how can i generate a White Gaussian Noise in Matlab 2009
in [Matlab]
|
Prev: A begginer's question - ButtonDownFcn
Next: load figure and then save with original figure's color??
From: Anyone on 26 May 2010 15:34 TideMan wrote on 26-May-10 03:49 ... > On May 26, 9:37 pm, "us " <u...(a)neurol.unizh.ch> wrote: >> TideMan <mul...(a)gmail.com> wrote in message <9267ac3c-bdc7-4582-b2d2-747404c2b...(a)y18g2000prn.googlegroups.com>... >>> On May 26, 9:22 pm, "Artur Racu" <arturr...(a)yahoo.com> wrote: >>>> "David Young" <d.s.young.notthis...(a)sussex.ac.uk> wrote in message <htio66$ch...(a)fred.mathworks.com>... >>>>> I don't know what Eurostag is, but in Matlab the function I would use to generate an approximation to white Gaussian noise is randn. >>>> Eurostag is a tool used for dynamic computation of power systems, i want to creat a signal(white gausian noise) so i can make a disturbance in power system and after to analyse different things; >>>> for example n=randn(10) if you take the mean of each column you will see that the mean is not zero(for white noise the main roule is t have the mean zero) >>>> could you please help me on that, >>>> thank you >>> Oh, don't be so silly!! >>> Of course the mean will be different from zero with only 10 numbers. >>> Now try this: >>> mean(randn(1000000,1)) >>> Is that close enough to zero for you? >>> If not, try this: >>> mean(randn(100000000,1)) >> but... derek, how mean(!)... >> >> mean(randn(100000000,1)) >> %{ >> ??? Error using ==> randn >> Out of memory. Type HELP MEMORY for your options. >> %} >> >> us > > Oh, Urs, you poor old thing. > You need to ask your employers to give you a better computer. > Here is what mine says: > >>> mean(randn(100000000,1)) > > ans = > > 2.1549e-005 >> mean(randn(100000000,1)) ans = -1.9690e-004 "neener-neener"
From: Bruno Luong on 26 May 2010 15:54 Anyone <""Zrqt\"@³@Qi§QUm![§H:\"&z25u/g`MpEr)\"{ETq[opZ\"Vz.az?"> wrote in message <4bfd77cd$0$32006$c3e8da3(a)news.astraweb.com>... > > > >> mean(randn(100000000,1)) Did someone mention this is a rich and slow way of doing >> randn/10000 Is there an obligation to run the expensive computer offered by the boss? Bruno
From: Walter Roberson on 26 May 2010 15:59 Bruno Luong wrote: > Anyone <""Zrqt\"@³@Qi§QUm![§H:\"&z25u/g`MpEr)\"{ETq[opZ\"Vz.az?"> wrote > in message <4bfd77cd$0$32006$c3e8da3(a)news.astraweb.com>... >> >> mean(randn(100000000,1)) > Did someone mention this is a rich and slow way of doing >>> randn/10000 No it isn't. By the Law Of Large Numbers, mean(randn(N,1)) for large N will have a much lower mean and standard deviation than randn(1,1)/sqrt(N) will.
From: Ken on 26 May 2010 17:09 > In theory, the mean of a set of numbers generated using RANDN will be 0. > In practice, it won't. > Here's a smaller example that may convince your chief that expecting exactly > the theoretical result in practice is not always a good idea. > > Take a six-sided die and roll it once. The mean value of the "sequence" of > numbers generated by that roll is whatever number came up on the die -- but > the theoretical mean is 3.5. That's not achievable with one roll of that > standard die. Even if you choose to roll that die a large odd number of > times, you will never be able to achieve the theoretical mean of that number > of rolls as all the numbers on the die are integers. > > If that argument doesn't work, then point him to Wikipedia (particularly the > second example on this page): > > http://en.wikipedia.org/wiki/Law_of_averages > > If nothing else that should occupy him for a few hours following links to > interesting topics :) > > -- > Steve Lord > sl...(a)mathworks.com > comp.soft-sys.matlab (CSSM) FAQ:http://matlabwiki.mathworks.com/MATLAB_FAQ > To contact Technical Support use the Contact Us link onhttp://www.mathworks.com hey, just to have fun being technical: in theory, the mean of a set of numbers generated using randn *is a random variable with a distribution centered at zero*, whose standard deviation tightens as you increase the size of the set. in the limit of infinite numbers in the set, the probability density function becomes a delta function at zero (probability 1). wouldn't be worth mentioning except you're implying a difference between theory and practice that doesn't exist. actually, in theory, you will NEVER get exactly zero with a finite set of numbers (because you're integrating the probability density function "from zero to zero"). artur: the true statement would be "for white noise, the EXPECTED mean is zero." also, if I'm understanding you right re: "gaussian curve values", use normpdf() or mvnpdf().
From: TideMan on 26 May 2010 17:20
I think I've come up with the perfect sop for Artur's boss who doesn't like the mean of 10 random numbers not being exactly zero. round(mean(randn(10,1))) |