由买买提看人间百态

topics

全部话题 - 话题: randomness
首页 上页 1 2 3 4 5 6 7 8 9 10 下页 末页 (共10页)
L****a
发帖数: 572
1
This is a basic quesiton of probability.
Y = (X+Y) - X = (X+Y) + (-X).
the pdf of the sum of two random variable is the convolution
of the pdf of the two random variable.

y
y)
j****8
发帖数: 1
2
Hi,
I have a probability question as follows:
x1, x2, ... ,xn are n indepedent random variables. if y1 is the largest value
of x1 to xn, y2 is the second largest vaule of x1 to xn, ..., yn is the
minimum value of x1 to xn. Obviously, y1, y2, ..., yn are n random vaibles as
well. The question is:
Are y1, y2, ..., yn indepedent to each other? How to prove it?
Thanks a lot!
Jeff
w**a
发帖数: 1024
3
来自主题: Mathematics版 - random variables problem
Given random variable X , Y
how to calculate E[(X-Y)^2]?
please correct me if i am wrong,
because X is a random variable, so it has a distribution Dx; Same to Y, Y 's
distribution is Dy. Then,
I can draw N ( N is large) numbers from Dx, x1, x2, x3, ..., xN
Then draw N numbers from Dy, we have y1,y2,y3,...,yN.
Then I do,
E'=\sum_{i=1}^{N}\sum_{j=1}^{N} (x_i - y_j )^2 /N^2
Will E' a good estimate of E[(X-Y)^2]? Thanks.
r****y
发帖数: 1437
4
来自主题: Mathematics版 - a question about random number generator
e.g., given a covariance matrix
cov = [ 1 0.02
0.02 1]
how to generate two random number series that have such covariance?
let's make it more specific, two random numbers both follow normal
distribution.
I guess there must be an algorithm invented already by someone for
such thing.
o******b
发帖数: 37
5
来自主题: Mathematics版 - Help on random integer generator
what you actually need is not a random number generator, what you need is a
random permutation of {1-N}. I think Knuth shuffle does exactly what you
want. Just search it, it is very easy to implement

jl
发帖数: 398
6
A random varible converge to -\infty in provability
是什么定义? 多谢!
A random varible converge to -\infty with probability 1 s
是什么定义? 多谢!
w*****g
发帖数: 9
7
来自主题: Physics版 - Random Walk question
I am afraid this is a hard problem and you need to be more specific. As far
as I know, for a Markov random walk without the potential, the diffusion
distance is proportional to D*t^2 where D is the diffusion constant. Then
the correlation function is in a exponential form:
exp(-\sqrt{D} \tau). It is easy to get the power spectrum by doing the
Fourier transform. I think it is delta function, but not sure.
For other random processes (e.g., Gaussian) or including the potential, I
don't have a clue.
r*****t
发帖数: 286
8
☆─────────────────────────────────────☆
terahertz (亲爱的) 于 (Sun Feb 18 09:42:41 2007) 提到:
被问到一个symmetric经典random walk问题。简化如下:
A symmetric random walk start from 0, will either hit level -a or b.
a) what's the probability it hit a before b;
b) what's the expected wait time to stopping time?
我知道a), 不过没搞定b), 考古也没有找到满意答案。不知道那位高人能给个提示?
我这里也有个提示:E[W^2-T]=E[0]=0 to solve for E[T]....
另外一道题:martingale, coin toss; if you throw HT, you got 2 dollar, if HHT
, you got 4 dollar, if HHHT, you got 8 dollar
g****y
发帖数: 71
9
convergence in probability usually refers to the law of large numbers. e.g.
average of iid Xi should converge in p to its expectation.
convergence in distribution usually refers to the central limit theorem.
e.g. square root of n times (average of iid Xi minus its expectation) will
behave like a normal distribution. it converges in distribution to a normal
random variable. but indeed it's not a normal random variable even in the
limit.
b***k
发帖数: 2673
10
☆─────────────────────────────────────☆
sinoers (某人) 于 (Fri Oct 12 13:30:45 2007) 提到:
假设一个random walk,左走一步概率是p,右走一步概率是q=1-p。
启示位置是0,请问在走到N步的时候,在之前都不触及边界
的前提条件下,走到位置 k 的概率。
两个边界分别是 Ra<0, Rb>0。
另外, Ra < k < Rb。
我只知道只有一个边界的情况下好解,用镜像法,但是
不知道两个边界的情况下如何解。
☆─────────────────────────────────────☆
sinoers (某人) 于 (Fri Oct 12 13:56:35 2007) 提到:
ok,自己解了,还是镜像法。

☆─────────────────────────────────────☆
gatsby (gatsby) 于 (Sat Oct 13 02:18:09 2007) 提到:
这个问题和random walk没什么关系,就是投N个硬币,最后有(N+k)/2个头的概
l*0
发帖数: 19
11
来自主题: Quant版 - 问个drift random walk问题
random walk problem.一个人在原点,距离左边距离a,右边距离b
如果此人向左向右follow standard brownian motion
问:1)在达到b之前,到达a的概率是多少
2) E(T)
这个答案大多数人都应该知道,分别是 b/(a+b), ab
现在问题是如果是shift random walk
dX_t = r dt + sigma dW_t
同样上面两个问题,答案应该分别是多少?请教了

D**u
发帖数: 204
12
来自主题: Quant版 - spherical random walk (转载)
【 以下文字转载自 Mathematics 讨论区 】
发信人: DuGu (火工头陀), 信区: Mathematics
标 题: spherical random walk
发信站: BBS 未名空间站 (Wed Nov 4 15:15:16 2009, 美东)
A guy starts walking from position x_0 on earth (we treat earth as a sphere
with radius R). He randomly chooses a direction and walk 1 meter (this is
the spherical distance) and reaches position x_1. He repeats the walk n
times and arrives at position x_n.
Question:
(1) what is E((x_n - x_0)^2)? (here ((x_n - x_0)^2 is the square of the
Euclidean distance, not t
w******l
发帖数: 58
13
来自主题: Quant版 - a random walk problem
Given a 1-D random walk starting at 0, with equal probability of going +1 or
-1, what is the probability P(n) that after n steps the random walk stayed
above 0, never hitting 0?
b**********5
发帖数: 51
14
来自主题: Quant版 - Lognormal Random Walk
Is Geometric Brownian Motion a Lognormal Random Walk? If so, why it's
lognormal, I
mean how could we know it's lognormal?
What are the usually used stochastic process with Lognormal Random Walk?
Finally what Geometric mean in the term Geometric Brownian Motion?
Thanks for idea in advance.
D**u
发帖数: 204
15
Randomly (also uniformly) choose 3 points A,B,C on a 2-dimensional sphere.
Let x be the spherical area of the spherical triangle ABC;
and y be the volume of the (Euclidean) 四面体 formed by A,B,C and O
(the center of the sphere).
If we treat x and y as 2 random varibles,
then what is the correlation between x and y?
y****n
发帖数: 60
16
给的答案是:
int num = 5*(rand5()-1) + (rand5()-1);
if(num<21) return ( num%7+1);
不明白为什么第一行代码要两个相加?前面一个不就是0 到20 的uniform
distribution 了吗?两个random number 加之后并不是uniform 的distribution 啊?
谢谢。
S*********k
发帖数: 507
N***c
发帖数: 1090
18
【 以下文字转载自 Physics 讨论区 】
发信人: Novac (Bcool), 信区: Physics
标 题: 一个关于random close pack的问题
发信站: BBS 未名空间站 (Fri May 30 15:45:49 2008)
最近在研究一个课题,是关于如何使不规则形状的物体更密实的堆垛的问题。这些物体
大多是立方的,片状的,也有可能是球形的,无法选择。这个题目也可以变化成什么样
的容器可以使这些物体堆得更密。
网上找了一圈没看到什么太相关的文献。感觉应该跟random close pack有关。大家觉
得这个题目跟哪个领域比较接近呢?
s**c
发帖数: 1247
19
来自主题: Statistics版 - random sampling in R
create 1000 uniform random samples, one for each obs
sort by the random sample, select the first 200
B******y
发帖数: 9065
20
呵呵,看来你对seed没有真正的了解。SAS不可能产生真正的Random Number,而只是
pseudo-random number,即使你用0做seed,也只是2^31-1个整数中的一个,其取决于
你运行程序那一刻的时间而定,你换个时间重新运行,看上去和原来不一样,但其实只
是重新挑了一个另一个整数而已。选择一个确定的正整数做seed可以方便别人复制你的
结果来检验你的程序的正确性,取0做seed和你每次任意换一个整数没有本质区别。
o****o
发帖数: 8077
21
in your sampling scheme, you will select all records that have the same
prefix number as the one you selected, which means the rest are not randomly
selected but only the first one, which amounts to select the numbers 1,...,
100
so that you use proc surveyselect to randomly choose 50 numbers and merge
with the original data on their numbering
s*r
发帖数: 2757
22
来自主题: Statistics版 - 这样还能算Randomized sample吗
randomized is not the same as random
s******n
发帖数: 134
23
Thanks very much for the reply. I really appreciate it. How to obtain a
random sample of size 1000 from f(x,y)=c? I know how to generate random
sample from normal distribution,dirichlet distribution,and so on. Thanks a
lot~
g********r
发帖数: 8017
24
我也是一知半解,欢迎拍砖:
比如说在板上随机挑出几个人来,到各大饭店品尝红烧肉.作为评价红烧肉好坏,Y就是各
人吃的量.X呢,我们感兴趣的是红烧肉的色泽,肥瘦比例之类的.
决定吃多少,还有个重要的参数就是各人的食量.比如版主大人,根据ID和雅号,那一定吃
的比较多.可是呢,这个不能当作一个fixed effect.尤其是观察量比较少的时候.操作上
来说,会吹大估计值的方差.哲学上说,我们不应该给板上好几百口子每人一个fixed
effect, 而应该把它当作从一个population里的抽样.所以就叫random effect了.
有的人,比如我老板,对random effect非常的不鸟,认为那是缩小方差的一个小手段.
F****n
发帖数: 3271
25
Effect == regression coefficient for a IV (or intercept)
Fixed Effect == the coefficient is a constant for your population
Random Effects == the coefficient is itself a random variable. Normally it
assumes there are sub-populations in your original population, each of which
has a different coefficient for a given IV, and can be estimated based on
some variables at the sub-population levels (group effects).
O*****y
发帖数: 222
26
来自主题: Statistics版 - Approximate random sample
Given a number of vectors if lengh m. The elements if every vector are from
the set{1,2,...,n} m resulted collection approximate a random sample of vectors. That is , one
can treat each vector in the collection as if it was obtained by randomly
select m integers from the set. Thanks.
n**m
发帖数: 156
27
来自主题: Statistics版 - repeated measures & random factor
学anova用的是ott的书,说的都是fixed factor, random factor。我今天读了
repeated measures的概念,觉得repeated measures 里面的subject(比如一个学生,
连续三年量身高),学生在这里就是 random factor。不知道说的对不对。
e*****t
发帖数: 642
28
有两个group的基因list,A组大概有3000个,B组有4000个。
他们之间有大概500的overlapping。怎么证明这500不是random的?
bionominal test?还是弄个contingency table? 但是这两个test怎么来说明
overlapping不是random的?null hypotheis 是什么?
e*****t
发帖数: 642
29
i said overlapping between these two groups, which means "chosen" is random
or not.
i think i don't worry about 3000 genes are randomly chosen to be in group A
or in group B. They've been already there. I am considering the probability
of choose 500 out of the pool.
I kinda have the idea. Probably conditional prob. will solve the problem.

the
of
g***l
发帖数: 22
30
来自主题: Statistics版 - fixed or random model
How can there be lots of levels? L,M and H are predetermined before
collecting data.
And if there are only three samples it doesn't make sense to use random
effect.

factors
effect.
random
g***l
发帖数: 22
31
来自主题: Statistics版 - fixed or random model
How can there be lots of levels? L,M and H are predetermined before
collecting data.
And if there are only three samples it doesn't make sense to use random
effect.

factors
effect.
random
S*********k
发帖数: 507
32
来自主题: Statistics版 - 求电子书: statistics of random processes
求电子书: statistics of random processes
by Liptser
http://www.amazon.com/Statistics-Random-Processes-Robert-Liptse
w*******9
发帖数: 1433
33
来自主题: Statistics版 - A definition about Random Correspondence
看不懂。大师能不能多说点?"In a given sample space, a random correspondence
could be DEFINED as x<--->y." 这只是记号,没看见定义啊。如果x和y不是
pairwisely observed, you can NEVER recover the relationship between x and y.
I assume your "random correspondence" is some sort of measure for
correlation.
h******1
发帖数: 1930
34
如题,小弟遇到些问题,诚心求帮助!
关于design & analysis of experiment.
randomized block design, completely Randomized Design, Latin Square Design.
如果哪位刚好懂这方面的知识,请私信联系!!!
c********h
发帖数: 330
35
如果一个factor的levels是fixed,比如gender,只有male和female,一般就
incoporate as fixed effect
如果一个factor的levels只是对应于某个实验或者observational study,比如说
batches,subjects,换了别的实验就不是这些levels了,那么一般用random effect
如果一个model have both fixed effects and random effects,那么就是mixed
effect model
F****r
发帖数: 151
36
X and Z are observed data (factor levels)
Beta are estimated fixed effects.
mu are estimated random effects.
You may notice the difference between beta and mu:
beta's are unknown constants to be estimated. this is the fixed effect.
mu is assumed to follow a distribution. This is the random effect.
c***z
发帖数: 6348
37
Weird, I found the Chinese version harder to understand...
If you are talking about panel data, here is a link
http://www.princeton.edu/~otorres/Panel101.pdf
In short:
The fixed-effects model controls for all time-invariant
differences between the individuals, so the estimated
coefficients of the fixed-effects models cannot be biased
because of omitted time-invariant characteristics…[like culture,
religion, gender, race, etc]
One side effect of the features of fixed-effects models is that
they c... 阅读全帖
b*********n
发帖数: 2975
38
use sample 1 after your code
or generate random number function and order it and top 1 like SAS did

I want to select randomly one observation from each group_id.
If it can be done in subqueries, it's even better.
Thanks.
b*********n
发帖数: 2975
39
sample 1 is from teradata itself, should very efficient,
however, if you want to have a very good random generator, you may try some
numerical libary by Fortran/C++ and then
insert the generated random value into your table, then top 1
it also not a big deal for teradata

this is a good way. Thanks.
BTW, is it efficient? I was hoping there is a generic way of doing it.
F8
发帖数: 348
40
来自主题: Statistics版 - a quesiton about random effect
我的问题是这种情况下用 nlmixed 怎么来 fit random study effect
glmer 相当于sas glimmix ?
能解释一下什么意思吗
哪个是fixed/random effect in this model
study 是categorical 还是 numeric
c**********e
发帖数: 2007
41
来自主题: Statistics版 - Use random effect model on panel data?
When the number of repeated measure is not many, random effect model is good
. I know people use mixed model often in pharmaceutical industry.
But in financial industry, the repeated measure is so many, say unemployment
rate each country in the last 20 years (240 monthly obs). Do we still use
mixed model or random effect model to do panel analysis? Or should we just
use linear regression?
Anybody has experience on this?
a**********0
发帖数: 422
42
【 以下文字转载自 JobHunting 讨论区 】
发信人: apprentice00 (数学学徒), 信区: JobHunting
标 题: weka有支持regression tree的random forest吗
发信站: BBS 未名空间站 (Sat Aug 23 00:08:21 2014, 美东)
我的问题不是classification 而是regression 所以需要由regression tree构成的
random forest 不知道weka有没有
t*****e
发帖数: 364
43
这个自己写一下应该不难吧:
1. boostrap 100-200 datasets from your training set
2. For each of the bootstrapped data, randomly select some features from
those features that are NOT in the important feature list (the number
selected could be a tuning parameter, usually random forest recommend p/3
for regression tree and sqrt(p) for classification tree).
3. Combine features obtained in step2 with the important several features
4. fit a decision tree using rpart (without pruning) and predict your test
set (eithe... 阅读全帖
t*****e
发帖数: 364
44
这个自己写一下应该不难吧:
1. boostrap 100-200 datasets from your training set
2. For each of the bootstrapped data, randomly select some features from
those features that are NOT in the important feature list (the number
selected could be a tuning parameter, usually random forest recommend p/3
for regression tree and sqrt(p) for classification tree).
3. Combine features obtained in step2 with the important several features
4. fit a decision tree using rpart (without pruning) and predict your test
set (eithe... 阅读全帖
w*******y
发帖数: 60932
45
13 in 1 USB 2.0 SD/SDHC/Micro SD Card Reader, Random Color:
http://www.meritline.com/13-in-1-sd-sdhc-microsd-card-reader---p-41418.aspx
$0.49 @Meritline shipped with coupon MLCK4006470831AL1, Limit 800 Order
only.
13 in 1 USB 2.0 SD/SDHC/Micro SD Card Reader, Random Color
Features
Support SD/MMC: SD,MiniSD,MMC,MMC,Plus,SDHC MMC,MMC Turbo,RS-MMC Mobile,RS-
MMC,MMC Turbo.
TF: T-Flash,MicroSD Card.
Windows98/ME/2000/XP/Vista,Mac OS 9.1 or later.
Draw Power directly from USB Port.
LED for bus acti
G*******s
发帖数: 10605
46
【 以下文字转载自 SmartShopper 俱乐部 】
发信人: GoRockets (火箭加油), 信区: SmartShopper
标 题: Old Navy random secret code 20% - 30% discount thru 6/28
发信站: BBS 未名空间站 (Sun Jun 26 08:08:35 2011, 美东)
The codes randomly generate when you visit old navy site: http://oldnavy.gap.com/
onsecret20 for 20% off
onsecret25 for 25% off
onsecret30 for 30% off
deal source and details:
http://userquote.com/forum/index.php?topic=524.msg733#msg733
w*******y
发帖数: 60932
47
Amazon will have the Serious Sam series on sale for 75% off individual games
, or deeper discounts on 4-packs. Plus customers who buy ANY of the
individual games or 4-packs will receive a $1.50 promotional code for
Serious Sam HD: The Second Encounter - Legend of the Beast DLC Pack:
http://iway.org/9558971
Here's the Deal of the Week landing page.:
http://iway.org/9558981

Serious Sam 3:
http://iway.org/9558991
- $9.99 (75% off)
Serious Sam 3 4 pack:
http://iway.org/9559001
- $30
Serious Sa... 阅读全帖
f*****g
发帖数: 15860
48
来自主题: _TexasHoldem版 - tell of random raise amount
online tournaments, noticed some players like to raise to a random/funny
number, like 888, 222, 2314, etc. i do this sometimes too.
pre-flop, i give it less value, just a common way to disguise/randomize your
"normal" bet ranges, or textbook-like 3.5xBB with XX hand.
post-flop towards the end, i give it more value or take it as an alert. my
immature (lol) theory behind this is:
1) the player is playing his hand in a relatively relaxed mood, and has the
"extra" attention to "make some fun" with t
w***f
发帖数: 20
49

结合其他报道,最近的random shoot并不只发生在晚上,中午下午时段均有发生,不只
针对成年人,伤亡者有不满1岁,6岁及十几岁的小孩
x*****i
发帖数: 287
50
【 以下文字转载自 EE 讨论区 】
发信人: xiguapi (guagua), 信区: EE
标 题: 请教:用Ansys模拟random media的有效permittivity
发信站: BBS 未名空间站 (Wed May 28 11:07:08 2014, 美东)
我用matlab做了一个三维空间矩阵50x50x50,里面随机分布着水、泥土还有黄铜三种物
质颗粒,一个网格就代表一个种物质。
我能把这个geometry导入到Ansys里面算高频的effective permittivity吗?
有没有关于这方面的instruct或者manual?
谢谢。
首页 上页 1 2 3 4 5 6 7 8 9 10 下页 末页 (共10页)