由买买提看人间百态

topics

全部话题 - 话题: linearized
首页 上页 1 2 3 4 5 6 7 8 9 10 下页 末页 (共10页)
R********n
发帖数: 5904
1
这段历史大概被彻底否毛及冷战给掩盖了。从一定程度说,夸克是从中国这窃取的idea?
http://www.nytimes.com/2006/12/05/science/05china.html?_r=1
China Pursues Major Role in Particle Physics
The superconducting magnet for the Large Hadron Collider near Geneva.
This was no idle diversion. According to natural dialectics, which formed
the philosophical underpinnings of Marxism, the entire universe, from top to
bottom, was seething with tension and change. As a result, Mao thought,
nature should be infinitely divisible.
“Take a footlong stic... 阅读全帖
S******8
发帖数: 24594
2
老方到美国后发的文章列表,你自己看吧:
1990
174. Biased clustering in a universe with hot dark matter and a cosmic
string, L.Z.Fang, S.P.Xiang and L.Yan, Astr. & Astrophys. 233, 1.
175. Periodicity of redshift distribution in a T-3 universe, L.Z.Fang, Astr.
& Astrophys. 239, 24.
176. An upper limit to quasar's peculiar velocity, L.Z.Fang, Y.Q.Chu, X.F.
Zhu and L.F. Wang, Inter. J. Mod. Phys., 5, 2993.
177. Quasar pair and quasar's peculiar velocity, L.Z.Fang, Developments in
general relativity, astrophysics and qu... 阅读全帖

发帖数: 1
3
来自主题: Military版 - 大家知道ee不光数学学的挺多
EE参考书目---数学类
LunarModule
来自: LunarModule (上海) 2012-10-06创建 2012-10-06更新
如何找好书:
1.去MIT OpenCourseWare(http://ocw.mit.edu/courses/),找相关课程的课本/参考书(Textbook/Referencebook)
2.看amazon.com上的评价 & 所列出的”同时购买”书单
3.好书后面的参考资料 & 推荐阅读
4.牛人写的书
5.豆瓣上的书评&豆列
6.网上热心人写的推荐书目
===================================================================
[数学类]
一、微积分/数学分析
1.《托马斯微积分》
2.《微积分和数学分析引论》 柯朗
3.《微积分学教程》 菲赫金哥尔茨
4.《数学分析习题集》 吉米多维奇
5.《吉米多维奇 数学分析习题集 学习指引》 谢惠民
6.《数学分析新讲》 张筑生
二、线性代数
7.《Introduction to Linear Algebra》 Gilbert Str... 阅读全帖

发帖数: 1
4
计算数学是指numerical analysis和computational approximation, 你说的这部分用
的是linear algebra,跟numerical linear algebra的知识段,并不一样的。
numerical analysis指的是Newton's method,Euler's formula这些。
而numerical linear algebra 指的是 lease square,Jacobian matrix 这类知识段
Convolutional Neural Networks 都用到我提的这些知识面了吗?
a********g
发帖数: 11
5
Checked Baggage Weight and Measurement
Baggage is measured by adding length, width and height to determine the
total number
of linear inches/centimeters.
Northwest will accept baggage up to a maximum of 50 pounds (23 kg) and a
total linear
dimension of 62 inches (158 cm) per piece.
Additional charges apply for baggage weighing more than 50 pounds (23 kg) or
with a
total linear dimension of more than 62 inches (158 cm) per piece.
For domestic and trans-Pacific travel, the maximum weight accepted
k*z
发帖数: 4704
6
来自主题: Classified版 - Entry level Data and Optimization Analyst
日常工作是简单的ETL和performance reporting,项目什么都有,segment,pricing
optimization, operation optimization, performance optimization, heat map.
工作语言是SQL和SAS,Reporting语言是Cognos+VBA+MDX/SSRS,不过以上这些不会不要
紧,可以培训。
需要有基础编程经验,Python, C++,Java,R, Matlab任何语言都可以,会写简单的
simulator和calculator. 这个会面试问到。
需要了解各种模型能解决运营实际问题,不需要知道如何具体apply,但是需要知道问题
应该在哪个track上解决。例如:信用估值(logistic),精准营销(cluster/decision
tree),预测需求 (time series),工作排班(linear programming),客户分类(
cluster),市场调查(marketing research). 工作会涉及到的有glm, ets,logistic
,
linear in... 阅读全帖
S*****n
发帖数: 4185
7
Journal Publications
M. Wang and D. P. Bertsekas, " A Unified Framework of Random Incremental
Methods for Large-Scale Convex Optimization ," In preparation.
M. Wang and D. P. Bertsekas, "Incremental Constraint Projection Methods for
Variational Inequalities," Lab. for Information and Decision Systems Report
LIDS-P-2898, MIT, December 2012; Submitted.
M. Wang and D. P. Bertsekas, "Stabilization of Stochastic Iterative Methods
for Singular and Nearly Singular Linear Systems," Lab. for Information ... 阅读全帖
t**d
发帖数: 352
8
来自主题: JobHunting版 - an interview question
Design a data structure for a text editor. require better than linear for
both line search and insert a new line.
for example, you can use String[] to store lines in a page. then search for
a particular line is constant time, but add a new line is linear.
you can chose to use LinkedList, but now insert is constant time while
search is linear.
p******x
发帖数: 441
9
得到一个strong at math but lack of clarity and knowledge on statistics的评语
,憋半天实在忍不住上来再吐槽一下。
Multicolinearity问题:
首先,最简单的模型 Y=Xb+e 的LS解是b_hat=(X’X)^(-1)X’Y, var (b_hat)=(X’X
)^(-1) sigma^2.
问题:什么是Multicolinearity,
答:如果承认X是rv,才能用”correlated “,否则只能用比较数学的linear
dependent,not of full column rank这种术语。
Multicolinearity又分2种,multicolinearity 和perfect multicolinearity,分别对
应的是X的column vectors 是 nearly linear dependent和 linear dependent(not
full ranked),分别对应的结果就是 (X’X) 是ill-conditioned 和singular. ,前者
是(X’X)^(... 阅读全帖
M******9
发帖数: 10
10
来自主题: JobHunting版 - 我的面试总结(FLGT+UPASD)和伪面经
基本都面完了,前一段刚注册了一个帐号,上来发文,大概说下自己的经历,抛个砖头
,希望对大家有用,也祝愿大家都能找到满意的工作。有NDA就不说onsite具体题目了
,感觉也没什么必要说,会大概说说面到的知识点,可能比较乱,大家将就着看。
基本情况:fresh cs phd, 找的都是SE的工作,为啥不找教职或者research lab这里就
不讨论了. FLGT(2 offers, 1家withdraw, 1家简历被刷), startups UPASD(2 offers,
2家电面挂,1家没申请)
pros:背景还不错,都是top school, GPA高。。(fresh貌似公司还是会稍微看看这个)
cons: 没有intern经验是硬伤,PhD期间,上完课后代码写得不多
package还没开始谈,initial offer都差不多200k+的样子,大公司hr明确表示等我都
面完了可以谈, startup都是late stage, 股票都是十万分之5-10, 感觉不好谈。LD目
前在一家大公司,说其实先去大公司几年也不错,比较稳定,貌似股票refresh也可能
不错,work/life... 阅读全帖
M******9
发帖数: 10
11
来自主题: JobHunting版 - 我的面试总结(FLGT+UPASD)和伪面经
基本都面完了,前一段刚注册了一个帐号,上来发文,大概说下自己的经历,抛个砖头
,希望对大家有用,也祝愿大家都能找到满意的工作。有NDA就不说onsite具体题目了
,感觉也没什么必要说,会大概说说面到的知识点,可能比较乱,大家将就着看。
基本情况:fresh cs phd, 找的都是SE的工作,为啥不找教职或者research lab这里就
不讨论了. FLGT(2 offers, 1家withdraw, 1家简历被刷), startups UPASD(2 offers,
2家电面挂,1家没申请)
pros:背景还不错,都是top school, GPA高。。(fresh貌似公司还是会稍微看看这个)
cons: 没有intern经验是硬伤,PhD期间,上完课后代码写得不多
package还没开始谈,initial offer都差不多200k+的样子,大公司hr明确表示等我都
面完了可以谈, startup感觉不好谈。LD目前在一家大公司,说其实先去大公司几年也
不错,比较稳定,貌似股票refresh也可能不错,work/life balance比较好。我自己是
想去startup, 但... 阅读全帖
a*******n
发帖数: 237
12
一直觉得喉咙不舒服,家庭医生说没问题,在我的坚持下照了个CT。
那个兄弟姐妹帮我看看下面说的是什么。名词太多,看不懂。
谢谢。
-------------------------------------
FINDINGS:
Limited images of the lower brain demonstrate no gross abnormality. In the

axial plane, there is an apparent ring enhancing 7 mm lesion in the left

cerebellum, but sagittal images suggest that this appearance is created by a

looping tortuous vessel rather than a true focal lesion.
Rounded soft tissue densities in the left maxillary sinus and left nasal
cavit... 阅读全帖
z***i
发帖数: 9880
13
来自主题: Parenting版 - 我眼里的老美的数学水平
Many business majors need to know linear regression. Like marketing,
operations, and finance.More and more jobs requires coding.
The most important benefit from math training is not how to calculae. It is
the way of logic thinking (based by my personal experience).
I have taught a lot of average U.S students and some average Chinese
students (coming from 2nd tier Universities of China). The difference on
math background between them is huge. Many of people in this board are only
experiencing top... 阅读全帖
b*****e
发帖数: 499
14
来自主题: Parenting版 - 我眼里的老美的数学水平
你说的这些还是归于数据分析这一块啊。
其实不是搞research的话,business majors的学生也未必真的很懂linear regression
,很多人run run 程序,写写报告就行了。
再说了,Obama多半不懂linear regression。
Steve Jobs懂不懂linear regression也不好说。

is
only
any
z***i
发帖数: 9880
15
来自主题: Parenting版 - 我眼里的老美的数学水平
Many business majors need to know linear regression. Like marketing,
operations, and finance.More and more jobs requires coding.
The most important benefit from math training is not how to calculae. It is
the way of logic thinking (based by my personal experience).
I have taught a lot of average U.S students and some average Chinese
students (coming from 2nd tier Universities of China). The difference on
math background between them is huge. Many of people in this board are only
experiencing top... 阅读全帖
b*****e
发帖数: 499
16
来自主题: Parenting版 - 我眼里的老美的数学水平
你说的这些还是归于数据分析这一块啊。
其实不是搞research的话,business majors的学生也未必真的很懂linear regression
,很多人run run 程序,写写报告就行了。
再说了,Obama多半不懂linear regression。
Steve Jobs懂不懂linear regression也不好说。

is
only
any
t*******r
发帖数: 22634
17
属实。数学分析是 Mathematical Analysis,比微积分涵盖面大得多。
比如马工的 Non-linear Programming,隶属于 Numerical Analysis,
而 Numerical Analysis 又隶属于 Mathematical Analysis。。。
但不知道哪个学校的微积分课会教 Non-linear Programming。。。
其实 Non-linear Programming 加个后处理,还可以用来解某一小部分
离散问题的。。。
A***u
发帖数: 3714
18
来自主题: Parenting版 - 爱好天文的孩子们有福了,
有哪个父母能不能找到转播的电视台。
GENERAL
Get ready to witness meteor shower next week
Camelopardalids is never-before-seen event, astronomers say
By Aftab Kazmi, Bureau Chief
May 18, 2014
Al Ain: Stargazers in the UAE will get a nature’s treat next week by
witnessing a never-before-seen meteor shower called Camelopardalids.
The spectacular shower will be visible on the morning of May 24 when the
earth, for the first time, will pass through comet 209P/Linear’s leftovers.
Astronomers have been making arrangement... 阅读全帖
t*******r
发帖数: 22634
19
来自主题: Parenting版 - [bssd] 小孩子做题蛮有意思的。
这题当然有逻辑。。。不过一来 linear regression 不是万金油,二来 linear
regression 本来就是代数范畴的,为啥你许用 linear regression 但不许用方程解?
这题不用方程就图解,分子不动分母动。。。所以先把分子对齐,这样就是 3/12 搞成
3/13,分母加上 1。。。这 1 对应于 24,所以统统放大 24 倍。。。所以是 24*3=
72 个白子。。。好像也不用啥追击电影。。。不过俺最近推代数和 science TV,不纠
结这个。。。

becomes
t******l
发帖数: 10908
20
来自主题: Parenting版 - 构建式数学在中国
另一方面,我觉得 “美式构建式数学教育”,在 代数/前代数/Formal-Operation 阶
段最大的问题,两个:
其一:是在 Formal Operation 阶段,没有考虑 Formal Operation 阶段的特点,照抄
小学阶段 Concrete Operation 的 spiral 教学法。这带来 (a) 教学进度太慢,教授
的基础知识太少,花太多时间在不必要的 spiral 上,(b) 在 Formal Operation 的
spiral 的循序渐进层面,也有问题。
这个原因是 Formal Operation 是有明显的阶段性和结构性的,而 spiral 的时候很难
越过阶段性和结构性的边界。举个例子就是:
Algebra 1 和 "linear" geometry / "linear" analytic geometry,比如 solve
linear equation, solve system equations, complete a square, solve quadratic
equation, equation of line, slope of ... 阅读全帖
c******8
发帖数: 3170
21
来自主题: Stock版 - 我的看法
http://www.springerlink.com/content/718052234449562k/
On the best linear approximation methods and the widths of certain classes
of analytic functions
S. B. Vakarchuk
Abstract
We discuss the best linear approximation methods in the Hardy spaceH q q≥1,
for classes of analytic functions studied by N. Ainulloev; these are
generalizations (in a certain sense) of function sets introduced by L. V.
Taikov. The exact values of their linear and Gelfandn-widths are obtained.
The exact values of the Kolmog
c******8
发帖数: 3170
22
来自主题: Stock版 - 我的看法
http://projecteuclid.org/DPubS?service=UI&version=1.0&verb=Display&handle=euclid.jap/1143936252
Journal of Applied Probability
Local linear approximations of jump diffusion processes
J. C. Jimenez and F. Carbonell
Source: J. Appl. Probab. Volume 43, Number 1 (2006), 185-194.
Abstract
Local linear approximations have been the main component in the construction
of a class of effective numerical integrators and inference methods for
diffusion processes. In this note, two local linear approximations
h****g
发帖数: 312
23
1)债券有非常多的种类,不能一概而论。你指的是不是t-bill,t-note和t-bond?
2)原油价格和index不是linear的。往往牛市中是positive linear,熊市中negative
linear。这也有很多期货交易的书讨论过原因。
k********8
发帖数: 7948
24
来自主题: Stock版 - Re: 学EE太不值了 (转载)
【 以下文字转载自 EE 讨论区 】
发信人: fockugcd (fockugcd), 信区: EE
标 题: Re: 学EE太不值了
发信站: BBS 未名空间站 (Fri Mar 8 19:27:43 2013, 美东)
几年前看linear的网站,写着profit sharing up to base的68%,假设一个很senior
的ic desinger拿15Wbase, 60% profit sharing,共24w,加stock,total应该25w.
semi里面应该linear和maxim这两家pay最好了。我还听人说过忘了是linear还是maxim
,如果是做design engineer的话,profit sahring是从30%起。
N*******g
发帖数: 1089
25
来自主题: Working版 - 不做技术,那你做什么?
你还没有回答我的问题。 坦率的说, 我不相信一个人可以涵盖医疗,通讯,能源所有
系统。 其次, 象我之前说的, 即使一个在其中一个领域, 比如能源, 分的也很细
, 结果也不一样。
当然, 这要看你们生产的是什么, 比如你的产品是芯片, 又或者你的芯片只用在电
源里面, 那么设计行业标准就少的多。 我给你举个例子, 我们设计的inverter 是用
在电动车上的, 但是客户说我能不能用你的inverter 在太阳能里面啊, 销售说行啊
。 问题来了, 太阳能的行业标准和电动车不一样, 比如客户用的thin film panel
那么dc-minus接地, 为什么? 你要知道thin film 的材料问题, 不接地的话,
panel 里面会起化学变化, 会腐蚀材料。 如果非要不接地的话,那么就要对dc用 d
ouble sulation , 可是客户是加州的, 加州又有特殊标准, 就是必
须dc要接地。 那么double insulation就又不能用了。这些都是小的
细节, 也是行业特点, 但是恰恰是这些细节, 决定了产品的应用, 比如我就要为他
们这些行业特点修改设计。 这些还都只... 阅读全帖
K******g
发帖数: 1870
26
However, it is also reported in other study that the xxxx can be represented
accurately by a linear function of the xxxx in a real environment [citing
my paper]. We obtained a similar result
in our preliminary tests as [citing my paper], so we use a simple
linear model in our paper.
好像有点悬。。。但是他绝对用的是我的model
还有这一个,提到了别人的paper:
"Previous studies [citing my paper], [citing somebody else's paper]
have shown that, for xxx, the difference relationship between xxxx in
successive intervals can be approxi... 阅读全帖
l****c
发帖数: 838
27
The history data is never linear, so linear extrapolation can't be correct.
I hope it is faster than linear for several reasons:
(1) CIR should produce something in a few months.
(2) Many companies had mass layoffs in 2008 and 2009, so I don't think many
people could apply for GC in those two years.
(3) Many people with PD of 2008, 2009 and 2010/4 already filed 485 last year,
so new cases should be less.
E******y
发帖数: 614
28
来自主题: Boston版 - Re: [转载] 哪位大侠数学好?

Some of them that I can think of:
极限 - Limit
连续 - Continuous
导数 - Derivitive
微分 - Differential
不定积分 - Integral
无穷极数 - Infinite Series
多元函数的微分 - Multi-Variable Differential
微分方程 - Differential Equation
行列式 - Determinant
矩阵 - Matrix
线性方程组 - Linear Equation
二次型 - Quadratic Form
矩阵的特征值 - Eigenvalue
线性规划 - Linear Programming
非线性规划 - Non-Linear Programming
多目标数学规划 - Multi-Objective Mathmatical Programming
动态规划 - Dynamic Programming
库存论 - Inventory Theory
排队论 - Queuing Theory
决策论 - Decision Theory
模型
p**********u
发帖数: 15479
29
这人一看就是真不懂linear algebra,商学院的mba那种。像daj那样做quant的都知道,
很多数学问题最后都归结到linear algebra。很多收敛之类的概念,都是从linear
algebra那里搞出来的。
D****R
发帖数: 6053
30
来自主题: Fishing版 - 问个问题,鱼竿怎么上飞机?
I believe the carry on linear limit is 45". If it is 1米1, should be
acceptable
The maximum combined linear measurement (length + width + height) of carry-
on bags must not exceed 14 inches x 9 inches x 22 inches (23 x 35 x 56 cm)
or 45 linear inches (114 cm).
S*********g
发帖数: 7653
31
17.6miles
Average pace: 8:21
Average speed: 7.2mile/h
Time: 2:27:24.37
Calories burnt 2138cal
2miles
Average pace 7:59
昨天实行了冬令时,这对我来说无疑是个好消息!
我一般不调表,也就是说,相对地看,我多出了一个小时,早上上课就推迟了一小时,
于是乎我早上
锻炼完就多一个小时recovery到最佳状态:)
不幸的是,我的手机的时间还是会变的,于是我的闹钟一律调到3:30AM。昨天在版上
看到大牛们的
半马全马report,深受刺激和鼓励啊XD。从中午开始carb load,吃大碗面条,下午一边
写paper
一边吃包子,芝麻饼,喝葡萄汁。晚上按夏令时11点就睡了。
早上按时3:30起床,吃了一个bagel,带上一个power bar,4:00AM开跑
carb load应该起作用了,而且感觉心肺从前天的tempo running恢复过来了,一路感觉
非常
easy,首先沿老路跑,然后开到高尔夫球场,接着返回,到linear park,路上就想着要
不要跑进
去,有点犹豫... 阅读全帖
c**i
发帖数: 6973
32
来自主题: Travel版 - Maglev for Tokyo-Nagoya Line
Dan Saltzstein, Japan Will Build World's Fastest Train. New York Times, Feb
20, 2011 (title in the print; date in the print).
http://intransit.blogs.nytimes.com/2011/02
/14/japan-plans-worlds-fastest-train/?scp=1&sq=japan%20train%20magnetic&st=
cse
(to be completed by 2027; 178 miles; in 40 minutes; will use magnetic
levitation
My comment:
(a) Central Japan Railway is private.
(b) Capital of Aichi Prefecture 愛知県, Nagoya is midway between Osaka in
the west and Tokyo in the east.
(c) Maglev (tran... 阅读全帖
z***y
发帖数: 473
33
从wiki上抄了一段好像能说明点问题:
"Both film and digital sensors exhibit non-linear responses to the amount of
light, and at the edges of the dynamic range, close to underexposure and
overexposure the media will exhibit particularly non-linear responses. The
non-linear dynamic response or saturation qualities of emulsion film are
often considered a desirable effect by photographers, and the distortion of
colour, contrast and brightness varies considerably between film stocks.
There is no limit to the number o... 阅读全帖
k**o
发帖数: 3006
34
☆─────────────────────────────────────☆
apolitical (无期徒刑) 于 (Thu Apr 11 12:52:03 2013, 美东) 提到:
趁板神在黑屋里关着,我来行为艺术啦。用希大师的eos 3拍了一场比赛。说明书上写
连拍是每秒4.7张,但实际操作达不到,不过我算了一下,一卷24张,洗扫花8块多,基
本按动快门每秒就扔出去一块多钱,连拍体育有肾上腺激素的涌动,这次跟着流动的还
有现金。
重点难题在对焦上,虽然能在胶片机上找到佳能现在数码对焦操作的影子,但我还没找
到感觉。对上焦很快很准,但追焦跟踪的难度很大。都说佳能技术储备强,我时光倒流
追抚奸商的过去,还需要更多时间摸索。我应该还会再谈谈对焦问题。
fuji superia 400
kodak ektar 100
顺便对比一下用7d拍的,光照条件相似。
对比胶片和数码,除了明显的不同色调外,胶片在明暗过度的处理上会不会好一些?条
件是大太阳直射,无云,7d给了明显的数码味,对比度很强,显得光很硬,胶片则显得
在细节上不是那么对比强烈,能看出是晴天,但不是那种让人睁不开眼... 阅读全帖
C**r
发帖数: 3
35
来自主题: WaterWorld版 - 1989年出生的Princeton AP (转载)
北京八中神童班的(初高中连读),14岁上清华,18岁清华3字班毕业,然后到MIT读博
,读书期间竟然拿到了美国绿卡。。。
贴个她的简历吧
来源:web.mit.edu/mdwang/www/CV.pdf‎
Mengdi Wang
Department of Electrical Engineering and Computer Science
Massachusetts Institute of Technology (MIT)
77 Mass. Ave., 32-D678
Cambridge, MA 02139
Mobile: 617-543-7694
Email: mdwang at mit dot edu
http://www.mit.edu/~mdwang/home.html
Research Interests
•Large-scale optimization problems that are driven by big data or
random processes, and related
stochastic methods.
R... 阅读全帖
d********f
发帖数: 43471
36
【 以下文字转载自 JobHunting 讨论区 】
发信人: forwhat2005 (风乱), 信区: JobHunting
标 题: 沾满鲜血的yahoo面经啊
发信站: BBS 未名空间站 (Thu Oct 24 17:35:15 2013, 美东)
代朋友发的yahoo面经,24小时内新鲜出炉的。
这哥们装纯说没有买买提的账号,刚注册了要三天后才给发帖,所以让我代发。我不懂
这些CS名词,也就发来让大家茶余饭后赏玩一番
——————————————————————————————————————
——————————————————————————————————————
————————————
真是郁闷,你帮我发吧,我没有账号,<>号内部分。
<背景是非CS领域Ph.D, CS M.S.。之前onsite面了四轮,哉在老印手上,估计在
boundary上所以又加了一轮电面。
面试官是个台湾人。上来先给了道coding小题,数组里找最大连续sum,说了半天不要
用brute force。我随首写了一个O(n)的,湾湾半信半疑,检查了老半天,自己还出了
很多test cas... 阅读全帖
b********s
发帖数: 6928
37
correct me if this isn't right
want to test the correlation of the two variables? use any basic softw
are package and look for the Pearson's product-moment coefficient? the
larger the value, the stronger the association of the two variables?
want to see how the dependent variable changes with the independent va
riable? try linear regression, test significance of the slope??? if do
esn't fit, add polynomial terms???
your idea is correct. if the graph shows a line that is almost linear,
it means t... 阅读全帖
a*******o
发帖数: 129
38
Hi, all,
I have a nonlinear constraint which i don't know how to linearize it.
it is in the form
x*y=z
x and y both have a bound.
Then, how can i linearize it? I hope I can achieve a LP or MILP form by
linearization.
Thank you very much.
r****c
发帖数: 2585
39
第一个问题的time complexity is linear, and space complexity is O(1)
只给一个简单可以理解的solution,
1. 先考虑所有长度为四的连续数字,找出最大的一个,一共需要linear time and O(1)
space. Let max be the sum of the max solution.
2. 考虑长度为四以上的数字,如果是最大,头和尾一定是正数. 从第一个正数开始linear
scan
a) assume first positive is A[i], consider x = A[i]+A[i+1]+A[i+2]+A[i+3]+A[i+4
], j=A+5
if it is negative then begin from A[j] and find first positive one.
b) If x is positive, then iteratively add x=x+A[j] until x become negative. In
this process,
if x > max, then se
d******e
发帖数: 7844
40
如果读applied statistics确实没什么用,里面的Regression,Anova那些东西和ML根本
不搭边。
数学课程有用的很多。比如Real Analysis,Linear Programming,PDE,Advanced
Linear Algebra。而统计的课程,修一修Probability Theory,Probability Model,
Multivariate, Linear Model, Statistical Inference也是很有必要的。
其实不是统计跳到CS来做东西,而是这东西本来就是共有的。
能自学数学的牛人,不修这个学位也罢。对于非牛人,读一个还是能学到不少东西的
r********3
发帖数: 2998
41
来自主题: CS版 - Valiant 是理论大牛
你的意思是说,研究了几十年的数据库技术都是小trick了? 在计算机领域里面,除了
最老的OS,System那个派系外,还没其他哪个派系刚这样说database领域。
超大规模数据处理的时候,即便是linear和sub-linear都不行的。KNN是可以直接做到
log级别的。话说回来,即便同样是linear,数据访问的顺序不同,也会造成几倍的差
别。你要是了解现在计算体系结构就明白了。
c*m
发帖数: 1114
42
来自主题: Hardware版 - 不用64位系统的理由?
你那篇文章我粗略读了一下。你现在这么一说我很不好不好意思,专门去仔细读了一下

文中的主要观点如下:
1. 32bit 可以用 PAE 用 >4GB内存。
2. 以前32bit OS不能用4GB内存是因为直接用Page file,所以只能access 2^32~= 4GB内
存。现在的PAE用来linear mapping技术,所以可以应用程序并不直接access内存,先在
PAE里面linear mapping了physical memory在用。
3. 64bit系统自然不用PAE, 32位系统需要多一层操作用PAE linear mapping physical
memory。 看起来32bit+PAE access内存会慢很多,事实上64bit因为PageTable比较大
,也有额外开销,这样看起来32bit的memory performance不比64bit的差。然后本文作
者用了windows 7 experience index来测试了一下,发现64bit的memory index只有5.0
, 32bit的memory index不管开不开PAE都有5.1. 所以..
s*i
发帖数: 5025
43
来自主题: Programming版 - Re: 请教一道题目
1. Make two groups
Group A, i<=j
Group B, i>j
linear time
2. Make a hashtable for group A, use i as key and j as value
(At the same time, do some mark work on identical nodes. i==j cases should
be finally calculated)
linear
3. for each in group B, use n as key to retrieve value in the table
and compare the value to m.
linear
w**z
发帖数: 8232
44
来自主题: Programming版 - 下一个热点是啥?
刚读了这篇文章,作者觉得yahoo 应该买Hulu。
https://www.linkedin.com/pulse/article/20140919134200-2539455-yahoo-finally-
buys-hulu-what-it-should-do-with-its-8-3-billion-alibaba-windfall?trk=tod-
home-art-list-small_2
Ready, set, GO! It is Alibaba IPO time ... right now. Set to be the biggest
IPO in US history (that’s $22 billion). And, prescient Yahoo! is set to get
$8.3 billion of it.
What should Yahoo! do with that windfall?
VIDEO, that’s what! Video is Yahoo!’s future, plain and simple. With this
massive cas... 阅读全帖
f*********g
发帖数: 25
45
来自主题: Programming版 - 一层卷积很容易理解...
如果convolution之后没有non-linear operation,根据Central-limit theorem,最终
的filter会变成Gaussian Filter。如果有non-linear operation,可以看一下这篇文章
https://www.di.ens.fr/~mallat/papiers/Bruna-Mallat-Pami-Scat.pdf
虽然它里面的non-linear operation用的是Fourier modulus针对的是complex value
filte和CNN的不一样,但是从single processing 和能量传播角度去看这个问题挺有意
思的
g****t
发帖数: 31659
46
来自主题: Programming版 - 一层卷积很容易理解...
Before finishing Einstein and Fourier's papers,
why we need to waste time on these papers?
I am old and I usually only checked modern papers' result and test data,
coding, 这些干货。


: 如果convolution之后没有non-linear operation,根据Central-limit
theorem
,最终

: 的filter会变成Gaussian Filter。如果有non-linear operation,可以
看一下
这篇文章

: https://www.di.ens.fr/~mallat/papiers/Bruna-Mallat-Pami-Scat.pdf

: 虽然它里面的non-linear operation用的是Fourier modulus针对的是
complex
value

: filte和CNN的不一样,但是从single processing ... 阅读全帖
g****t
发帖数: 31659
47
来自主题: Programming版 - 装个机快被搞死了
恕我直言,你这属于十窍通了九窍,一窍不通。
数学方法如何用,数学模型干什么的都没入门。
还多了个乱套名词的毛病。拉格朗日都出来了。
按照你的那个所谓“论证”,
我儿子现在初中学的linear model更没用。
你尽可以说:
Linear model is doomed, in long term
Linear regression is doomed, in long term
This bbs had already been doomed, in long term
...
如果你找到一个更好的办法分析minst或者别的
什么,分享一下,那技术上好事。
在意义上随便延伸,那就是三观问题。


: 有空看看我的题目 如何?

m****r
发帖数: 141
48
I am compiling a LaTeX file with 10 chapters and a table of contents. I have
added the following in the first page.
But why is the table of contents still missing?
thanks !!!
Any help will be appreciated.
This is the main file.
\pdfbookmark[1]{TABLE OF CONTENTS}{table}
\tableofcontents
\addtocontents{toc}{\def\protect\@chapapp{}} \cleardoublepage \
phantomsection
\addcontentsline{toc}{chapter}{LIST OF TABLES}
\listoftables
\cleardoublepage \phantomsection \addcontentsline{toc}{chapter}{LIST OF
F... 阅读全帖
m****r
发帖数: 141
49
I am compiling a LaTeX file with 10 chapters and a table of contents. I have
added the following in the first page.
But why is the table of contents still missing?
thanks !!!
Any help will be appreciated.
This is the main file.
\pdfbookmark[1]{TABLE OF CONTENTS}{table}
\tableofcontents
\addtocontents{toc}{\def\protect\@chapapp{}} \cleardoublepage \
phantomsection
\addcontentsline{toc}{chapter}{LIST OF TABLES}
\listoftables
\cleardoublepage \phantomsection \addcontentsline{toc}{chapter}{LIST OF
F... 阅读全帖
I***d
发帖数: 1023
50
来自主题: AnthroLing版 - 最近看aegean的东西比较多
剛剛看完一本書,Oliver Dickinson的The Aegean Bronze Age。這是同另外兩本愛琴文
明的書一起借回來的。因為這一本在愛琴考古裏面比較經典比較全面,所以從它開始看。
還是毅力不夠,看了兩個月才看完。
看過有很多零雜的感觸,最主要的一點,是對愛琴文明和考古學現狀有了一個比以前
更全面而且清楚些的認識。那時候才看linear B的書的時候,比較興奮,期望值升得很高
,大約也是因為linear B的解讀證明它是希臘語系的文字罷。看過這本愛琴文明的書,期
望值回落了一些。從克里特島到邁錫尼,linear B還只是宮廷當中最有限的記錄符號,不
能多方面的反映當時的文化。換言之,愛琴文明比起近東其它更古老的文明,還是相對粗
糙原始一些的。
即使這樣,還是調落了。看書的時候,雖然隔著紙,隱隱有一種能觸摸的感覺。書上
講一個墓葬,就能零散的想到很多東西。對考古學還是懷了一種敬佩感,尤其近東,在西
方學界,真是研究得很熱鬧。想到我的拉丁文老師之一,就曾經在Pompeii工作過一年多
。我去Pompeii的時候,就只是在人群裏面,一點點看那些剝開的痕跡,同樣的
首页 上页 1 2 3 4 5 6 7 8 9 10 下页 末页 (共10页)