L****8 发帖数: 3938 | 1 本质是对各种scale translation进行穷举(convolution)
然后把结果综合起来 |
|
c*******9 发帖数: 9032 | 2 Translation 不算伪概念吧,因为convolution的功用不仅仅是穷举Translation(更不
是最主要目的)。 |
|
|
|
|
L****8 发帖数: 3938 | 6 translation invariant 似乎没有了 |
|
h**********c 发帖数: 4120 | 7 从数值计算的模型来讲,
一个动力系统,又过去好多年了,现在回忆一下。建立起模型,比较理想是一个ODE,
常微分,invariant不变。这样的模型不是说好算,而是可以分析能不能算。
比如RUNGE KUTTA 方法,可以根据前面的数据推算出将来的状态,还有Euler 方法,
implicit Euler 等。本质是这样的系统是可导的。
但是稳定性分析,可以分析一个ODE系统的计算结果是不是收敛的,不收敛越算误差越
大。最后结果是荒谬的。
稳定行分析 还会分心一个系统是否会有混沌行为,比如strange attractor.混沌系统
就是轻微的干扰,系统就会产生强烈随机的反馈。
啰嗦了一气说明,在特定条件即便是可以见模建的很好的ODE系统模型,也会会这样或
那样的无法预测。
另外一个我个人的观点就是在同步处理当中,只有一个线程或进程死掉了,可以作出线
程状态的结论,否则你绝对不会知道你下次读的状态。就是活物没法预测。 |
|
h**********c 发帖数: 4120 | 8 觉得这本质是一个哲学问题,预测是
1.决定你不做什么或做什么,你基本会输
2.不能决定你做什么不做什么,你一定会赢
从数值计算的模型来讲,
一个动力系统,又过去好多年了,现在回忆一下。建立起模型,比较理想是一个ODE,
常微分,invariant不变。这样的模型不是说好算,而是可以分析能不能算。
比如RUNGE KUTTA 方法,可以根据前面的数据推算出将来的状态,还有Euler 方法,
implicit Euler 等。本质是这样的系统是可导的。
但是稳定性分析,可以分析一个ODE系统的计算结果是不是收敛的,不收敛越算误差越
大。最后结果是荒谬的。
稳定行分析 还会分心一个系统是否会有混沌行为,比如strange attractor.混沌系统
就是轻微的干扰,系统就会产生强烈随机的反馈。
啰嗦了一气说明,在特定条件即便是可以见模建的很好的ODE系统模型,也会会这样或
那样的无法预测。
另外一个我个人的观点就是在同步处理当中,只有一个线程或进程死掉了,可以作出线
程状态的结论,否则你绝对不会知道你下次读的状态。就是活物没法预测。 |
|
b**********h 发帖数: 419 | 9 感觉这些概念是编程课里某个重要Topic里的
有什么书可以系统学习这些概念的, 最好是Java的书,
谢谢 |
|
|
|
n*w 发帖数: 3393 | 12 Java应该不支持covariance?除了c#, 主流的还有哪个? |
|
T*******x 发帖数: 8565 | 13 这个道理我还是没明白:如果C是A的subtype,为什么在Java中,list of C里不能存一
个type A的object,而在haskell里就可以。 |
|
n*w 发帖数: 3393 | 14 Java的list中又进又出。haskell只出 |
|
|
|
T*******x 发帖数: 8565 | 17 list(C) 是 list(A) 的 sub type?
这样的话就是sub type关系了。 |
|
n*w 发帖数: 3393 | 18 yes, 反直观的是contravariant. |
|
T*******x 发帖数: 8565 | 19 谢谢。这个地方我还是没想明白,但是我不知道该问什么问题好了。 |
|
h**********c 发帖数: 4120 | 20 看过一个讲座,忘了,
我理解covariant是change the ceiling,contravariante是有一个floor.
实际中没用过,不过JEE里好多吗喜欢这么用。 |
|
n*w 发帖数: 3393 | 21 想到一个contravariant解释,假如有一个function eat,接受一个参数,类型为鱼。
那么吃动物应该是其子类。大吃一鲸不是。 |
|
l******t 发帖数: 55733 | 22 这是oo继承固有的问题。fp没这个问题。fp的category是mixin的。 |
|
g****t 发帖数: 31659 | 23 来自主题: Programming版 - 数值计算 我的意思很简单,就是空间里有一块地方,simulation很多步,然而结果走的极慢。但
是它又不是invariant manifold,所以会有数值困难。
: 我不是太理解这个slow manifold的概念,就说我论文里的计算,我导师自己给
个概念
: 叫orbit continuation,在平衡点附近grow 一个周期轨道,不断算这个轨道的
: continuation,因为连接两个不同平衡点的周期轨道的轨道也是异构环。我算的
都是2
: manifold.我又查了一下,你说的slow manifold不是homotopy,可能开会的时候
听一个
: 人问过collision,是不是这个意思?
: 我这人挨溜号,搞计算的时候图形玩的多,当了马公,还老想着计算的东西。
: 感觉还是有点多动症,还好没有发展成自闭或弱智。
|
|
h**********c 发帖数: 4120 | 24 来自主题: Programming版 - 数值计算 我老当年学传热学时候公式比这复杂多,不过老师也不明白,或着她认为我们不需要明
白,知道算什么查那本书,不过传热的计算多数可以在工程许用范围内验证,不是蒙人
的随机数。工程上科一半要做实验,验证理论,每个人都能重复。
你不能建立你的模型象auto 手册,搞成invariant Ode模型,search auto07p.pdf 或
auto2000.pdf
比如象bruta problem,Sandstede 的模型.
我说的仅供参考。 |
|
g****t 发帖数: 31659 | 25 来自主题: Programming版 - 数值计算 我的意思很简单,就是空间里有一块地方,simulation很多步,然而结果走的极慢。但
是它又不是invariant manifold,所以会有数值困难。
: 我不是太理解这个slow manifold的概念,就说我论文里的计算,我导师自己给
个概念
: 叫orbit continuation,在平衡点附近grow 一个周期轨道,不断算这个轨道的
: continuation,因为连接两个不同平衡点的周期轨道的轨道也是异构环。我算的
都是2
: manifold.我又查了一下,你说的slow manifold不是homotopy,可能开会的时候
听一个
: 人问过collision,是不是这个意思?
: 我这人挨溜号,搞计算的时候图形玩的多,当了马公,还老想着计算的东西。
: 感觉还是有点多动症,还好没有发展成自闭或弱智。
|
|
h**********c 发帖数: 4120 | 26 来自主题: Programming版 - 数值计算 我老当年学传热学时候公式比这复杂多,不过老师也不明白,或着她认为我们不需要明
白,知道算什么查那本书,不过传热的计算多数可以在工程许用范围内验证,不是蒙人
的随机数。工程上科一半要做实验,验证理论,每个人都能重复。
你不能建立你的模型象auto 手册,搞成invariant Ode模型,search auto07p.pdf 或
auto2000.pdf
比如象bruta problem,Sandstede 的模型.
我说的仅供参考。 |
|
T********i 发帖数: 2416 | 27 差不多10年前了。C++。彻底的kernel bypass。NUMA, huge TLB,CPU isolation。
当时连靠谱的NUMA allocation机制都找不到。发现最可靠的就是reserve一个huge
page然后根据CPU socket平均分配一下pool。
custom memory allocator。用的最简单的binary。allocate的时候找最接近的2^n空闲
块,如果找不到就在pool里顺序分配一个。free的时候把这个块归还n的list里边。
allocate和free都是个位数指令。理论上内存浪费可能多一倍。实际上没那么多。其实
系统死循环运行,memory usage pattern可以证明是固定的。
我个人经验。这种lock free架构和传统架构是格格不入的。根本没有任何办法能够把
他们整合起来。多核间通信都是invariant message passing。传递的都是指针。这是
硬实时架构,延迟是guaranteed。除非你哪里写错了,那种情况直接主动崩溃是最好的
选择。如果你写对了,消息都是在一个简单的循环队列里边,在消息被擦除前,保... 阅读全帖 |
|
f*********l 发帖数: 176 | 28 【 以下文字转载自 TV 讨论区 】
【 原文由 feelingfool 所发表 】
经过invariable兄两天以来的不懈努力
终于完成了!
all the files have been checked.
刚才从Serv-U转换到G6 FTP
开始时候导致若干同志们连接失败
这里说声sorry!
feelingfool.2y.net |
|
发帖数: 1 | 29 WHY IDO?
http://idolinguo.org.uk/whyido.htm
NOTE: Because of the difficulty of showing Esperanto's special accented
letters, where a letter which in Esperanto should carry a circumflex it is
shown here followed by a circumflex (^), and where the letter 'u' should
carry a breve it is followed by a tilde (~).
The idea of using a constructed language as a medium of international
communication - not replacing existing languages but supplementing them - is
far from new. The first such language to ach... 阅读全帖 |
|
发帖数: 1 | 30 The international language Ido - a general description
It would be very useful if we could talk with people in other countries, or
correspond with them, as we can with people in our own country. However, the
language barrier often makes this difficult if not impossible.
The answer to this problem given by many people is: let them (that is,
everyone else) learn English! Certainly English is the most widely spoken
language in the world, but it requires a lot of time and some skill to learn
it at a... 阅读全帖 |
|
c**i 发帖数: 6973 | 31 Waiiiiiit. I am from Massachusetts. We are No. 1--to a certain extent.
I know little about UK or Oxford University. But we in Massachusetts are
proud of the achievements of our state.
(1) First I question veracity of data for Baylor College of Medicine (BCM).
I have rarely heard of it, and have no idea whether it is good or bad. But
the author of the original posting never cites the source of his data--and I
think he is off mark.
Biomedical funding in US can come from National Institute of Healt... 阅读全帖 |
|
c**i 发帖数: 6973 | 32 Waiiiiiit. I am from Massachusetts. We are No. 1--to a certain extent.
I know little about UK or Oxford University. But we in Massachusetts are
proud of the achievements of our state.
(1) First I question veracity of data for Baylor College of Medicine (BCM).
I have rarely heard of it, and have no idea whether it is good or bad. But
the author of the original posting never cites the source of his data--and I
think he is off mark.
Biomedical funding in US can come from National Institute of Healt... 阅读全帖 |
|
w*********s 发帖数: 2136 | 33 Why Almost Everything You Hear About Medicine Is Wrong
http://www.newsweek.com/2011/01/23/why-almost-everything-you-he
If you follow the news about health research, you risk whiplash. First
garlic lowers bad cholesterol, then—after more study—it doesn’t. Hormone
replacement reduces the risk of heart disease in postmenopausal women, until
a huge study finds that it doesn’t (and that it raises the risk of breast
cancer to boot). Eating a big breakfast cuts your total daily calories, or
not—as a st... 阅读全帖 |
|
l**********1 发帖数: 5204 | 34 那是 Poincaré 龐加萊猜想的解铃人 俄罗斯数学家 Perelman 佩雷尔曼
//en.wikipedia.org/wiki/Grigori_Perelman
please go to
//en.wikipedia.org/wiki/Henri_Poincaré
//en.wikipedia.org/wiki/Poincaré_conjecture
and
名稱三體問題
英語名稱:three-body problem
N體問題及三體問題的概念N體問題:
N體問題可以用一句話寫出來:在三維空間中給定N個質點,如果在它們之間只有萬
有引力的作用,那麽在給定它們的初始位置和速度的條件下,它們會怎樣在空間中運動。
三體問題:
最簡單的例子就是太陽系中太陽,地球和月球的運動。在浩瀚的宇宙中,星球的大
小可以忽略不記,所以我們可以把它們看成質點。如果不計太陽系其他星球的影響,那
麽它們的運動就只是在引力的作用下産生的,所以我們就可以把它們的運動看成一個三
體問題。
天體力學中的基本力學模型。研究三個可視爲質點的天體在相互之間萬有引力作用
下的運動規律問題。這三個天體的質量、初始位置和... 阅读全帖 |
|
w*****r 发帖数: 2061 | 35 "While this chronology depicts an extraordinary life from humble beginnings
to the dais in Stockholm for the Nobel Prize, it does not do justice to his
remarkable personal qualities. Julie was the ultimate mentor. His laboratory
was not simply a place where good science was done, but one where the
fellows learned how to do good science. Julie's government-issue steel desk
was strategically placed in the laboratory, where it was 4 feet from the
reagent scale and 10 feet from the scintillation cou... 阅读全帖 |
|
m*****0 发帖数: 72 | 36 需要刺激人的cd4 T细胞,需要合成一个多肽,loading 到MHC II分子上。需和合成这
么几种多肽中一种:
Fibrinogen-α (P02671) 78 to 91, Aggrecan (P16112) 84 to 103, Vimentin 66
to 78 , Collagen II (P02458) 1,236 to 1,249 以及 CLIP, CD74,(P04233)
Invariant chain 106-120。请问哪种能 most commonly recognized amongst healthy
donors MHC-II DR1 or DR4?
thanks a lot.. |
|
D*a 发帖数: 6830 | 37 原标题: A bridge to business
Peter Fiske
Nature 530, 243–245 (11 February 2016) doi:10.1038/nj7589-243a
PhD holders should not underestimate their value to industry and the
business sector, says Peter Fiske.
Subject terms: Careers Business Education
The glossy poster at my university career planning and placement centre both
intrigued and perplexed me: “PhDs: come learn about a career in management
consulting — recruiting reception tonight!”
As a PhD student in geology, I was dimly aware of the nam... 阅读全帖 |
|
D*a 发帖数: 6830 | 38 原标题: A bridge to business
Peter Fiske
Nature 530, 243–245 (11 February 2016) doi:10.1038/nj7589-243a
PhD holders should not underestimate their value to industry and the
business sector, says Peter Fiske.
Subject terms: Careers Business Education
The glossy poster at my university career planning and placement centre both
intrigued and perplexed me: “PhDs: come learn about a career in management
consulting — recruiting reception tonight!”
As a PhD student in geology, I was dimly aware of the nam... 阅读全帖 |
|
c******x 发帖数: 438 | 39 I know a lot of papers I am asking, Thanks in advance.
1. Comparison of hyperelastic models for rubber-like materials
MARCKMANN G. (1) ; VERRON E. (1) ;
Rubber chemistry and technology, 2006, vol. 79, no5, pp. 835-858 [24 page(s)
(article)] (74 ref.)
2.Constitutive equations for amended non-Gaussian network models of rubber
elasticity
Alex Elías-Zúñiga, a and Millard F. Beatty
International Journal of Engineering Science
Volume 40, Issue 20, December 2002, Pages 2265-2294
3.Modeling hyper... 阅读全帖 |
|
n********r 发帖数: 195 | 40 http://www.phrma.org/catalyst/ours-lose-not-just-jobs-medical-p
Ours to Lose: Not Just Jobs, But Medical Progress
by Kate Connors on 7/13/2011
Monday, we discussed how policies that undercut the biopharmaceutical
research sector could lead to huge job losses, with an annual reduction in
sector revenue of $20 billion potentially leading to as many as 260,000 lost
jobs.
The example I used – government-mandated rebates in Medicare Part D –
could be just such a job-killing policy.
But those 260,000 ... 阅读全帖 |
|
M******n 发帖数: 43051 | 41 Basic research has long been the missing foundation of China’s aspiration
to develop a world-class drug industry. The country has become a global
powerhouse in the manufacture of bulk pharmaceutical chemicals. It is a
preferred destination for major drug companies that outsource research work
to reduce costs. To date, however, none of the world’s major drugs has been
either invented or developed in China.
The country wants to change that in a big way. The government is spending $7
billion over t... 阅读全帖 |
|
x*****u 发帖数: 3419 | 42 http://thread.gmane.org/gmane.comp.version-control.git/57643/fo
From: Linus Torvalds linux-foundation.org>
Subject: Re: [RFC] Convert builin-mailinfo.c to use The Better String
Library.
Newsgroups: gmane.comp.version-control.git
Date: 2007-09-06 17:50:28 GMT (3 years, 38 weeks, 3 hours and 54 minutes ago)
On Wed, 5 Sep 2007, Dmitry Kakurin wrote:
>
> When I first looked at Git source code two things struck me as odd:
> 1. Pure C as opposed to C++. No idea why. Please don't talk ab... 阅读全帖 |
|
x*****u 发帖数: 3419 | 43 http://thread.gmane.org/gmane.comp.version-control.git/57643/fo
From: Linus Torvalds linux-foundation.org>
Subject: Re: [RFC] Convert builin-mailinfo.c to use The Better String
Library.
Newsgroups: gmane.comp.version-control.git
Date: 2007-09-06 17:50:28 GMT (3 years, 38 weeks, 3 hours and 54 minutes ago)
On Wed, 5 Sep 2007, Dmitry Kakurin wrote:
>
> When I first looked at Git source code two things struck me as odd:
> 1. Pure C as opposed to C++. No idea why. Please don't talk ab... 阅读全帖 |
|
h*****0 发帖数: 145 | 44 下面的最新经济学杂志排名(2004年)使用了最复杂和详细的方法,是至今发表在最高级
别刊物上Econometrica(2004年)的排名,纠正了以往排名的缺陷,这些方法包括
1. 校正了引用杂志的水平(比如一篇文章被AER引用和被CHINA ECONOMIC REVIEW引
用的得分是不一样的),
2. 使用更长的时间段。SSCI 的影响力因子(IMPACT FACTOR)只用了过去两年发表
在某杂志上的论文在今年被引用的平均次数,使那些时髦杂志如Journal of Financial
Economics,Journal of Health Economics占了便宜(JHE的影响力因子有时可以排经
济学前5)。而这个排名用了过去7年发表的论文。
3. 采用的公理化(axiomatic)的体系。该文证明了其采用的排名的方法是唯一满足4
个公理的方法。这四个公理是 reference intensity, weak homogeneity, weak
consistency,and invariance to splitting of journals。
该文名字是“THE ME |
|
k***g 发帖数: 7244 | 45
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Markov strategies 只是状态(state)的函数,不
是时间(t)的函数:fix state,Markov strategies 是 time-invariant 的。 |
|
|
t***g 发帖数: 60 | 47 最近看别人做的continuous time delta-sigma 仿真,DAC 波形是return-to-zero (
RZ)。若RZ宽度改变,则只需相应反比例改变反馈系数大小。我的疑惑是:一旦改变
dac波形,按照数字和模拟impulse invariance等价转换,整个模拟环路filter参数都
得变,而不仅仅变反馈系数大小就可以了。后者是否只是一种近似方法?初学者,请多
指教。 |
|
B*B 发帖数: 87 | 48 帮忙下载paper
在springer 的Lecture Notes in Mathematics 系列,学校的图书馆只能下最新的,这
个老paper不知道这里有没有人能下到
H.Brehm
Description of spherically invariant random processes by means of g-
functions
Lecture Notes in Mathematics, 1982
PP.39-73
感谢万分!
email: g**************[email protected] |
|
r*****f 发帖数: 247 | 49 Exponential sums as discrete fourier transform with invariant phase
functions
是Springer Berlin / Heidelberg的
x********[email protected]
谢谢 |
|
m********7 发帖数: 54 | 50 g(t) = f(t)*cos(w0*t)
该系统不是LTI系统(确切的说,是linear system,但不是time-invariant 系统),所
以你不能
用输入和冲击响应的卷积来求时域系统响应,相应的,你也不能用频域乘积求其频域响
应。 |
|