower7, adventure09, tennisgirl, wishhope,
coalpilerd, shaitthis, rhsh, difflife, IcyLi, David1231, tennisfun,
Andrea111, chuyue, luckyou, xiaoxiaoren, tangdoudou, linjiangxian, mesao
奖励版面:(Midlife)20伪币成功
奖励版面:(Fishi... 阅读全帖 |
|
g********n
发帖数: 2314 | 38 【 以下文字转载自 SanFrancisco 讨论区 】
发信人: dadabear (bless you), 信区: SanFrancisco
标 题: [合集] goldenrain的观点未免偏颇
发信站: BBS 未名空间站 (Thu Feb 8 16:24:51 2007), 站内
☆─────────────────────────────────────☆
MSJ (Mission San Jose) 于 (Tue Feb 6 00:42:38 2007) 提到:
看完他blog的《现在在哪里买房风险小》，我只能笑着摇摇头。在我眼里，这个列表与
其说是“风险小”列表，不如说是“房子难涨地区列表”。试论证一二：
1. 常言道“强者恒强，弱者恒弱”。NYC, SF地区房价高企不是一年两年了，支撑房价
的不是那个affordability的percentage，而是能afford的人口数和房子总数的比例。
这里看百分比是没有多大意义的。只要还有人能买得起还愿意买，即使affordability
只有15%又如何呢？
2. 正如high tech公司的P/E要远大于... 阅读全帖 |
|
u**h
发帖数: 509 | 39 【 以下文字转载自 Military 讨论区 】
发信人: aldernetwork (alder), 信区: Military
标 题: 重温《波茨坦公告》
发信站: BBS 未名空间站 (Fri Sep 14 00:10:23 2012, 美东)
１９４５年７月，美、英、中三国首脑和外长在柏林西南波茨坦举行会议，２６日
发表《波茨坦公告》，敦促日本投降。全文如下：
美、英、中三国政府领袖公告：
（一）余等：美国总统、中国国民政府主席及英国 首相代表余等亿万国民，业
经会商，并同意对日本应予以一机会，以结束此次战事。
（二）美国、英帝国及中国之庞大陆、海、军部队，业已增强多倍，其由西方
调来之军队及空军，即将予日本以最后之打击，彼等之武力受所有联合国之决心之
支持及鼓励，对日作战，不至其停止抵抗不止。
（三）德国无效果及无意识抵抗全世界激起之自由人之力量，所得之结果，彰
彰在前，可为日本人民之殷鉴。此种力量当其对付抵抗之纳粹时不得不将德国人民
全体之土地、工业及其生活方式摧残殆尽。但现在集中对待日本之星则较之更为庞
大，不可衡量。吾等之军力，加以吾人... 阅读全帖 |
|
M***e
发帖数: 836 | 40 China moves quickly to deal with deadly earthquake's aftermath
35 minutes ago
BEIJING — Within 24 hours of China's deadly quake, some 20,000 troops
converged on the disaster area to help dig out the dead and injured, and
military planes and trucks ferried in another 30,000 reinforcements.
The rapid mobilization to stricken Sichuan province reflected the priority
that China's leadership places on delivering efficient disaster relief while
showing the world it stands ready for anything that may co |
|
c****s
发帖数: 5892 | 41 圣诞和新年即将到来，各商家已经开始瞄准市场，准备把握住这一年当中最好的商机，无线通讯市场亦不例外。这个以往似乎仅限于罗渣士(Rogers)、贝尔(Bell)和研科通讯(Telus Mobile)这三大巨头之间的竞争，却由于更多以省钱、无合约等为促销手段的小型公司的介入而变得更加激烈起来。
不久前，本国政府通过市场调查了解到，认为国民为通讯所缴纳费用负担过重，希望看到该市场中多一些竞争，以减少民众的负担。于是，从去年年底开始，一些新型的无线通讯公司如 Public Mobile、Mobilicity和Wind Mobile等相继出现，并一致推出无合约和无时限通话的诱人条件，将竞争矛头直接指向现时无线通信市场公认的三大巨头——罗渣士、贝尔和研科通讯。
Convergence Consulting Group的电讯分析专家Brahm Eiley指出，自从这些新型的电讯公司出现的一年以来，一直不断的推出各种折扣，这些折扣包括将通话、短讯和上网费用联合在一起，可以吸引像罗渣士这样的大公司现有的超过一半的用户。
上网费节省逾75%
他说：「仅上网费用(data plan)这一项，就能够节省超过75... 阅读全帖 |
|
O*******d
发帖数: 20343 | 42 If you trace back your linage you will find you are at the terminal of a
continues female line and a continues male line. Everyone's female line
converge to a single female individual living in Africa some 200000 years
ago. This individual is the lucky person who's daughter gives birth to a
daughter who gives birth to a daughter who gives birth to a daughter .......
... who gives birth to your mother. All other female individuals at the
time have broken female linkage in their offspring.
The b |
|
k***g
发帖数: 7244 | 43 多个啥，2002年的时候差不多 18% 左右，最近的数据没有留心看过，但是可以肯定的
是只少不多，估计会 converge 到欧洲 8% 的水平。 |
|
w**********y
发帖数: 1691 | 44 然后你们讨论的那个姓的概率模型:
假设一开始有100对夫妻:99对夫姓王,1对夫姓李.
每一对生俩个孩子,男女比例相当.
如果生男的,抓一个胡人来结婚,跟这个男的姓
如果生女的,也抓一个胡人来结婚,跟胡人的姓(比如姓金)
那么中原第二代有200对夫妻: 平均值是 99王+1李+100金
第三代有400对夫妻: 99王+1李+300金
非要写统计模型的话:
第n代王姓夫妻的比例是 P= 99%*W1*W2*...*Wn=99%*exp(logW1+logW2+..logWn)
Wi是下一代生的孩子是男的比例(不是概率)..比例是个random variable,W~Binomial(100*2^n,0.5)/(100*2^n)
根据大样本理论(这里需要更高级的大样本理论,因为Wi不是同分布的,需要重新构造),(logW1+...+logWn)/n converge in probability to E(logW)
E(logW)<0.那么最终的结果是P 收敛到0.
第n代的李姓也是一样...
长此下去王和李都灭亡了..大家都一样....
这说明什么...这个模型是shit..很多假设都 |
|
m***y
发帖数: 14763 | 45 Second this, playing dumb won't go far.
Especially on Bullet 2, Biao was clear (actually everybody knew) that Mao's
roadmap is
anarchy->Mao's dictatorship-> balanced collective oligarchy in the near
future after his death -> power converges to Mao's family (Qing Jiang->
Yuanxin->Xinyu)
But it's gonna be hard. Mao changed his style after 1949, from uniting his
sidekicks together to turning them against each other. And this roadmap only
works out by maintaining delicate power balance between fract... 阅读全帖 |
|
k*********g
发帖数: 791 | 46 not really.
the generic drift is kind of random walk process, which drifts (except one
dimensional random walks).
but why not so many chinese surnames is due to a converging mechanism: a
surname is either passed to next generation, or terminated, but not created.
hence, that not so many chinese surnames tells us china is an ancient
civilization with uninterrupted culture.
will |
|
p***r
发帖数: 20570 | 47 参数当然有，但是不同计算都是用这些共同的参数。还有一些参数在一定范围内就
converge，不影响你的结果了。 |
|
w****j
发帖数: 5581 | 48 突然想起陶行知的名字改来改去，最后converge on行知，这下所有可能的组合被遍历
了啊。知然后行，行然而知和知性合一。呵呵 |
|
T*****y
发帖数: 18592 | 49 聚合与离散（convergence and divergence）是全球史另一个重要研究视角，青铜与游
牧文化的离散和中国民族文化的聚合是两个极好的例证。从青铜时代世界体系的中心来
看，西亚青铜冶炼技术和中亚游牧文化全球扩散。青铜技术起源于西亚，首先扩散到中
亚、地中海地区（南欧和北亚）、南亚，然后扩散到欧洲大部、东亚和东南亚、南部非
洲，乃至整个旧大陆及其附近岛屿；与此相关的文化要素有小麦、砖和黄金崇拜。游牧
文化形成于中亚，然后向四面八方扩散，与印欧文化的形成和印欧人的扩张密切相关。
青铜时代世界体系是由西亚中亚向周围扩张形成的，五千年前局限于西亚及其附近地区
，四千年前扩展到中亚地区，三千年前普及到几乎整个旧大陆。另一方面三代文化是聚
合而成。本土起源的猪、狗、鸡和外来的牛、马、羊组成了六畜，外来的麦与本土的稻
、粟、黍、豆构成了五谷。牛耕姗姗来迟，耒耜或锄头一直是基本的生产工具，犁耕与
锄耕互补形成了东亚农业传统。丝绸西传，毛毯东播，丝毯是东西文化交织的象征。舟
船源自中国，车马来自中亚，舟船车马在中原交汇。夯筑是本土起源，秦砖或周砖源自
西亚，夯土城墙外砌砖边是东西合璧。土葬... 阅读全帖 |
|
z**********e
发帖数: 22064 | 50 http://www.peruviantimes.com/06/beyond-machu-picchu-choquequira
Beyond Machu Picchu -- Choquequirao, Lost City in the Clouds
By Gary Ziegler --
The first rays of morning sunlight illuminate the great stone altar,
streaming through a square opening over my head. “Inti camac sumac,”
chants the priest. Soaked in sweat, I fight the bindings holding me to the
stone as the grinning, looming, scarlet-cloaked figure slowly brings down a
gleaming, bloodstained bronze knife toward my heaving chest.
“Jefe,... 阅读全帖 |
|
