k*********g
发帖数: 791 | 1 这几天我一直忙着做网站。
Discontinuous的目的好处：
1 flux control across inter-element boundaries，这样有利于local
conservation; local conservation is beneficial for a better nature of the
discrete system, so that less iterations are required for convergence. also,
flux control is essential for shock/crack caturing schemes.
2 有利于并行计算; the reason behind is a little sophisticated but numerical
practice obviously demonstrates this fact.
DG, discontinuous galerkin, 在计算固体力学领域经常成为 DG FEM 的代名词，但是
，有限元做成 discontinuous 是没有太多意... 阅读全帖 |
|
w***g
发帖数: 5958 | 2 看了眼那paper，极水。文中先假设了欧式空间，先做k-means （或者k-d tree)，然后
对cluster center做spectral cluttering。如果数据真的非常大，那确实得这么做。
但这方法也太显然了，需要发paper吗？
然后spectral clustering最大的好处就是可以处理非欧式空间的数据，甚至都不需要
测度，只要相似度就可以。假设了欧式空间就把spectral clustering这个好处给去掉
了。那还不如直接就k-means得了。
我说的方法目前没人用过。你要有兴趣可以evaluate一下，好的话你可以自己写paper。
注意要写paper的话选dataset的时候一定要选一些非欧式空间的数据，比如我paper中
用的那个EMD。 |
|
I****M
发帖数: 143 | 3 launch Matlab， 进入start》Matlab》time series tools
然后在help中搜索 spectral，using time series tools， help窗口就会出来一个说
明，并shows an example of how to create a spectral plot by dragging the
onboard time series onto the Spectral Plots node，按部就班，希望能够work out。 |
|
m****s
发帖数: 79 | 4 【 以下文字转载自 Immigration 讨论区 】
发信人: majors (minor), 信区: Immigration
标 题: 审稿机会：spectral subtraction denoising
发信站: BBS 未名空间站 (Sun Oct 13 14:36:36 2013, 美东)
接到一个审稿邀请。不过我没时间；审稿邀请说可以推荐。
审稿机会：spectral subtraction denoising
杂志：BioMedical Engineering OnLine
有意者可回信，我会把摘要给你。如果你觉得合适，我就把你推荐给编辑。 |
|
m****s
发帖数: 79 | 5 接到一个审稿邀请。不过我没时间；审稿邀请说可以推荐。
审稿机会：spectral subtraction denoising
杂志：BioMedical Engineering OnLine
有意者可回信，我会把摘要给你。如果你觉得合适，我就把你推荐给编辑。 |
|
w***g
发帖数: 5958 | 6 Spectral clustering算法的瓶颈在于算NxN的相似性矩阵O(N^2)以及对其作特征值分解
。提高速度的关键就是对NxN的矩阵进行稀疏化。可以对每个点算其K-nearest
neighbor，然后矩阵只存K-NN对应的那些值，剩余的全都置0。 然后对稀疏矩阵用迭代
法进行特征值分解。对所有点在所有点上求K-NN (K-NN graph)可以用我在WWW'11上发
表的方法进行加速。 |
|
k**o
发帖数: 8 | 7 兄弟居然还搞实名制啊，你做得不错啊。
目前有哪些作者用了你的knn 算法作spectral clustering 啊？
速度以及内存消耗跟michael jordan 一伙人2009年提出的算法比较怎么样？ |
|
D*****k
发帖数: 176 | 8 I don't think sparsigying the similarity matrix is a mainstream approach to
scale up spectral clustering. Low rank approximation like nystrom method
looks much more promising |
|
w*******g
发帖数: 99 | 9 Could anyone recommend some good books about collocation methods and
spectral methods? Introductory level will be better... Thank you! |
|
t****g
发帖数: 715 | 10 哪位有这样功能的matlab代码么：plot time series data on the spectral domain.
多谢 |
|
y****y
发帖数: 179 | 11 如果有一个随机信号f(t),该怎么求他的Power spectral density S(f)呢？
是直接fft(f(t))吗？ |
|
r********g
发帖数: 868 | 12 你好，我是2012年optical imaging方向毕业的博士，有做过一些spectral anaylysis
的工作，能否把摘要发给我，看看合不合适，谢谢 |
|
k*********g
发帖数: 791 | 13 这几天我一直忙着做网站。就你关于sem的评论说一下。
Discontinuous的目的好处：
1 flux control，这样有利于local conservation
2 有利于并行计算
DG, discontinuous galerkin, 在计算固体力学领域经常成为 DG FEM 的代名词，但是
，有限元做成 discontinuous 是没有太多意义的，性能提高很少，却麻烦了很多。为
啥，我有点忘了为啥，但这是我几年前下的结论。如果真的要用有限元，还是
classical的continuous finite element method性价比高。
在scientific computing领域，galerkin的定义比较混乱，经常这一波人说的galerkin
和那一波人说的galerkin不是同一个意思。我认为，galerkin这么定义比较canonical
：用best fitting方式、主体在transform space计算的称之为galerkin。
这个定义是照顾了spectral method community的习惯，有点anti有限元的，因为在有
限元，g... 阅读全帖 |
|
m******i
发帖数: 834 | 14 ☆─────────────────────────────────────☆
kennkqzhang (kenn) 于 (Mon Sep 27 16:20:44 2010, 美东) 提到:
即便在有限元擅长的regime、也恰好是性态很好的很容易求解的regime，如elliptic、
parabolic问题，也有一种数值方法比它快10~100倍；
让藕匪夷所思的是，为什么这么retard的东西会存在50+years。。。
☆─────────────────────────────────────☆
kennkqzhang (kenn) 于 (Sun Oct 31 06:47:05 2010, 美东) 提到:
一个产生的矩阵有几百个非零元素的数值方法；
比一个bandwidth只有几个元素的方法；
要慢10到100倍；
（一般来说）；
☆─────────────────────────────────────☆
kindlefan (maxwell) 于 (Sun Oct 31 14:34:49 2010, 美东) 提到:
请展开说说.
有限元的成熟... 阅读全帖 |
|
b*s
发帖数: 82482 | 15 http://skyserver.sdss.org/dr1/en/proj/advanced/galaxies/spectra
When you look at the spectrum of a galaxy, you are really looking at the
combination of spectra from the millions of stars in the galaxy. So studying
the features of a galaxy spectrum tells you about the types of stars the
galaxy contains, and the relative abundances of each type of star.
Galaxy spectra also clearly show you whether a galaxy contains star-forming
regions called HII regions. HII is a spectral emission line that corre... 阅读全帖 |
|
N**D
发帖数: 10322 | 16 所以说你还很年轻嘛
在多看看书，多看看paper, 就知道vapnic有多牛了，就知道为啥SVM这么厉害，
spectral clustering, 如果没记错的话，好像最后还是要用k-means吧。说白了，炒作。
类似的想法很早就有，只不过是这个大牛在某次nips提出来，就好象很火。这个大牛的
一个弟子，认为凡是和他们想法不同的就不是spectral clutsering， fourier
analysis 也不是spectral 的。
希望牛人们使劲在nips灌水，别到其它地方害人了。 |
|
N**D
发帖数: 10322 | 17 所以说你还很年轻嘛
在多看看书，多看看paper, 就知道vapnic有多牛了，就知道为啥SVM这么厉害，
spectral clustering, 如果没记错的话，好像最后还是要用k-means吧。说白了，炒作。
类似的想法很早就有，只不过是这个大牛在某次nips提出来，就好象很火。这个大牛的
一个弟子，认为凡是和他们想法不同的就不是spectral clutsering， fourier
analysis 也不是spectral 的。
希望牛人们使劲在nips灌水，别到其它地方害人了。 |
|
x****h
发帖数: 300 | 18 都是中文文献：
1. SPECTROSCOPY AND SPECTRAL ANALYSIS , volume 25, issue 5 2005 P 792-795
authors:Kang WJ, Liang SX, Jia LH, et al.
title: Determination of trace cadmium in environment water samples by
flame atomic absorption spectrometry using flow injection on-line
preconcentration with double microcolumns
2. SPECTROSCOPY AND SPECTRAL ANALYSIS, volume 26, issue 7, 2006 P 1345-1348
authors: Wang AX, Guo LP, Wu DM
title: Determination of Pb and Cd in environmental water sample by on-
line... 阅读全帖 |
|
k*********g
发帖数: 791 | 19 canonical definitions of galerkin methods and collocations methods.
the definition of galerkin methods vary in different communities of
scientific computing.
since spectral-based methods will be the future of numerical computation for
continuum, the canonical definition of galerkin sort of follows the bible "
C. Canuto, M. Y. Hussaini, A. Quarteroni, and T. A. Zang. Spectral Methods,
Fundamentals in Single Domains. Springer-Verlag, 2006":
if the core computation is conducted in the transformed s... 阅读全帖 |
|
c******k
发帖数: 1140 | 20 I tried to measure the spectral width of the pulse using OSA when
propagating a 1m fiber and a 3m fiber,respectively.
The spectral width@ FWHM is decreased to be 20nm with the 3m fiber compared
to the spectral width 32nm with the 1m fiber. |
|
B**e
发帖数: 60 | 21 Normal matrix.
p(A)<= ||A||, for any matrix norm ||.||
For normal matrix, the spectral norm is the same as the spectral radius of
the matrix. Thus, let ||.|| be the spectral norm, then
p(A+B)<=||A+B||<=||A||+||B||=p(A)+p(B)
If need more details, the textbook Matrix Analysis by Horn R A and Johnson C
R is recommended. |
|
c******k
发帖数: 1140 | 22 这种方法，脉冲得spectal width 是否变化。飞秒脉冲得spectral width 是24nm,经过
光线色散，变成皮秒脉冲，此时spectral width 是多少？是否根据time-bandwidth
product, 皮秒脉冲得spectral width 变小了？？ |
|
c******k
发帖数: 1140 | 23 我们有30nm的spectral width （FWHM）的 laser pulse, 经过一段光纤色散后，由于
色散脉冲宽度展宽了，请问这时的它的spectral width有变化吗？
后来自己也用光谱分析仪测了色散前和色散后的脉冲的spectral width. 结论是变窄了
（15nm）。但不清楚自己的测量是否正确。请牛人说说理论上是怎样的？谢谢啦 |
|
b***y
发帖数: 14281 | 24 既然有人争论方的学术水平，就上网搜了一下。应该说这个水平对于一个junior
faculty来说可以算是很牛B了，但是对于老方这种senior的已经搞了一辈子的人来说，
就只能算so so。不要说是院士，离aps fellow的级别也还差了不少。引用率最高的一
篇文章还是95年到了arizona之后跟老外合作的，被引118次，这也是方唯一一篇被引用
上百次的文章。可见方励之89之前确实没有做出过什么重要的工作，他的名声和地位绝
对是靠政治得来的，当然其中部分的也是80年在中国搞科普所得，必须承认科普工作也是
有重要的社会意义的。
(BTW，有人说你这个搜索未必完整。不错，确实不能保证100％毫无遗漏。但是我
用的search engine is THE search engine everybody in this community uses today.
So, if any article is not found by this search engine, sadly, it simply doesn't matter,
because no one would ever n... 阅读全帖 |
|
m**a
发帖数: 1840 | 25 【 以下文字转载自 EE 讨论区 】
发信人: mola (Super Baby), 信区: EE
标 题: 为极解密：如何看待华为拿下5G“短码”方案？
发信站: BBS 未名空间站 (Thu Dec 8 12:21:31 2016, 美东)
朋友大作，发在知乎。欢迎点赞和转发。
https://www.zhihu.com/question/52732376
美国当地时间11月17日凌晨0点45分，在刚刚结束的3GPP RAN1 87次会议的5G信道编码
方案讨论中，经过艰苦卓绝的努力和万分残酷的竞争，以中国华为公司主推的Polar
Code（极化码）方案，成为5G控制信道eMBB场景编码方案。
编者按
“华为极化码事件”内幕：
极化码获得的并非之前大肆炒作的“短码”。
极化码打败的对手并不是LDPC，而是在控制信道上取代4G现有技术TBCC。
若非极化码最终靠非技术手腕挤进控制信道，华为投资几十亿的5G“三神器”在5G NR
第一阶段业已全部打水漂。
整个5G信道编码又经历一番怎样波澜壮阔的争斗，这背后又隐藏着多少暗流汹涌的阴谋
诡计。“为极解密” 为您最全方位的解析。
为极解... 阅读全帖 |
|
C*S
发帖数: 1658 | 26 http://iceer.iem.cn/Prof.%20Shih-Chi_Liu.htm
Curriculum Vitae:
Summary
Dr. Shih-Chi Liu has been a charismatic and creative leader, leveraging
multilateral collaboration to increase the safety of our society and the
reliability of civil infrastructure and building international understanding
and goodwill among engineers and scientists throughout the world. In his
role as the Director for Earthquake Hazard Mitigation Program (1975-2001) at
the National Science Foundation (NSF), he initiated, dev... 阅读全帖 |
|
Z*****o
发帖数: 4090 | 27 http://www.stereophile.com/features/368/index.html
By Robert Harley • Posted: Nov 1, 1993
Not that long ago, digital audio was considered perfect if all the bits
could be stored and retrieved without data errors. If the data coming off
the disc were the same as what went on the disc, how could there be a sound-
quality difference with the same digital/analog converter? This "bits is
bits" mentality scoffs at sonic differences between CD transports, digital
interfaces, and CD tweaks. Becaus... 阅读全帖 |
|
U**********b
发帖数: 13 | 28 Two Post-doc Positions in Molecular Spectroscopy at the University of
Louisville and the Ohio State University.
Two post-doctoral positions in experimental and theoretical molecular
spectroscopy are available in the research groups of Prof. Jinjun Liu at the
University of Louisville (UofL) and Prof. Terry A. Miller at the Ohio State
University (OSU), respectively.
1. Post-doc position at UofL:
The first position is presently available in the laboratory of Dr. Jinjun
Liu https://sites.go... 阅读全帖 |
|
d**y
发帖数: 32 | 29 有两篇，请有兴趣且能审稿的站内信联系。
1：题目：Modeling and simulation an anti-reflective coating of ZnO and ZnS
for silicon solar cells using Silvaco software
摘要
In this paper simulated single layer anti-reflective coating on silicon
solar cell that based on the refractive index limit of silicon dioxide (SiO2
), zinc oxide (ZnO) and zinc sulphide (ZnS)
are presented. Two simulations of ZnO and ZnS coating were simulated to
compare with SiO2
anti-reflective coating on silicon solar cell surface. These simulations
car... 阅读全帖 |
|
m**a
发帖数: 1840 | 30 【 以下文字转载自 EE 讨论区 】
发信人: mola (Super Baby), 信区: EE
标 题: 为极解密：如何看待华为拿下5G“短码”方案？
发信站: BBS 未名空间站 (Thu Dec 8 12:21:31 2016, 美东)
朋友大作，发在知乎。欢迎点赞和转发。
https://www.zhihu.com/question/52732376
美国当地时间11月17日凌晨0点45分，在刚刚结束的3GPP RAN1 87次会议的5G信道编码
方案讨论中，经过艰苦卓绝的努力和万分残酷的竞争，以中国华为公司主推的Polar
Code（极化码）方案，成为5G控制信道eMBB场景编码方案。
编者按
“华为极化码事件”内幕：
极化码获得的并非之前大肆炒作的“短码”。
极化码打败的对手并不是LDPC，而是在控制信道上取代4G现有技术TBCC。
若非极化码最终靠非技术手腕挤进控制信道，华为投资几十亿的5G“三神器”在5G NR
第一阶段业已全部打水漂。
整个5G信道编码又经历一番怎样波澜壮阔的争斗，这背后又隐藏着多少暗流汹涌的阴谋
诡计。“为极解密” 为您最全方位的解析。
为极解... 阅读全帖 |
|
z****6
发帖数: 10776 | 31
这种胶片真没有对应的target
Because no standards exist to define the colored filter sets
that film scanners use to capture the red, green, and blue
information of the film image, each manufacturer’s scanner
has its own characteristic output. The output depends on the
scanner’s sensitivity to the dyes in the film. This sensitivity
is determined by the spectral distribution of the colored filter
sets and/or the spectral sensitivity of the
charge-coupled-device (CCD). |
|
k******a
发帖数: 2436 | 32 "绿色的光谱跟红色光谱重叠严重"你指的是human spectral sensitivity curve?
我提到的误解是认为整个光谱的数据都被测量，但实际上照相机的原理是只测量三原色
周围很窄的3个区间的能量。接近于human spectral sensitivity curve里面3个波峰。
根据实验证明只要reproduce这三个小区间的颜色就可以引起眼睛产生近似于原场景的
生物反应 |
|
k*********g
发帖数: 791 | 33 目前最优秀的dsem（disocntinuous spectral element method）是：
chebyshev collocation spectral；
但是藕相信还有一种更好的存在；
还需要做很多研究； |
|
x****n
发帖数: 43 | 34 Besides SNR, high field will increase spectral resolution, since 1) larger
spectral width for the same ppm range
and 2) J-coupling does not scale with main field. |
|
A******y
发帖数: 2041 | 35 "The author hopes that this incident inspires publishers, editors, reviewers
and authors to advance more effective measures to check and require
spectral data sets as a part of the publication process. In particular,
there is currently a lack of tools for presenting raw data files along with
publications." Xenobe Research Institute AKA James J. La Clair. Check your
spectral data synthetic chemists (at least look at the spectrum...with a
huge water peak!). |
|
U**********b
发帖数: 13 | 36 【 以下文字转载自 Postdoc 讨论区 】
发信人: UofLlaserlab (UofLlaserlab), 信区: Postdoc
标 题: Post-doc Positions in Molecular Spectroscopy
发信站: BBS 未名空间站 (Mon May 23 14:08:21 2016, 美东)
Two Post-doc Positions in Molecular Spectroscopy at the University of
Louisville and the Ohio State University.
Two post-doctoral positions in experimental and theoretical molecular
spectroscopy are available in the research groups of Prof. Jinjun Liu at the
University of Louisville (UofL) and Prof. Terry A. Miller at the Ohio Stat... 阅读全帖 |
|
j***h
发帖数: 4412 | 37 再看Northridge, CA(34.277N,-118.544W), For the 1994 Northridge earthquake,
the intensities MMI VI to IX cover the areas of damage. 然而据USGS的Uniform
Hazard Response Spetra, 其IX度区的设计基本地加速度为0.583，高于GB50011-2001
中最高的9度设防的设计基本地加速度0.40g，更是远远高于《GB50011-2008建筑抗震设
计规范局部修订》中地震局某些“专业”人士出于经济考虑，将此次地震灾区的最高设
防烈度比照唐山市设为八度的设计基本地加速度0.20g。
Conterminous 48 States
2003 NEHRP Seismic Design Provisions
Latitude = 34.277
Longitude = -118.544
Spectral Response Accelerations Ss and S1
Ss and S1 = Mapped Spectral Acceleratio |
|
w*******U
发帖数: 256 | 38 I'm a beginner of the 3D DNS with the pseudo-spectral method. I calculated
the freely decaying flow. If the nonlinear inertial force is set to be zero
then the velocity divergence (sum |k dot u_k|, where u_k is velocity in
spectral space) is zero; but if the inertial force (or Re) increases then
the divergence (sum |k dot u_k|) increases. Then I calculated the flow
driven by a prescribed forcing and the divergence is zero.
So it seems that it is more difficult to resolve the freely decaying flow... 阅读全帖 |
|
c*****i
发帖数: 51 | 39 then try spectral analysis first.
use some triangle kernel to smooth the spectrum estimation first.
Usually, I find the cycle identified by spectral analysis pretty accurate.
it is also strong to incorporate cross-sectional variation in the cycles. |
|
s**o
发帖数: 29 | 40 基本上，在无线通信中有两大类四种small-scale fading
1. Time dispersion due to "multipath delay spread"
a) flat fading
b) frequency selective fading
2. Frequency despersion due to "Doppler spread"
a) fast fading
b) slow fading
以上这两大类fading基本上是independent of one another
关于flat fading，通俗讲，就是传输信号的带宽小于信道的相干带宽，所以，信号
经过衰落信道后，在receiver处，信号的spectral characteristics依旧不变。
其对应的则是frequency selective fading，其信号的带宽大于信道的相干带宽，
所以，在RX处，接收信号在不同的频率其衰落特性也不同，即spectral characteristics
变了。
具体的，可以参考 Rappaport的"wireless com |
|
f*****g
发帖数: 30 | 41 It has been a point of debating whether OFDM provides high spectral
efficiency. For a well-controlled channel such as a telephone line, both
OFDM and single-carrier modulations can achieve spectral efficiencies close
to the Shannon bound. The advantage of OFDM is low complexity. But even
that is debatable depending on application conditions. For wireless
channels, there are many more issues. Probably OFDM is better in handling
some of them. But I doubt that in general, OFDM can achieve bet |
|
c******k
发帖数: 1140 | 42 比如10nm的spectral width的 laser pulse, 经过一段光纤色散后，脉冲宽度展宽了，
请问这时的它的spectral width有变化吗？ |
|
A*u
发帖数: 111 | 43 fiber的transmission在10nm之内应该可以看作是平的，
所以我觉得应该不会有spectral width的变化
但如果有非线性效应， 如supercontinuum
spectral width会变宽 |
|
c******k
发帖数: 1140 | 44 前几个星期，我发贴讨论“激光脉冲经过光纤色散后的频谱宽度变化吗？”非常感谢大
家的讨论和有价值的建议。现在就这个问题我把实验结果再次公布一下，因为以前的实
验有问题。
Pulse information (femtosecond fiber laser):
Spectral width: 30nm
Repetition rate:36MHz
Pulse width: 85fs
Average power of pulse: 45mW
Delta_Lambda=29.9nm after output from 1m of fiber, and Delta_Lambda=30.2nm
after output from a 11m of fiber. The spectral width of our pulse is
indepdent on the fiber length (SMF-28 fiber) without considering the
nonlinear effect of the fiber. The previous reulst for a reduced spe |
|
d*******n
发帖数: 369 | 45 For MP1 and MP2, 假设input frame is 512 samples. after it is fed into 32个
bandpass PQMF filter， the output is 512 spectral components grouped into 32
banks or 512 time-domain samples?
Since the input is time-domain samples, the output of the bandpass filters
should be also time-domain samples, right? If so, quantizer will take the
output of the bandpass filters as input but the quantizer needs the input to be spectral components!!! I didn't see an FFT between the filter bank and the quantizer that |
|
a**i
发帖数: 419 | 46 不是特别懂。但我理解这是用bandpass filter的响应来代替spectral components。
理论上它们携带的信息是等效的。
32
to be spectral components!!! I didn't see an FFT between the filter bank and
the quantizer that transforms the time-domain samples to frequency domain.
Confused??
time-domain input samples transformed to frequency via the bandpass filters?
bandpass filters so they are transformed to the frequency domain by the MDCT |
|
L*****e
发帖数: 169 | 47 Power spectral density and four transform are different.
The autocorrelation of white noise is actually delta funtion.
Power spectral density is the fourier transform of the autocorrelation
function.
Get it?
not because of phase. you are comparing apples to oranges. |
|
