l*****0
发帖数: 118 | 1 We currently have 2 Ph.D. funded positions available in our research group.
Our work is mostly focused on secure and advanced memory systems and
architectures. The new Ph.D. students will mostly focus on investigating the
design of secure architectures that employ low-overhead encryption and data
integrity verification mechanisms. Our research group helps reshaping the
design of future of computing systems in the presence of emerging emerging
memory technologies and new security requirements. Fo... 阅读全帖 |
|
r*******g
发帖数: 1707 | 2 【 以下文字转载自 Beijing 讨论区 】
发信人: reshaping (宁鸣而死 不默而生～学习宽恕与遗忘), 信区: Beijing
标 题: 问个更初级的endnote问题
发信站: BBS 未名空间站 (Thu Jan 15 11:54:13 2009), 转信
正文里的reference, 为啥显示的不一样？
有的是 (smith, 2009)
有的是 (j. a. smith, 2009)
有的是 (jack ass smith, 2009)
一篇文章里什么样的都有，很乱
可是看了半天也不知道怎么改 |
|
g****y
发帖数: 75 | 3 by Gregory Clark
Dismal scientists: how the crash is reshaping economics
With the chattering classes consumed by concern for the devastated value of
their 401K funds, and their suddenly precarious lifestyles, there has been
much anger and scorn directed at those former masters of the universe,
financiers.
But the shock to the world of finance has been echoed by a shock to the
world of academic economics that is just as profound.
In the long post WWII boom, as free market ideology triumphed, econ |
|
z********r
发帖数: 313 | 4 做通信系统仿真
fm,fm模块都要求输入的数据是scalar,但是我前面的模块的输出是1*7的
试了 reshape,buffer模块都不能把1*7的vector转化成scalar
不知道大家有什么建议？
谢谢！ |
|
w*******p
发帖数: 253 | 5 Usually the smaller step, the better accuracy, but more time-consuming.
Rsoft doesn't support localized mesh reshape, which means you can make the
meshes smaller in some critical region. So you have to live with that.
This is one of the software calculation tolerances I mentioned to you in
previous message, if the difference is within a certain range (for example,
diff steps give 0.1% difference in Neff), you can choose the bigger step to
save time.
Think about this, the refractive index you use |
|
l*****0
发帖数: 118 | 6 We currently have 2 Ph.D. funded positions available in our research group.
Our work is mostly focused on secure and advanced memory systems and
architectures. The new Ph.D. students will mostly focus on investigating the
design of secure architectures that employ low-overhead encryption and data
integrity verification mechanisms. Our research group helps reshaping the
design of future of computing systems in the presence of emerging emerging
memory technologies and new security requirements. Fo... 阅读全帖 |
|
a*****a
发帖数: 1038 | 7 来自主题: Environmental版 - 海龟机会 Points taken. Again, no offense in my reply below.
There is a big change for AECOM in the near future. AECOM just bought Earth
Tech a couple of weeks ago. Right now a dozen of subcompanies under the
AECOM, such as M&E, TCB, Mansseull, ENSR, EDAW, and earth tech will be
restructured and reorganized into a big company. There will be a lot of
changes and uncertainty for this company in the United States. It is not
sure how AECOM reshapes its offices worldwide. |
|
Q***5
发帖数: 994 | 8 koox is right, you have to prepare a matrix first.
If your `f' function can only take two variables, you can do the following
to come up with that matrix:
[x,y] = meshgrid(-2:0.1:2,-2:0.1:2);
x1 = x(:);
y1 = y(:);
for k = 1:length(x1)
z1(k,1) = f(x1(k), y1(k));
end
z = reshape(z1, size(x));
surf(x,y,z) |
|
T******a
发帖数: 195 | 9 难是肯定的，不光是当医生。看看各行各业跟10年前相比，那个不是难上加难。全球的
经济如果都能会到克林顿时那个速度，顾及哪个行业都好很多。所以这些咱都管不了。
其实我们大多数img最能拼的时智商，就是考个高分。而“情商”就是 socialized
connection会差很多，尤其时cmg。不是能力问题，而是语言，文化等等问题。看看那
些能拿很多iv的relative old cmg 来说，多是情商高手，也就是“混”在江湖的人。
跟人混了，人家才拿你当朋友，才会有好的lor。当然说了容易，做了难，归根到底还
是personality问题。所以如果能考个relative ok 的分数，“混”将是决定因素。从
这个角度来说，如果你至少愿意去“混”，在这个当医生的长征路上才会更好的"
reshape youself". 高智商（分数）的“混子”（connection）估计会有更多的机会。
既然不是每个人都能最中买吃，就想开点，尽力就好。这辈子不行，下辈子早点“觉悟
“。 |
|
d****y
发帖数: 2180 | 10 ☆─────────────────────────────────────☆
LittleThree (xiao san) 于 (Thu Sep 20 10:40:48 2012, 美东) 提到:
１）各个PGM里CMG的数量在下降。很多PGM，PGY３里还有CMG，２里就很少，１里更是
没有了。这样的PGM不是少数。趋势就是众多PGM越来越不愿意招收CMG。比如，德州某
医院过去每年招收CMG，从去年开始，因为某些原因拒绝所有CMG。与此成对比的是IMG
的PGM甚至得到了强化。很多老印PGM过去还招点其它国家毕业生点缀，现在已经是１０
０％老印了。
２）CMG能选择的专业越来越窄。早年还能有CMG进入较好的专业，比如麻醉等。现在IM
基本就是最好选择了。当然，这个也许对所有非AMG都一样。
３）关系的作用越来越大。关系的重要性大过一切。如果你的老板是有权势的MD，且和
你关系良好；那么你几乎成功大半了。分数，背景，绿卡都不那么重要了。查某些好的
PGM，清一色AMG之中居然有个CMG，前无古人，后无来者。不敢说１００％，９０％的
可能性都是关系。
４）中国人在各PGM里... 阅读全帖 |
|
J**********i
发帖数: 925 | 11 Good old days gone for Biotech
http://www.signonsandiego.com/news/business/biotech/20061126-9999-1n26biotech.html
The business of biotechnology is undergoing a seismic shift, and for some
companies there's no surviving the jolt. Take Discovery Partners
International. The San Diego company ceased to exist in September, a victim
of trends that are reshaping the biotech industry and, at least in the short
term, threatening innovation and job creation in the nation's third-largest
biotech cluster.
I |
|
f*******e
发帖数: 689 | 12 Are you a pharmacist or a faculty now? If neither of them, I think you getta
make a choice-being a pharmacist or a faculty. Both of them are very tough
jobs when you are on early years of your career. For me, I spent a lot of
years in school being a student on science. However, as a pharmacist intern,
it was very challenging for me. I gotta learn some very basic things and
even started to learn correctly pronunciation with my ENGLISH tutor. My
knowledge has been reshaped for being a pharmacist. |
|
J*******3
发帖数: 1651 | 13 One Hundred Years of Quantum Physics
Daniel Kleppner and Roman Jackiw*
An informed list of the most profound scientific developments of the 20th
century is likely to include general relativity, quantum mechanics, big bang
cosmology, the unraveling of the genetic code, evolutionary biology, and
perhaps a few other topics of the reader's choice. Among these, quantum
mechanics is unique because of its profoundly radical quality. Quantum
mechanics forced physicists to reshape their ideas of reality, |
|
A**********n
发帖数: 824 | 14 下面是网传杨振宁获1994年鲍尔奖的官方表扬词。哪位知道原文出处？杨振宁获鲍尔奖
的官方信息的链接：https://www.fi.edu/laureates/chen-ning-yang
For the formulation of a general field theory which synthesizes the physical
laws of nature and provides us with an understanding of the fundamental
forces of the universe. As one of the conceptual masterpieces of the
twentieth century explaining the interaction of subatomic particles, his
theory has profoundly reshaped the development of physics and modern
geometry during the last forty years. This the... 阅读全帖 |
|
q*******t
发帖数: 17 | 15 http://www.quantspot.com/jobs/list/18162-Shanghai-Quantitative-Finance-Program
Morgan Stanley Innovative Data, Environments, Analytics & Systems (IDEAS),
is an integrated quantitative and technology organization formed to create a
sustainable, commercial advantage for Morgan Stanley by reshaping the firm
’s businesses around innovative people, processes and systems. IDEAS
includes revenue-generating, business unit-embedded desk strategist teams
and platform and technology teams with a broad rang |
|
f*******d
发帖数: 339 | 16 This article did not mention Wang's experiment. However, it
seems that
these experiments are similar. The superluminal velocity is
only apparant,
it is produced by sending and receiving particularly shaped
"signals". These
signals, although apparantly maintaining their shape,
actually were "reshaped"
by absorption, interference, diffraction, or other effects
en route, thus a naive
reading of the measurement yiels superluminal velocity.
I think these velocities are kind of "generalized phase
velo |
|
D****n
发帖数: 21 | 17 One Hundred Years of Quantum Physics
Daniel Kleppner and Roman Jackiw*
An informed list of the most profound scientific developments of the 20th ce
ntury is likely to include general relativity, quantum mechanics, big bang c
osmology, the unraveling of the genetic code, evolutionary biology, and perh
aps a few other topics of the reader's choice. Among these, quantum mechanic
s is unique because of its profoundly radical quality. Quantum mechanics for
ced physicists to reshape their ideas of rea |
|
q**j
发帖数: 10612 | 18 同一个事情先用SAS干了，然后用R干了。有些感想，有些问题。跑来跟大家交流一下啊。
1. R里面改个变量名这么这么难？
需要一个reshape pacakge
data.frame = rename(data.frame, c(oldname = "newname"))
不用这个package就更麻烦了。
2. 控制时间，日期怎么这么难？
以前“告别棒球场”说了，最好不要用日期，全部用字符。临到用的时候才换。就算这
样还是问题多多。比如 “1990-12-04”，怎么样轻松的把它变成 “1990-12-01”（月
初），“1990-12-31”（月末）。往前，往后移动几个月？我写了一个function来干这
个。但是非常慢（yes, I know it.)而且不能往以前移动。我在SAS里面经常用intnx函
数在proc sql的时候来specify where condition。这个用R实在是非常，非常不方便。
# t=intnx("1 mon","2007-12-15",2,'e')
# DateVector = c("2006-02-03","2004-05-06")
# Ne |
|
b********y
发帖数: 63 | 19 Not sure how SAS is more convenient on those issues, but they should not be a
problem for R at all.
同一个事情先用SAS干了，然后用R干了。有些感想，有些问题。跑来跟大家交流一下啊。
1. R里面改个变量名这么这么难？
需要一个reshape pacakge
data.frame = rename(data.frame, c(oldname = "newname"))
不用这个package就更麻烦了。
you can just change the name: colnames(x)[1] = newName1.
2. 控制时间，日期怎么这么难？
以前“告别棒球场”说了，最好不要用日期，全部用字符。临到用的时候才换。就算这
样还是问题多多。比如 “1990-12-04”，怎么样轻松的把它变成 “1990-12-01”（月
初），“1990-12-31”（月末）。往前，往后移动几个月？我写了一个function来干这
个。但是非常慢（yes, I know it.)而且不能往以前移动。我 |
|
c**d
发帖数: 104 | 20 这个属于reshape data
1）PROC transpose
2) using array in data step. |
|
d*******1
发帖数: 854 | 21 有一个dataframe, 几百个column, 准备用RESHAPE TRANSPOSE一下(只有头两个
VARIABLE 还保留在结果的TABLE里, 剩下的column都要压缩成一条COLUMN, 但是好象要
用VARYING=来规定所要压缩的COLUMN位置, 怎么用R先把要压缩的VARIABLE的启示和终
止位置抓到呢? |
|
a********s
发帖数: 188 | 22 library(reshape)
x <- data.frame(id=1:2, var1 = 1:2, var2=3:4, var3=5:6)
melt(x, id=c("id"))
这样应该可以符合你例子的要求。 |
|
D******n
发帖数: 2836 | 23
reshape(data,varying=names(data)[-1],direction="long",timevar="vname",v.name
s="value",times=names(data)[-1]) |
|
d*******1
发帖数: 854 | 24 谢谢了, 这个RESHAPE的帮助看得我头大,你的例子就容易懂多了. 不过这么好的一个
FUNCTION的NAMING CONVENTION可太掉价了, 什么啊TIME TIMEVAR的, 完全不只所云啊.
name |
|
|
|
a****m
发帖数: 693 | 27 谢谢，主要是行数太多，前62个数（row wise）是一组，如果matrix 置换会把所有的
行弄混，结果txt转换成cvs时候，把空格给变成0了，用genelowvalfilter 好像也不能
去掉，只好这样处理了：
谢谢
[n,p]=size(data);
for i=1:2000
gene(i,:)=reshape(data(((i-1)*16+1):16*i,1:4),1,64);
end
data1=gene(:,1:47);
data2=gene(:,49:63);
data=[data1,data2];
这个filter 不起作用。
[mask,genes] = genelowvalfilter(gene,'absval',1); |
|
d*******1
发帖数: 854 | 28 看到你的答案我也FT了， HEHE. 那么names(data)[1]到底给出了什么呢？ 我这个IDEA
是从你的那个RESHAPE(VARYING=NAMES(DATA)[2].....)来的， 呵呵 |
|
d*******1
发帖数: 854 | 29 我有个data如下：
gene patient1 patient2 patient3.............
_________________________________________________
gene1 10 20 50
gene2 20 40 60
gene3 100 200 300
................
我要做的是取gene1的数据和其他所有的基因做pearson regression. 我的第一个方案
是搞一个运行50000次的explicit loop, 每次取一行gene数据和gene1的数据行做
regression, 这个太慢了（耗时将近1小时）。
我的另一个想法是用reshape把上面这个矮胖的数据转成下面的瘦长样子：
gene patientID value gene1 |
|
A*******s
发帖数: 3942 | 30 google proc transpose, reshape data, long to wide, wide to long |
|
hs
发帖数: 1549 | 31 reshape(randperm(5*5),5,5)
然后把最小的n个变成-1，最大的m个变成+1，其它变成0 |
|
s****y
发帖数: 297 | 32 一数据，一列是各values， 另一列是这些values的frequency， 怎样才能对这些
values求quantile呢？
比如，
values Frequency
0.2 4
0.3 0
0.4 8
。 。
。 。
。 。
第二列表示数据中有4 个0.2， 0个0.3， 8个0.4 ，等等。。。
我需要求这些数的quantile，R里面有什么简便的方法吗？
谢谢了！！！
另外，这些技巧什么的可以在哪里学到呢？有没有高手给推荐本书啥的？
有的时候不知道技巧的关键字，搜索的结果都不是很理想。。。 |
|
d***n
发帖数: 43 | 33 a<-c(0.2, 0.3, 0.4)
b<-c(4,0,8)
c<-c()
for (i in 1:3)
{
d<-rep(a[i], b[i])
c<-c (d, c)
}
Then sort the c vector and calculate the quantile. |
|
a***d
发帖数: 336 | 34 hehe, 直接 c <- rep(a, b) and no sorting needed. |
|
k*******a
发帖数: 772 | 35 不好意思，那个好像也不行，不过你可以先 reshape data比如
data a;
do i=1 to 8;
x=1; y=1;output;
end;
run;
data b(keep=var z);
set a(rename=(x=z) in=inx) a(rename=(y=z) in=iny);
if inx then var='x';
else if iny then var='y';
run;
proc means data=b noprint;
by var;
var z;
output out=c(drop=_freq_ _type_) n=n nmiss=nmiss min=min max=max;
run; |
|
j******4
发帖数: 6090 | 36 variables: Age(in month) Female(dummy), Smoke(dummy)
outcome(counts): time1 time2 time3 time4
Reshaped into long form then the data looks like this:
ID time Y Age Female Smoke
1 1 0 30 1 1
1 2 1 30 1 1
1 3 5 30 1 1
1 4 2 30 1 1
...
可以用GEE或者mixed model，但是据我观察，在 time=1 的时候，会有很多 Y = 0，我
在想 Zero-Inflated-Poisson(ZIP) 是否可以用来分析这种repeated measures的data，要
如何用呢？
多谢各位~ |
|
s******1
发帖数: 39 | 37 试试EXCEL PIVOT-TABLE?用STATA的话RESHAPE命令也可以搞定了。 |
|
|
|
c***z
发帖数: 6348 | 40 【 以下文字转载自 DataSciences 讨论区 】
发信人: chaoz (面朝大海，吃碗凉皮), 信区: DataSciences
标 题: 最近的一些面经
发信站: BBS 未名空间站 (Tue Jun 10 01:52:38 2014, 美东)
首先说说教训。如果你是技术骨干，那么在介绍项目的时候，得悠着点，注意不能说涉
及保密的东西。这个对于有些人来说是常识，但是对于比较呆的人（比如我），却是需
要栽几个跟头才能学到的。
然后是面经。大概有这么几类的题目：
1，理论类。Netflix爱问，追着我问Random Forest怎么split，怎么cross validate，
还有Neural Network怎么back propagate, 怎么feature selection，用的什么包裹，
什么原理，一直问到survival analysis的公式，很久没用于是歇菜了（和treatment
model混淆了）。对不起内推的大哥了。
2，实践类。小公司爱问，类似于码工面试的设计题。这种情况怎么搞，那种情况怎么
搞，没见到data只能胡说，就说以前我有一个项目，情况有点... 阅读全帖 |
|
k*******a
发帖数: 772 | 41 try reshape function
or reshape2 R package |
|
w*****1
发帖数: 473 | 42 I tried reshape and it works! Thank you very much! |
|
w*******n
发帖数: 469 | 43 need R code, something like cast, melt, reshape ... |
|
c**r
发帖数: 2019 | 44 从2003年6月开始作版副，到现在已经两年零四个月，共计整理精华区五次，封ID四
个。。。推荐现任版副reshaping出任新版主。
谢谢广大网友的厚爱，
chrr |
|
r*******g
发帖数: 1707 | 45 申请人：reshaping
版面：我爱北京天安门
理由：灌水很久，应版主蓝猫邀请，协助版务工作
日期：2/28/2008 |
|
S**********8
发帖数: 30 | 46 Thanks a lot!
Happy New Year~
, centaur, clavichord, coolhunter, DabaoSODBoy, dcw, Discovery, dongchengqu,
Drake, drew, enantiomers, FRJJ2009, geon, guaiguaicong, haiduc, happydoudou
, icbchsbc, isisl414, jjneimei, Kakaa, kicolly, largewolf, LingLingCool,
lingyunxie, Logan601, LuckygirlIam, makejian, ManCity, maoyi, Maxstudio,
Mercedes, meso, Mobot, mochamocha, nash1031, normanwu, oriclechao, paotangde
, pyruvate, Reduce, reshaping, scoreboard, shopper2: , steve, superbao,
tianyumi, Tody, Tomos |
|
b****p
发帖数: 239 | 47 最近读到一篇关于ZocDoc的文章，感受很深．省时间的可以看看我自己的总结如下，原
文链接（http://mashable.com/2013/01/08/zocdoc/）读起来感觉更好．
分享出来，仅以抛砖引玉，望多指点．
业务模式总结：
其聪明之处把握住了，急性疾病但没有生命危险，病人的痛苦程度取决于ER和预约医生
的延迟，大城市非常突出．所以起点是NewYork的牙医，利用医生的空余slot来满足突
发病人的需求．
不知道更深一步做此结论是否正确：Zocdoc更是给年轻医生一个机会和一个新的渠道，
来更快的建立起自己的病人群体．而已经做得很好的医生因为病人群体大，空余时间少
，加盟Zocdoc的经济意义不大．
公司发展的概要：
A short journey of ZocDoc：
1. A painful experience from a well-trained consultant
In 2007 a McKinsey consultant, Massoumi, had a sinus infection. As the plane
descended, he f... 阅读全帖 |
|
c***z
发帖数: 6348 | 48 首先说说教训。如果你是技术骨干，那么在介绍项目的时候，得悠着点，注意不能说涉
及保密的东西。这个对于有些人来说是常识，但是对于比较呆的人（比如我），却是需
要栽几个跟头才能学到的。
然后是面经。大概有这么几类的题目：
1，理论类。Netflix爱问，追着我问Random Forest怎么split，怎么cross validate，
还有Neural Network怎么back propagate, 怎么feature selection，用的什么包裹，
什么原理，一直问到survival analysis的公式，很久没用于是歇菜了（和treatment
model混淆了）。对不起内推的大哥了。
2，实践类。小公司爱问，类似于码工面试的设计题。这种情况怎么搞，那种情况怎么
搞，没见到data只能胡说，就说以前我有一个项目，情况有点类似，我大体是这么做的
，但是细节不能说 ：P
3，基础操作类。大公司爱问，基本上SQL解决。我最近有点轻敌，没有重视SQL，觉得
这东西老古董。结果被人问了一个SQL很容易而R，Pig和Scala都很难解决的数据操作的
问题。还有Linkedin的三姐让我1... 阅读全帖 |
|
l******n
发帖数: 9344 | 49 来自主题: DataSciences版 - R问题请教 添0把没有的元素补齐，然后把count reshape就行了。 |
|
w*******n
发帖数: 469 | 50 Thanks.
I would like use some function in R like reshape, melt or cast/decast,
supposing I have quite a lot of columns like this.
Or is there any way to do the rbind but ignoring the name difference? |
|
