h***o
发帖数: 26 | 1 Is there a fast way to obtain the largest/smallest eigenvalues of a
symmetric matrix? or the upper bound for the largest and lower bound for the
smallest?
The matrix is large and sparse.
Thanks a lot |
|
S*******t
发帖数: 97 | 2 在做newton-raphon迭代的时候经常要inverse matrix，
经常是进行到4，5个循环的时候matlab就报warning,
因为matrix基本上已经是singular了(det = 0).
请问有经验的大侠是如何处理这个问题的？ |
|
X*****r
发帖数: 2521 | 3 【 以下文字转载自 EE 讨论区 】
发信人: Xfilter (支持南开的兄弟们), 信区: EE
标 题: 请教，Markov chain里，如果transition matrix
发信站: BBS 未名空间站 (Wed Feb 13 22:27:05 2008), 转信
是random的，怎么处理？
就是整个transition matrix是时变得，random的
怎么处理这种问题？
有没有什么相关的分析结果，以及相关的MDP solution？
多谢指点！ |
|
n*s
发帖数: 752 | 4 I have a matrix and it turns out
if lambda is eigenvalue, 2-lambda is also eigenvalue
is this a kind of symmetry? does it help solving the eigenvalue problem?
for example, does this property help in decomposing the matrix? |
|
X*****r
发帖数: 2521 | 5 比如变量是一个psd matrix
有专门对付这种变量是一个matrix的optimization类型吗？
多谢指点！ |
|
h****r
发帖数: 391 | 6 Matrix Mathematics
by Dennis S. Bernstein
Princeton Univ. Press
2005
这本书里有好几种计算matrix exponential的方法 |
|
g****t
发帖数: 31659 | 7 这本书好象几百页的索引，晕死。
Matrix Mathematics
by Dennis S. Bernstein
Princeton Univ. Press
2005
这本书里有好几种计算matrix exponential的方法 |
|
m**a
发帖数: 1840 | 8 比如函数值是matrix,把不等号换成positive semidefinite partial ordering,然后证
明存在一个值是 all zero matrix. |
|
f**n
发帖数: 401 | 9 Assume matrix A is a m*n matrix and r=rank(A)
deficient.
What is the best/quickest way to get a set of r columns (or rows) in A so
that A1,...,A_r are independent from each other?
I know gaussian elimination can achieve this. But I cannot find it with a
reasonable numerical routine.
P.S. LAPACK is available for me. But I do not see any routine that
conveniently does this.
Many thanks! |
|
p**o
发帖数: 3409 | 10 A是一个real symmetric matrix, 非负对角阵D是A的degree matrix
(i.e. D = diag[d_i], where d_i=\sum_j |a_{ij}| )
记 A' = D^{-1/2} A D^{-1/2}，显然也是一个对称阵
请问如何证明A'的最大特征值是1 ？
PS1. 我已经构造出一个x_0使得 A'x_0 = x_0，也就是说1是A'的一个特征值，
但不知道怎么证明它是最大特征值……
PS2. 根据Rayleigh-Ritz定理，最大特征值等于 max_{x^T x=1} x^T A' x
也就是说我只需要证x^T A'x <= 1 for all unit vector x
有高人能提供一些思路么？
先谢过～～ |
|
c**a
发帖数: 316 | 11 Let C is known and C is positive definite.
How to approximate C as
C = x x' + D where x is a vector and D is a diagonal matrix
i.e.
min || C - x x' - D ||^2
s. t. D is a diagonal matrix
Is there an explicit solution ?
Is the problem convex?
Thanks. |
|
a****a
发帖数: 5763 | 12 一般的解hermitian matrix, numerical recipes上有一小节专门讲
这个问题,可以直接用jacobian transformation，或者用Householder
转换成三角阵然后用QL迭代
或者干脆写成 2nx2n的实matrix 求解
if C=A+iB is hermitian
(A+iB)\cdot (u+iv) =\lambda(u+iv)
then it is equivalent to write as
|A -B | \cdot |u| = \lambda |u |
|B A | |v| |v |
不太清楚解得性质如何，不过后者应该能保证eigenvalue是实数吧
具体参见Numberical Recipes third edtion page 590
I
including |
|
b**g
发帖数: 949 | 13 this is ok for a temporary walk around, but would cause some potential issue.
for instance, this allows Matrix[i].push_back(T&) or Matrix[i].clear() which is usually not nice. |
|
A*******s
发帖数: 3942 | 14 root matrix that has the stochastic (transition) matrix property (row sum=1
and non negative real elements) |
|
A*******s
发帖数: 3942 | 15 if B^i=A, , B and A are both square matrix, B is A's i-th root matrix |
|
a****o
发帖数: 42 | 16 For such large scale problem, before you do anything
concrete,
it might be a good thing to first try to find some way of
not
inverse the matrix.
Some thing you might consider:
1. If the matrix sparse? I guess it should be.
2. Are you really want to have a 1/A, or you just want an
answer to AX=b?
3. Except for symmetry, are their other characteristic you
can
exploit?
If all the answer is NO, perhaps what yhht said is right: do
the
inverse directly using some method, the reason is that for
such
la |
|
B***y
发帖数: 83 | 17 这是道非常有意思的题目。涉及到凸集结构。
首先在一个 Banach space B 中，
我们定义一个凸集 C 为
1) If x, y \in C, then for t \in [0, 1], tx + (1-t)y \in C.
现在定义 C = set of all n*n seme-positive definite matrix A = (aij), with each
element aij \in [0, 1].
定义 C_k = set of all n*n seme-positive definite matrix A = (aij), with each
element aij \in [0, 1], with rank >= k.
那么 C 在所有 n*n 对称矩阵中，给出了一个凸集。更重要的是 C_k 也是凸集。
凸集的一般事实是：如果存在于有限维空间中，则总存在 basis, 即线性
无关向量集。( reference: 张恭庆，关肇直：线性泛函分析讲义，convex analysis,
1977
J.T. Marti.)
Theorem ( Minkows |
|
S******y
发帖数: 1123 | 18 What is the difference between
table
data frame
matrix
?
Can I just perform matrix computation (such as product, addition,
subtraction) between two data frames?
Thanks |
|
s*****n
发帖数: 2174 | 19 1. There is no "table" in R.
2. Data frame and matrix are totally different things. A data frame is
basically a list, with some extra attributes. A matrix is a vector with some
extra attributes.
3. You can convert them in certain cases, but not all. |
|
o******6
发帖数: 538 | 20 ☆─────────────────────────────────────☆
qian091 (november) 于 (Wed Mar 5 12:09:13 2008) 提到:
请教：已知error属于multivariate t distribution,covariance matrix 的MLE是什么
形式？和normal下的（n-1)/n * sample covariace matrix不同吧。
thanks!!
☆─────────────────────────────────────☆
jesonchang (jesonchang) 于 (Wed Mar 5 21:09:07 2008) 提到:
t is actually a scaled mixture normal. It seems that there is no closed form
solution for MLE, but EM algorithm should work definitely.
☆─────────────────────────────────────☆ |
|
a***r
发帖数: 420 | 21 谢谢~我明白你的意思了
不过我的主要问题其实是要把一个matrix里的数值和相应信息extract出来，成为一个
简明的list，因为原来的matrix很大，不可能直接看 |
|
n******0
发帖数: 298 | 22 matrix有2m列，要把它分成两个matrix，第一个含前m个列，第二个含后m列。谢谢。 |
|
G**Y
发帖数: 33224 | 23 create a new matrix, something like
y1 = matrix(rep(y, each=5), 5, 2)
then
x/y1 |
|
y*****h
发帖数: 34 | 24 做simulation，有true covariance matrix
还有 estimated covariance matrix
请问各位高手，用什么来衡量估计的好不好啊，在矩阵的情况下，相对应的bias, MSE,
应该怎么计算啊。。
谢谢！ |
|
s*******d
发帖数: 132 | 25 假设A 为 matrix
在R里，solve(A) 会出错。
有啥语句判断singular matrix |
|
f******9
发帖数: 267 | 26 请教一个关于R的问题,有一个matrix，对角线两侧是对称的，如何把它变成只含有对
角线一侧的matrix，并且再如何把它
变成vector，如下所示，谢谢啦！
[,1] [,2] [,3] [,4] [,5] [,6]
[1,] 1.000 0.032 0.118 0.029 0.008 0.007
[2,] 0.032 1.000 0.057 0.166 0.002 0.232
[3,] 0.118 0.057 1.000 0.005 0.002 0.001
[4,] 0.029 0.166 0.005 1.000 0.042 0.188
[5,] 0.008 0.002 0.002 0.042 1.000 0.022
[6,] 0.007 0.232 0.001 0.188 0.022 1.000
[,1] [,2] [,3] [,4] [,5] [,6]
[1,] 1.000
[2,] 0.032 1.000
[3,] 0.118 0.057 1 |
|
f******9
发帖数: 267 | 27 How to find the numbers which are more than 0.5 in a matrix (and with their
rownames and colnames)?
for example, in this matrix:
[,1] [,2] [,3] [,4] [,5]
[1,] 1.0000000000 0.0320075758 0.1183712121 0.0293040293 0.0083333333
[2,] 0.0320075758 1.0000000000 0.0571428571 0.1668650794 0.0004662005
[3,] 0.1183712121 0.0571428571 1.0000000000 0.0053418803 0.0028846154
[4,] 0.0293040293 0.1668650794 0.0053418803 1.0000000000 0.0426767677
[5,] 0.008333 |
|
R*********r
发帖数: 225 | 28 which(X>0.5)
> X <- matrix(rnorm(25, 0,1),nrow=5)
> X
[,1] [,2] [,3] [,4] [,5]
[1,] 1.1050564 2.3754090 1.3176173 -1.5279143 1.1355724
[2,] 0.4067519 0.2387128 -1.1444963 -0.6691630 -0.8591197
[3,] 0.4040506 0.3869183 0.2218467 -0.3432441 0.9440614
[4,] -0.5409871 0.4164543 1.1699040 -0.2741125 0.5039738
[5,] -0.5846117 1.0009905 -1.9204391 1.0342027 0.1003464
In R, matrix's entries are indexed column by column.
thus
X[1,2] == X[6]
Not recommended but |
|
r********e
发帖数: 33 | 29 Where to get free C++ library for matrix computation?
I need to do matrix multiplication, inverse, trace, diagonal, rowSum,
colSum etc, eigen value/vector, Hessian, SVD, etc..
Are they available for public for free?
Thanks!
Happy New Year! |
|
b*****n
发帖数: 685 | 30 我数学不是很好，老是靠经验，试了试：
eigen(cov(matrix(rnorm(10), nrow = 2)))
eigen(cov(matrix(rnorm(10), ncol = 2)))
您数学牛，给解释一二？ |
|
q**j
发帖数: 10612 | 31 the model is for srue temporal. i have aobut 400 time period to estimate a
12 by 12 matrix, and i want the estimate to be smooth. if this is too large,
i can reduce the matrix to 8 by 8. any good suggestions? i am open to any
suggestions, bayesian model or not.
estimation |
|
g**a
发帖数: 2129 | 32 A <-read.table("D:/Course/matrix.csv", header = T, sep = ",",as.is=T) |
|
c*****a
发帖数: 808 | 33 这个我就不会了。
如果一是matrix什么的，你可以加上一个新的时间column啊。as.matrix和 as.data.
frame之类乱convert下，左边会变成obs 1,2,3,4... |
|
w*******9
发帖数: 1433 | 34 So why do you need it to be a matrix? Most of the time the vector is enough.
Did you try the transpose function t()? |
|
D**u
发帖数: 288 | 35 weiwei859 is right
>t(as.matrix(g[2,]))
[,1] [,2] [,3]
[1,] 2 5 8 |
|
s*****r
发帖数: 223 | 36 刚开始学学了一节R， 不要笑话我问这么初级的问题
x <- matrix(1:6, 2, 3)
x[,2] <- c(7,1)
a <- c(4, 6, 8)
问题1
x[(a>5|x[1,]==2)]
我怎么都不理解这个输出是
[1] 2 7 5 6
我可以理解
a>5|x[1,]==2
[1] FALSE TRUE TRUE
X是matrix
这样X[FALSE TRUE TRUE]为什么就是
[1] 2 7 5 6
问题2
b) x[(a>5|x[1,]==2)[1:2],]
这又是怎么得到
[1] 2 1 6
谢谢 |
|
v*******e
发帖数: 11604 | 37
1. x如果当作一个matrix来index的话，需要两个参数，一个行一个列。如果你只提供
一个参数，那么不是把x当作matrix来index的，而是当成一个一维的vector。你求x[c(
FALSE,TRUE,TRUE)]的时候，只有一个参数，所以当成vector来index的，由于x长度是6
，你的index长度是3，所以index是复用的，相当于x[c(c(FALSE,TRUE,TRUE),c(FALSE,
TRUE,TRUE))]，这样在x的值c(1,2,7,1,5,6)中选出了第2,3,5,6个，即得到2,7,5,6
2.你用两个参数来index的，第一个是c(FALSE,TRUE)，选了第二行；第二个参数是在逗
号后面，没给出，所以缺省是选择所有的列，所以你得到c(2,1,6) |
|
s********1
发帖数: 235 | 38 【 以下文字转载自 Database 讨论区 】
发信人: someone111 (some), 信区: Database
标 题: 如何把一个correlation matrix 按照一个table 输入一个 database 里？
发信站: BBS 未名空间站 (Tue Dec 8 16:13:07 2015, 美东)
如何把一个correlation matrix 按照一个table 输入一个 database 里？有大约1000,
000 X 1000, 这么多vectors ，每一对vectors 算一个 correlation, 要输到database
的表里，每一个vector 都有一个常量string 对应，所以不牵扯把vector 输到
database table 里。怎么弄比较好？主要以后可能要拿这些代表 vector 的常量
string 做key join table. |
|
发帖数: 1 | 39 统计学的不好，请教一下大家
问题 ： 给定一组random variable (f1, ..., fm), 如何根据sample 来估计
covariance matrix呢？基于什么度量分析误差呢？
当observation data 数目小于 m 时，该如何处理呢？
sample covariance matrix 可以算作一个选项。
有相关note 或 书请推荐了。谢谢 |
|
w*******y
发帖数: 60932 | 40 Like the titles states, Transformers: 25th Anniversary Matrix of Leadership
Edition [Complete Series]:
http://kids.woot.com/
for $64.99 shipped from Transformers: 25th Anniversary Matrix of
Leadership Edition [Complete Series]:
http://kids.woot.com/
. Brand new.
Currently $152.99 on Amazon and lowest it's been on Amazon was just under $
100.
|
|
|
|
|
|
w*******y
发帖数: 60932 | 45 Bestbuy has Toshiba - Satellite Laptop / IntelCore i3 Processor / 15.6"
Display - Matrix Graphite for $379.
oToshiba Satellite Laptop / IntelCore(TM) i3 Processor / 15.6" Display /
4GB Memory / 500GB Hard Drive
Link:
http://www.bestbuy.com/site/Toshiba - Satellite Laptop / Intel® Core™ i3 Processor / 15.6" Display - Matrix Graphite/4034101.p?skuId=4034101&productCategoryId=pcmcat247400050000&id=1218450210501
|
|
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: The Ultimate Matrix Collection [Blu-ray] $25.49
发信站: BBS 未名空间站 (Sun Jan 22 08:52:50 2012, 美东)
Amazon.com has The Ultimate Matrix Collection (Blu-ray) for $25.49 with free
shipping. Best price for this set.
http://userquote.com/forum/index.php?topic=1049.msg1302#msg1302 |
|
w*******y
发帖数: 60932 | 47 Did a search and didn't find anything on this sale. Frys.com is having a Blu
-ray sale online. There's tax in CA, might be tax for other states. But
pretty good prices on some titles. This deals is also in store for those
that have a Frys near by. (limit one per customer online)
Apparently there's a glitch on the site and free shipping will only show up
properly if you order one at a time.
$7.99 Blu-rays with free shipping:
http://iway.org/9557891
Unforgiven [Blu-ray] $7.99
The Hangover [Rated/... 阅读全帖 |
|
s****n
发帖数: 4902 | 48 【 以下文字转载自 Movie 讨论区 】
发信人: shhyin (pat), 信区: Movie
标 题: 有人来谈谈Matrix的虚实转换么？
发信站: BBS 未名空间站 (Thu Feb 11 14:36:18 2010, 美东)
后面一截似乎好懂，就是进去又出来什么的，可是前面neo被trinity叫走究竟在matrix
里还是2199的真实世界？Neo的感觉和观众的感觉究竟是个什么关系。似乎逻辑上总有
脱节之处啊。
手里有蓝光碟，随时可以探讨哪节讲啥。 |
|
a*****i
发帖数: 4391 | 49 【 以下文字转载自 Movie 讨论区,原文如下 】
发信人: ayanami (螃蟹,老子横着爬), 信区: Movie
标 题: [zz] Matrix Review
发信站: The unknown SPACE (Mon May 19 15:03:07 2003), 转信
Saw Matrix Reloaded this afternoon.
Executive Summary: It sucked.
For a more detailed review of this 138 minute suckfest, read within.
========================================================================
First, a confession: I haven't been to a movie in years. It's not that I
don't like movies; I love them. It's just that I don't particularly
care for t |
|
a**o
发帖数: 59 | 50 我觉得他们都算不上suck呀。暂且不论世界上头号大傻帽阿汤哥，我觉得minority
report 里构想的未来世界很酷啊。我本是很讨厌科幻片的，最烦什么一堆大怪物啥的，
但是minority report 里的科技既不过分离奇，又让人眼前一亮，连我这种痛恨计算机的
都很向往。我觉得很大一部分原因在于它给人的感觉是这样的技术离我们并不遥远，所以
并不荒谬。
我觉得matrix的作用也是在于它向人们展示了什么叫酷。我看完就一个感觉，够酷。其他
的毛病都可以不追究了。作为一个blockbuster,我觉得它完全达到了他的目的。作为一个
科幻的blockbuster,他也足够炫啦。matrix的确有很多没用的情节，尤其是光屁股的orgy
, 不过大家要求不能太高了嘛。也说不准某些现在看着没用的情节到了下一级就有用了(
比如说agent smith哪天冒充尼奥帅哥和催你替苟且，结果他家老二上少长一颗痣就被揭
穿了）。我承认打斗太多，还打得那么象电子游戏，人都变了色了，腰都变细了。不过，
你不能不服，他打得就是酷。你再给我找出来一个打得比这高的。我不是action迷但我也
绝对佩服MR的acti |
|
