A*********u 发帖数: 8976 | 1 ☆─────────────────────────────────────☆
jiahe (JiaHe) 于 (Tue Jan 29 02:31:20 2008) 提到:
是不是有点混淆了 最大概然数 跟平均数?
能否给个直观的解释?
☆─────────────────────────────────────☆
geography (地理) 于 (Tue Jan 29 09:55:55 2008) 提到:
Ex=X*P{x
☆─────────────────────────────────────☆
bdr (spruce) 于 (Tue Jan 29 11:37:54 2008) 提到:
X takes values 2^1, 2^2, 2^3, ....
Each with prob. 1/2, 1/2^2, 1/2^3, ...
You have , P(X< inf) = 1; EX = inf.
☆─────────────────────────────────────☆
kosine (仰视星辰) 于 (Tue J |
|
w****g 发帖数: 727 | 2 【 以下文字转载自 FleaMarket 讨论区 】
发信人: wzhang (wzhang), 信区: FleaMarket
标 题: [fs] Crysis 3 + Bioshock Inf.coupon
发信站: BBS 未名空间站 (Mon Apr 8 12:09:17 2013, 美东)
我想卖的物品:
【出售】Crysis 3 + Bioshock Inf.coupon
http://www.newegg.com/Product/Product.aspx?Item=N82E16800995145
单张面值:
169
可接受的价格(必须明码标价!):
40
物品新旧要求:
new
邮寄方式要求:
YL
买卖双方谁承担邮寄损失(Required if not code only):
付款方式说明:
Bill Pay or check
or Paypal
其他补充说明:
广告的有效期:
物品来源:
我的联系方式:
bbs mail
二手交易风险自负!请自行验证是否合法和一手卡!: |
|
s******i 发帖数: 9 | 3 在桌面放了bcmwl6.inf还有另外一个文件
学着网上sudo ndiswrapper -i ~/Desktop/bcmwl6.inf
一直在报错,完全傻掉。。。。今天想第一次开始试用Linux呢,结果这个东西就折腾
好久
希望大侠赐教!多谢了! |
|
s******i 发帖数: 9 | 4 错误信息是
julietay@XPS:~$ sudo ndiswrapper -i ~/Desktop/bcmwl6.inf
[sudo] password for julietay:
sudo: ndiswrapper: command not found
julietay@XPS:~$ ndiswrapper -i ~/Desktop/bcmwl6.inf
The program 'ndiswrapper' is currently not installed. You can install it by
typing:
sudo apt-get install ndiswrapper-common
ndiswrapper: command not found
julietay@XPS:~$
大家赐教! |
|
p********y 发帖数: 111 | 5 【 以下文字转载自 Hardware 讨论区 】
发信人: physicsboy (ONE物理男人), 信区: Hardware
标 题: INF文件在那里--laptop不能从stand by恢复问题
发信站: BBS 未名空间站 (Fri May 1 18:06:37 2009)
网上查了很多,一般大家认为要用如下操作:
http://www.laptopvideo2go.com/forum/index.php?showtopic=6779
The 2 settings in the INF that control the GPU side of the power states are:
MapOSD3ToNV For Laptops based on all CPU other than Centrino.
MapOSD4ToNV This is for Centrino based laptops and newer.
Both these with a setting of 3 places the system in it's deepest sleep (
hig |
|
|
a****6 发帖数: 342 | 7 请问各位一个电脑的问题:
我们新买了一个seagate (Free Agent Go driver)的移动硬盘来备份server 上的东
西(设定的是每天自动备份,旧的移动硬盘一直没问题,都能成功备份)。
把这个新移动硬盘连上server后,电脑上的杀毒软件Symantec Endpoint Protection
总是 block 这 个Autorun.inf , 所以备份总是不成功。
请问怎么解决阿?该怎么改这个杀毒软件的设置阿?
非常感谢。 |
|
m*****0 发帖数: 55 | 8 大家好,问个学术问题。
我在使用EM算法的时候,在30个iterations之后,我的log-likelihood就变成了-
infinity。不知道是什么原因造成的。我的EM就是MLE + logistic regression。感觉
没啥问题啊。
M step是为了maximize expection of log-likelihood, 我这里变成-inf,岂不是EM算
法在这个iteration完全没有用?
谢谢了。 |
|
K****n 发帖数: 5970 | 9 先说说model细节吧,logistic regression怎么用的em? 是有prior还是有mixture? 只
要有一个data point的p是0, log likelihood就是-inf,就像debug一样,你可以研究
一下这30个interation是怎么让它变0的。原理上em的likelihood应该是单增的,如果
你没用package的话也说不定是哪儿写错了 |
|
p********y 发帖数: 111 | 10 网上查了很多,一般大家认为要用如下操作:
http://www.laptopvideo2go.com/forum/index.php?showtopic=6779
The 2 settings in the INF that control the GPU side of the power states are:
MapOSD3ToNV For Laptops based on all CPU other than Centrino.
MapOSD4ToNV This is for Centrino based laptops and newer.
Both these with a setting of 3 places the system in it's deepest sleep (
higher is
better)
But some sytems can't resume from this mode and end up with black screen or
lines
or other things.
This requires the MapOSD3ToNV t |
|
g******t 发帖数: 18158 | 11
以下这个是某网络军史爱好者的考证,我不完全赞同他的观点,他提供的某些史料也不
一定就100%准确。但是他还是提供了一些原始资料的,可以一看,大家批判拍砖:
#######################################################
作者:小辣就可以啦 发表日期:2005-4-21 1:40:00
回复
各位看官先看此文的出处
顽强截击在松骨峰
http://202.84.17.11/world/htm/20001004/155999.htm
新华网首页 =国际舞台 =专题报道集 =新华网专题报道
2000/09
抗美援朝大回放
为纪念中国人民志愿军抗美援朝出国作战50周年,新华社20多位军事记者
分赴全国各地,寻访一批健在的战功卓著的高级将领;新华网特设《战将访问记》、《
亲历抗美援朝》、《见证重要战事》等专栏,刊载这组系列专访。本网站所刊登的新华
社及新华网各种新闻﹑信息和各种专题专栏资料,均为新华通讯社版权所有,经协议授
权,禁止下载使用。
请看原文,50年代鼓舞人心的文宣是这样讲述松骨峰之战的
顽强截击在松骨... 阅读全帖 |
|
q********g 发帖数: 10694 | 12 相关搜索: 热导率, 文件, 计算
作者: zhxlhdd2008 于 2010-10-28 16:23
看到有不少人在找热导率计算方面的in文件,我就贡献三个in文件吧,仅供参考。
同时,附件里贴出了我的计算结果。EMD的输出结果(compute heat/flux command
+compute tc command的计算结果)中, “ac.dat”(见附件中的"ac.wmf")是热流自
相关函数(我已经修改了compute_tc.cpp,目前输出的是normalized HCACF,但结果中
给出的还是没有归一化的热流自相关函数,但形状和归一化的是一样的,请注意!)随
m的变化,"tc.dat"(见附件中的"tc.wmf")是热导率随m的变化(m的涵义请参看热导率
计算的Green Kubo离散化公式,见附件"Comparison of atomic-level simulation
methods for computing thermal conductivity”中的(9)式),"tc_time.dat"(见附
件中的"tc_time.wmf")是热导率随时间的变... 阅读全帖 |
|
i**********e 发帖数: 1145 | 13 The recurrence relation for LIS is this:
Let Li be the length of longest increasing subsequence that ends at Ai.
Li = max Lj + 1, where 0 <= j < i and Aj < Ai
Then, the longest increasing subsequence of A is:
max { Li }, 0 <= i < n.
To improve from O(n^2) to O(n log n), the key is to avoid the linear search
to find max Lj. Using a clever table indexing, we can apply binary search.
Using your example,
A={3, 0, 1, 7, 2, 4, 9, 10, 5, 6, 8}
Create another table called B.
B[k] basically answers the f... 阅读全帖 |
|
c****t 发帖数: 19049 | 14 这不是Grandi's series吗。1/2是所谓的Cesàro sum
这思路不对因为S1=(1+1+...)-(1+1+...)=Inf-Inf。Inf和Inf-Inf都是未定值,不
follow实数代数体的rules,不定义Inf及其计算规则前无法assign value
1-S1=1-Inf+Inf=Inf-Inf是未定值
S2=Inf
S2+S1=Inf+Inf-Inf=Inf-Inf是未定值
S2-S1=Inf-Inf+Inf=Inf-Inf是未定值
2x(1+1+1......)=Inf是未定值; 2+2+2+2......=Inf是未定值。除非你定义如此,这两
个未定值当然不等
定义2x(1+1+1......)=2+2+2+2......并不等同于定义2x(1+1+1......)=2S2 |
|
c****t 发帖数: 19049 | 15 这不是Grandi's series吗。1/2是所谓的Cesàro sum
这思路不对因为S1=(1+1+...)-(1+1+...)=Inf-Inf。Inf和Inf-Inf都是未定值,不
follow实数代数体的rules,不定义Inf及其计算规则前无法assign value
1-S1=1-Inf+Inf=Inf-Inf是未定值
S2=Inf
S2+S1=Inf+Inf-Inf=Inf-Inf是未定值
S2-S1=Inf-Inf+Inf=Inf-Inf是未定值
2x(1+1+1......)=Inf是未定值; 2+2+2+2......=Inf是未定值。除非你定义如此,这两
个未定值当然不等
定义2x(1+1+1......)=2+2+2+2......并不等同于定义2x(1+1+1......)=2S2 |
|
d****d 发帖数: 2919 | 16 不矛盾,
P( Ta < inf ) = 1,
不是说Ta=inf的路径绝对没有,实际上永远到不了a的路径也是无穷多条(比如在原地
打转就行了)。只是到不了a的路径数跟Ta<inf的路径数比起来是底阶的inf。
(就跟在实数轴上随机取一点,取到无理数的概率是1一个套路,这不等于说实数轴上
全是无理数。)
但是算E[Ta]的时候,这些永远到不了a的项也有贡献,
E[Ta]= Ta<inf 的项*Ta<inf的概率 + Ta=inf的那些项*Ta=inf 的概率。
Ta<inf的肯定是有限的,但后面那项,其实就是 inf*0。
inf*0 是 inf,有限,或者0都行,就看谁更高阶了。
实际上是那个inf 更高阶,所以最后E[Ta]=inf。 |
|
y**i 发帖数: 1112 | 17
两个数组相差是奇数一样可以填充
比如你的例子
a1 = 1 2 3 4 5 6
a2 = 3
可以把a2写成-Inf -Inf 3 Inf Inf Inf,(这里用Inf表示无穷大,整数的上限)
那么同样的方法(比较中位数扔掉一半)最后得到两个中位数取小的(因为多加了一个
Inf在右边)
如果把a2写成-Inf -Inf -Inf 3 Inf Inf,那么最后得到两个中位数取大的(因为多加
了一个-Inf在左边)
不过对于第二种方法,对这个例子比较特殊,没有等到最后得到两个中位数的时候已经
遇到了a1的中位数3等于a2的中位数3了,所以直接就返回3了
换另一个例子:a1 = 1 2 3 4 5 6 7 8, a2 = 0, 第一种方法返回4和5中的4,第二种
方法返回3和4中的4。 |
|
n**d 发帖数: 112 | 18 不好意思 我们也没预料到居然有50人的上限 而且还满了。。。比赛结束 我们赢了
MnM 7:4 Inflow
01 MnM.Frankie(Z) < Inf.Jerran(T) @SW
02 MnM.iCC(P) > Inf.Jerran(T) @SW
03 MnM.iCC(P) > Inf.Copperhead(P) @LT
04 MnM.iCC(P) < Inf.Looky(Z) @SS
05 MnM.Ember(T) > Inf.Looky(Z) @LT
06 MnM.Ember(T) < Inf.Spades(T) @MO
07 MnM.DPR(Z) > Inf.Spades(T) @MO
08 MnM.DPR(Z) > Inf.Lexink(Z) @BS
09 MnM.DPR(Z) > Inf.Goodfight(Z) @LT
10 MnM.DPR(Z) < Inf.JTPROG(T) @SW
11 MnM.Magician(P)> Inf.JTPROG(Z) @DQ |
|
l***u 发帖数: 1 | 19 1. Fourier Inversion Theorem
Th: If \int_{-\inf}^{\inf} \abs(g_{X}(t))dt <\inf, then X is absolutely
continuous with density
f(x)=1/(2*pi) \int_{\-inf}^{\inf} exp(-itx) g(t)dt
2. The Ch. f. of Laplace distribution, which has density is 1/2*exp(-\abs(x)),
is
g(t)=1/(1+t^2)
This is easy to verify by standard calculation.
3. So, by Fourier Inversion Theorem,
1/2*exp(-\abs(x))=1/(2*pi) \int_{\-inf}^{\inf} exp(-itx)*1/(1+t^2) dt
=1/(2*pi) \int_{\-inf}^{\inf} exp(itx)*1/(1+t^2) dt
|
|
m****n 发帖数: 45 | 20 t^2/(e^t-1) = e^(-t)t^2/(1-e^(-t))=e^(-t)t^2 sum_{k=0..inf}e^(-kt)=
=t^2 sum_{k=1..inf}e^(-kt)
so int_{t=0..inf}t^2/(e^t-1)=sum_{k=1..inf}int_{t=0..inf}t^2 e^(-kt)dt
let kt=x, then
int_{t=0..inf}t^2 e^(-kt)dt=1/k^3 int_{x=0..inf}x^2 e^(-x)dx=1/k^3 Gamma(3)
=2/k^3
hence
int_{t=0..inf}t^2/(e^t-1)=2*sum_{k=1..inf} 1/k^3 |
|
m*****n 发帖数: 74 | 21 来自主题: BrainTeaser版 - 一道数学题 it's nice, but sometimes there is a new V(U(V(U(A)))),
for example, A=[1,2)+(2,3]
U(A)=[1,3]
V(U(A))=(-inf,1)+(3,inf)
U(V(U(A)))=(-inf,1]+[3,inf)
V(U(V(U(A))))=(1,3)
also you have
V(A)=(-inf,1)+{2}+(3,inf)
U(V(A))=(-inf,1]+{2}+[3,inf)
V(U(V(A)))=(1,2)+(2,3)
8 in total, and indeed 10 if you take A=[1,2)+(2,3]+{4}
+ |
|
z*********e 发帖数: 10149 | 22 我的tomcat manager里面显示是附件这样的
一定要deploy .war文件吗?我看了其他几个application也没有.war文件。
文件目录
tomcat/webapps/springapp/下面有
META-INF folder
WEB-INF folder
hello.htm
hello.jsp
index.jsp
其中tomcat/webapps/springapp/META-INF folder下面有
MANIFEST.MF
tomcat/webapps/springapp/WEB-INF folder下面有
classes folder
lib folder
springapp-servlet.xml
web.xml
tomcat/webapps/springapp/WEB-INF/classes下面有
tomcat/webapps/springapp/WEB-INF/classes/springapp/web/HelloController.class
tomcat/webapps/springapp/WEB-INF/classes/springapp/we... 阅读全帖 |
|
I***O 发帖数: 188 | 23 D盘右键第一项是自动播放,双击不能打开.
是Autorun.inf这个文件的问题,就像有些光盘的自启动一样,你双击不一定能进入光盘,
而是弹出一个安装的画面,就是Autorun.inf这个文件里做了设置,很多病毒都会在这个
文件里动手脚,先杀病毒。其他盘正常吗?
确定无毒后打开我的电脑-工具-文件夹选项-文件类型, 找到“驱动器”,
点下方的“高级”-点选“编辑文件类型”里的“新建”-操作里填写“open”(这个
可随意填写)-用于执行操作的应用程序里填写explorer.exe-确定
应该能解决问题!
还有个方法:鼠标双击我的电脑-工具-文件夹选项-查看-显示所有文件和文件夹,
然后进入d盘,把autorun.inf删除,重启即可。
如果找不到autorun.inf,那么
1、开始-->运行-->cmd(打开命令提示符)
2、dir autorun.inf /a (没有参数a是看不到的,a是显示所有的意思),此时你会发
现一个autorun.inf文件
3、attrib autorun.inf -s -h -r 去掉autorun.inf文件的系统、只读、隐藏属性,否
则无法删除。 |
|
r*******y 发帖数: 1081 | 24 Still something confusing. Using residual theorem we can show that
\int_{-inf}^{inf} e^{ix} / x dx = i* PI so we can have
\int_{-inf}^{inf} sin(x) / x dx =PI
but we also can have
\int_{-inf}^{inf} cos(x) / x dx = 0 which is very confusing
since \int_{-inf}^{inf} cos(x) / x dx will diverge
Just look at \int_{0}^{1} cos(x) / x dx which is infinity |
|
c*m 发帖数: 1114 | 25 Assuming:
Event A: x1>a
Event B: x2>a
{x1,x2}是multivariate normal with correlation, with pdf:f(x1,x2)
要求的CDF:
p( A or B )= P(A)+P(B)-P(A and B)
P(A) 和 P(B) 可以用N(0,1)的CDF求出来。
P(A and B)=\int_{a}^{inf} \int_{a}^{inf} f(x1,x2) dx1 dx2
或者直接用f(x1,x2)来算。
P(A or B)=\int_{a}^{inf} \int_{-inf}^{a} f(x1,x2) dx1 dx2
+ \int_{-inf}^{a} \int_{a}^{inf} f(x1,x2) dx1 dx2
+ \int_{a}^{inf} \int_{a}^{inf} f(x1,x2) dx1 dx2
求pdf时可以用Leibniz rule for three dimensional case, 应该能够得到解析解。
see : http://en.wikipedia.org/... 阅读全帖 |
|
g*******y 发帖数: 1930 | 26 how about
note: inf = infinity
numbers: inf, inf/2,.....10,inf*0.75
all your L[0][] L[1][] will select the first number:"inf"
when it comes to the last number "inf*0.75":
which number to sacrifice?
obviously the answer should abandon the first "inf", but if you abandon the first number, your L[][] result will become useless
this is exactly what I meant in my post:
You don't define a "state"(i.e,subproblem) well enough if you don't specify whether the first number is chosen or not. |
|
t****t 发帖数: 6806 | 27 我用的symbolic toolbox,是和maple等价的吧?手里没有maple.
是这样的函数
第一个
p=(exp(-(x-m)^2)+exp(-(x+m)^2))/2;
q=p*log(p);
int(q, x, -inf, inf)
这个就搞不定了. m是正实参量,换成数字也不行.
这一步其实我已经用matlab算过了,但是m比较大时matlab会把有效的部分漏掉.
第二个
p=(exp(-(x-m)^2-y^2)+exp(-(x+m)^2-y^2)+exp(-x^2+(y-m)^2)+exp(-x^2+(y+m)^2))/
4;
q=p*log(p);
int(int(q, x, -inf, inf), y, -inf, inf)
matlab一下子就说maple溢出了. |
|
J*******g 发帖数: 267 | 28 here is my solution, not sure if it is correct
d(Bu*exp(-ut))
= (dBu)*exp(-ut) + Bu*d(exp(-ut)) + dBu*d(exp(-ut))
= exp(-ut)*dBu - t*exp(-ut)*Bu*du
therefore,
\int_0^\inf d(Bu*exp(-ut)) = \int_0^\inf exp(-ut) dBu - t*\int_0^\inf exp(-
ut)*Bu du
=> X_t = \int_0^\inf exp(-ut)*Bu du = (1/t)*\int_0^\inf exp(-ut) dBu
Hence,
E[X_t] = 0
E[X_t*X_s] = (1/t*s)*\int_0^\inf exp(-u(t+s)) du = 1/(t*s*(t+s)) |
|
r*******y 发帖数: 1081 | 29 a little more clear now.
using residual theorem, we can show \int_{-inf}^{-\epsilon} sin(x) / x dx
+ \int_{\epsilon}{inf} sin(x) / x dx (in the sense of v.p.)---> PI as
\epsilon ---> 0. But we also can show \int_{-inf}{inf} sin(x) / x dx
converge
so we can claim \int_{-inf}{inf} sin(x) / x dx = PI. |
|
s*****n 发帖数: 2174 | 30 这个按照定义写出来, 很容易证明.
E(X|X>-k) = \int_{x=-k}^{+Inf}(x * pdf(x|x>-k))dx
=\int_{x=-k}^{+Inf}(x * pdf(x))dx / \int_{x=-k}^{+Inf}(pdf(x))dx
=\int_{x=-k}^{+Inf}(x * pdf(x))dx / \int_{x=-Inf}^{k}(pdf(x))dx
=\int_{x=-k}^{+Inf}(x * pdf(x))dx / F(k)
下面把第一项做一个简单的变量替换 y = x^2
易得第一项就是f(k) |
|
d****d 发帖数: 2919 | 31 我中学数学里学的log是把(0 +inf)一一对应到(-inf, +inf),
你说的可以把 (0,1)一 一对应到(-inf, +inf)是对的,
但觉得log这例子不对。。。 |
|
i******c 发帖数: 9350 | 32 找到了这个, 凑合试试:
Detailed Solution Steps
1. Go to C:\Windows\System32\DriverStore\FileRepository
Find files named usbstore…
2. Choose the most recent usbstor folder
Open the folder
You’ll see a number of files…
3. Select and copy usbstor.inf
4. Now go to C:\Windows\Inf
Paste the usbstor.inf file into this folder (On my system there was no
usbstor.inf file already in this folder).
5. Now try again to install the reader
When Vista reports it cannot find a driver direct it to C:\Windows\Inf
If all goes... 阅读全帖 |
|
m**c 发帖数: 90 | 33
Sorry I didn't make myself very clear in my post: You need to have a file
called "web.xml" in "${TOMCAT_HOME}/webapps/myapp/WEB-INF". You can copy
"web.xml" from "${TOMCAT_HOME}/webapps/examples/WEB-INF" to
"${TOMCAT_HOME}/webapps/myapp/WEB-INF", then modify it a little bit. Why
don't post the "web.xml" from "${TOMCAT_HOME}/webapps/examples/WEB-INF" to
this board so that I can take a look?
place
it?
examples.
面
文件放在examples目录下(D:\jakarta-tomcat-4.1.24\webapps\examples\WEB-INF\class |
|
R*****l 发帖数: 310 | 34 最近频繁遇到。不知道是否算“落雪”的变种,但比落雪凶悍多了。
症状:所有逻辑分区根目录下全部被加上autorun.inf和pagefile.pif
两个文件。在system32/com/下面多了两个木马程序lsass.exe和smss.exe,
一直后台运行,无法杀掉。在startup里面加了一个~MSDOS启动程序。
机器运行爆慢,所有exe程序被锁住,运行任何exe都将启动木马进程,
双击盘符也将启动木马进程,防火墙和杀毒软件均被屏蔽。
解决方法: 重起到安全模式下,
start->run->cmd打开command prompt窗口,
删掉每个分区下面的autorun.inf和pagefile.pif。
attrib -r -s -h autorun.inf
attrib -r -s -h pagefile.pif
del -f autorun.inf
del -f pagefile.inf
去system32/com/下面删了lsass.exe和smss.exe,去startup下面删除~MSDOS。
重起后,search最近被修改过的文件。发现所有分区的很多html/htm文 |
|
z***e 发帖数: 1757 | 35 【 以下文字转载自 PKU 讨论区 】
【 原文由 zhyue 所发表 】
3.since every element of K is isolated, then for each element k of K
we can find r(k) s.t. intersection of open ball Br(k) and K is {k}.
Then all these open balls forms an open cover of K. suppose K is not
finite, then this open cover has no finite subcover, contradict with
K is compact. QED
2.a) is easy.
b) suppose inf(p*B1)+...+inf(p*Bm) is not inf(p*B), i.e there exists
b of B, s.t. p*b
p*b=p*b1+...+p*bm, a |
|
j*****y 发帖数: 1071 | 36 先化成
\sum_{n = 0}^{\inf} (n + 1) x^(2n) (1/(n + 1)!) (1/(1-x^2))^(n + 1)
+ \sum_{n = 0}^{\inf} (n + 1) x^(2n) \sum_{k = 0}^{n} (1/k!) (1/(1-x^2))^
k
第一个 sum 是 1/(1-x^2) e^(x^2/(1- x^2))
下面算第二个,需要交换 n 和 k的指标,交换以后就变成
\sum_{k = 0}^{\inf} (1/k!) (1/(1-x^2))^k \sum_{n = k}^{\inf} (n + 1) x^(2n)
其中 \sum_{n = k}^{\inf} (n + 1) x^(2n) 可以看成是 (x^2)^(k + 1) +
(x^2)^(k + 2) + ... 的导数,也就是 (x^2)^(k + 1)/ (1 - x^2) 的导数
剩下的你自己弄吧 |
|
c****o 发帖数: 1280 | 37 yes, p<1. Think about the following martingale, let s_0>1 be such that
s_0+s_0^-2=2, then E(s_0^x)=1 ans s_0^(S_n) is a martingale, let tau be the
stopping time, then E(s_0^(S_min(n,tau))=1,(by OST) take the limit, we have
p(tau
-inf) then solve p(tau
is why we can take the limit inside, and also why p<1.
prob |
|
s********1 发帖数: 54 | 38 Compare data inf123 and data inf, who can explain to me why outputs from
these two codes are not the same? Data inf is the right one. In terms of the
difference, they are represented as follows:
In data inf123
length readd $10.;/*先把長度都開好*/
input group id_n name $ job $12. st_time $10.;
In data inf
length job $12 readd $10 st_time $10;/*先把長度都開好*/
input group id_n name $ job $ st_time $;
______________________________________________
data inf123 ;
length readd $10.;/*先把長度都開好*/
... 阅读全帖 |
|
g*******y 发帖数: 1930 | 39 have you tried my test case?
inf, inf*0.5, ...(normal numbers),inf*0.7
you can use 9999999 as inf.
237 |
|
s***h 发帖数: 662 | 40 来自主题: JobHunting版 - 一个dp题 有一条线段上有点1, 2, 3, ...n, 均匀分布, 每两个之间间隔为1,从一点出发
到另一点需要花费时间t = 1.
假定你从某点s出发, 需要遍历所有的点. 每点i都有一个时间ti与之相关,必须在
ti之前到达点i.
如何在最短时间里面遍历所有的点, 并且满足ti的限制? 提示(可以使用S(i,j,x)表示遍历点i->j, 满足ti,且终止在点x(x = i or j).)
这个好像可以写成
OPT = min(S(1,n,1), S(1,n,n))
S(1,n,n) = min{((S(1,i,i) + (n-i)) > tn) ? inf : keep the value, or
((S(1,i,1) + n) > tn) ? inf: keep the value}
...
但是从1出发就trivial了.出发点不是1, 这个解没有用. 如果i!=1, S(i,j,x)怎么表示呢? 如果写成
S(i,j,j) = min{((S(i,k,k) + (j-k)) > tj) ? inf : keep the value, or
... 阅读全帖 |
|
w**z 发帖数: 8232 | 41 http://stackoverflow.com/questions/825221/where-can-i-find-the-
It's a math problem more than CS. Well, don't know how far you want to go..
That is the implementation from math lib
public static double sqrt(double a) {
return StrictMath.sqrt(a); // default impl. delegates to StrictMath
// Note that hardware sqrt instructions
// frequently can be directly used by JITs
// and should be much faster than doing
// Math.... 阅读全帖 |
|
m******s 发帖数: 165 | 42 好久不刷题了,写一个玩玩,估计多半哪儿有小错。。。
int getmin(string str, vector dic)
{
int n = str.size();
const int INF = n + 1;
vector dp(n + 1, INF);
dp[0] = 0;
for (int i = 0; i < n; ++i)
if (dp[i] < INF)
for (int j = 0; j < dic.size(); ++j)
if (str.substr(i, dic[j].size()) == dic[j])
dp[i + dic[j].size()] = min(dp[i + dic[j].size()], dp[i] + 1);
return dp[n] < INF ? dp[n] : -1;
} |
|
t*******r 发帖数: 22634 | 43 俺开完会就没力气开火了。。。
其实从形式主义和 cosmology 的角度看,连牛顿莱布尼兹的极限概念,
都不是最基本的。
极限最最基本的概念,说白了,就是无限表达式的结果要满足两个条件:
(1)能够自圆其说,并且不自相矛盾。(形式主义的要求)。
(2)能够有用。(cosmology 的要求,best-fit modelling)。
对于(1)的形式主义的要求:
比如前面人提到的 (1 - 1 + 1 - 1 + 1 ... ) 。。。这个导致“发散”
的意思,其实是结果不唯一,所以自相矛盾。
其实这个 (1 - 1 + 1 - 1 + 1 ...) 也不是没结果的,也就是
死板主义数学家不愿意给个结果而已。。。形式主义者完全
可以给这玩意儿也有个结果,这个结果就是 NaN (Not-a-Number)。
(indeterminable 就是 NaN 里的一种情况)。
其实 (2 + 2^2 + 2^3 + 2^4 + ...) 也可以有个结果,这个结果
就是 Inf(或者 Positive-Inf)。
那么 (Inf - Inf) 的结果是啥?形式主义者说,是 NaN。。。
这也能自... 阅读全帖 |
|
d*********2 发帖数: 48111 | 44 wy never really think about this:
in theory or in real life, mobilized inf is for sure faster than infantry.
There is no doubt for that.
Then when you consider using the mobilized inf or infantry during the war,
their
speed is decided by their speed of supply.
If you have unlimited supplies instantly(like everything with local supply
by robbery or government supply), mobilized inf is faster. Otherwise
mobilized inf and
infantry's speed are solely decided by the speed of your supply. In the old
t |
|
h***i 发帖数: 3844 | 45 不知道对否
1. by law of large number, empirical risk will converge to theoretic risk
this is why we like to minimize empirical risk. called ERM
2. but inf of empirical risk will not always converge to inf of theoretic
risk. so
1 is not good enough.
3. under some constrain for example, uniform converge condition, inf of
empirical risk will converge to inf of theoretic risk.
4. Glivenko-Cantelli theorem.
5. VC theorem(a generalization of GC theorem), the famous inequality,
theoretic risk <=empirical ri |
|
l*********r 发帖数: 80 | 46 【 以下文字转载自 JobHunting 讨论区,原文如下 】
发信人: littlelover (大肚子情剩), 信区: JobHunting
标 题: Urgent Help needed about Java Servlet?
发信站: Unknown Space - 未名空间 (Sat Oct 25 17:13:21 2003) WWW-POST
UserAdmin.java is in C:\jwsdp-1.3\webapps\HW3example\WEB-INF\classes
hw1_lib.jar is in C:\jwsdp-1.3\webapps\HW3example\WEB-INF\lib
我是这样compile的:
C:\jwsdp-1.3\webapps\HW3example\WEB-INF\classes>javac -classpath
C:\jwsdp-1.3\we
bapps\HW3example\WEB-INF\lib\hw1_lib.jar;c:\jwsdp-1.3\common\lib\servlet-api.j
ar
UserAdmin.ja |
|
s***8 发帖数: 1136 | 47 Which statement(s) are true with regard to web applications in JavaEE 6?
A, A WAR file must contain a web.xml under WEB-INF/
B, A WAR file must contain a web.xml under META-INF/
C, EJB bean classes may be packaged under WEB-INF/classes
D, One or more ejb-jar files may be packaged under WEB-INF/lib
E, A servlet can be mapped to 0, 1, or more url-patterns |
|
q*********u 发帖数: 280 | 48 a c d?
Which statement(s) are true with regard to web applications in JavaEE 6?
A, A WAR file must contain a web.xml under WEB-INF/
B, A WAR file must contain a web.xml under META-INF/
C, EJB bean classes may be packaged under WEB-INF/classes
D, One or more ejb-jar files may be packaged under WEB-INF/lib
E, A servlet can be mapped to 0, 1, or more url-patterns |
|
m*****f 发帖数: 2 | 49 Here are my problems, any advice on how to kill them?
C:\WINNT\inf\new.exe=>(Instyler)=>%appfolder%\crack.exe=>(Instyler)=>%appfolde
r%\sysconfig.ocx Infected IRC-Worm.Randon.I
C:\WINNT\inf\new.exe=>(Instyler)=>%appfolder%\crack.exe=>(Instyler)=>%appfolde
r%\sysconfig.ocx Disinfection failed - Trying second action
C:\WINNT\inf\new.exe=>(Instyler)=>%appfolder%\crack.exe=>(Instyler)=>%appfolde
r%\sysconfig.ocx Move failed
C:\WINNT\inf\new.exe=>(Instyler)=>%appfolder%\crack.exe=>(Instyler)=>%appfol |
|
y*********r 发帖数: 587 | 50 Three steps to remove Messeng v4.7:
1. open a command window, type
"rundll32 advpack.dll,LaunchINFSection %systemRoot%\INF\msmsgs.inf,BLC.Remove"
2. type
"rundll32 setupapi,InstallHinfSection BLC.Remove 128
%systemRoot%\INF\msmsgs.inf"
3. reboot the computer
folders, |
|