On Dec 29, 12:59�pm, charanchar <charanc...@gmail.com> wrote:
> On Dec 29, 10:43�pm, Oli Charlesworth <ca...@olifilth.co.uk> wrote:
>
>
>
> > charanchar wrote:
> > > On Dec 29, 10:22 pm, Tim Wescott <t...@seemywebsite.com> wrote:
> > >> charanchar wrote:
> > >>> On Dec 29, 10:13 pm, Greg Berchin <gberc...@comicast.net.invalid>
> > >>> wrote:
> > >>>>> have u got any solution for getting gaussian noise(100 samples) with
> > >>>>> exactly zero mean and variance=1 in matlab?
> > >>>> help randn
> > >>> it will not give the random noise with mean=0 and var=1. mean and var
> > >>> will change as seed value changes. they are not fixed
> > >> I think you are confusing a random process with a sample thereof.
>
> > >> If you take a Gaussian random process with mean = 0 and variance = 1,
> > >> and take 100 samples, you will _not_ get sample sets with mean = 0 and
> > >> variance = 1. �What you'll get is samples whose means and variances are
> > >> themselves random variables, with _expected_ values of 0 and 1,
> > >> respectively.
>
> > >> A Gaussian process that is guaranteed to generate a mean = 0 and
> > >> variance = 1 for every 100 sample set would not be i.i.r., if it were
> > >> even possible in general.
>
> > >> --
>
> > >> Tim Wescott
> > >> Wescott Design Serviceshttp://www.wescottdesign.com
>
> > >> Do you need to implement control loops in software?
> > >> "Applied Control Theory for Embedded Systems" gives you just what it says.
> > >> See details athttp://www.wescottdesign.com/actfes/actfes.html
> > > thank u Tim, that is if we run the process several times, then the
> > > average of all means will be zero and var will be 1. am i correct?
>
> > No. �You can only guarantee a particular mean and variance by generating
> > an infinite number of samples.
>
> > However, given a set of 100 samples, you can normalise them to have zero
> > sample-mean and unit sample-variance.
>
> > --
> > Oli
>
> yes that is correct.
> my problem is for phase noise correction algorithm in ofdm system, i
> want to introduce phase noise as wiener process. for example if we set
> variance(using randn in matlab) as .1 that var is changing from symbol
> to symbol by + or - .02. this small change will affect the estimation
> badly. becoz the ofdm is more sensitive to phase noise
On Dec 29, 11:52�am, charanchar <charanc...@gmail.com> wrote:
> have u got any solution for getting gaussian noise(100 samples) with
> exactly zero mean and variance=1 in matlab?
If you have 100 samples of actual Gaussian noise, you can easily force
the mean and variance of the 100 samples to be exactly 0 and 1
respectively. It is a trivial problem.
Dirk
Reply by Oli Charlesworth●December 29, 20082008-12-29
charanchar wrote:
> On Dec 29, 10:43 pm, Oli Charlesworth <ca...@olifilth.co.uk> wrote:
>> charanchar wrote:
>>> On Dec 29, 10:22 pm, Tim Wescott <t...@seemywebsite.com> wrote:
>>>> charanchar wrote:
>>>>> On Dec 29, 10:13 pm, Greg Berchin <gberc...@comicast.net.invalid>
>>>>> wrote:
>>>>>>> have u got any solution for getting gaussian noise(100 samples) with
>>>>>>> exactly zero mean and variance=1 in matlab?
>>>>>> help randn
>>>>> it will not give the random noise with mean=0 and var=1. mean and var
>>>>> will change as seed value changes. they are not fixed
>>>> I think you are confusing a random process with a sample thereof.
>>>> If you take a Gaussian random process with mean = 0 and variance = 1,
>>>> and take 100 samples, you will _not_ get sample sets with mean = 0 and
>>>> variance = 1. What you'll get is samples whose means and variances are
>>>> themselves random variables, with _expected_ values of 0 and 1,
>>>> respectively.
>>>> A Gaussian process that is guaranteed to generate a mean = 0 and
>>>> variance = 1 for every 100 sample set would not be i.i.r., if it were
>>>> even possible in general.
>>>> --
>>>> Tim Wescott
>>>> Wescott Design Serviceshttp://www.wescottdesign.com
>>>> Do you need to implement control loops in software?
>>>> "Applied Control Theory for Embedded Systems" gives you just what it says.
>>>> See details athttp://www.wescottdesign.com/actfes/actfes.html
>>> thank u Tim, that is if we run the process several times, then the
>>> average of all means will be zero and var will be 1. am i correct?
>> No. You can only guarantee a particular mean and variance by generating
>> an infinite number of samples.
>>
>> However, given a set of 100 samples, you can normalise them to have zero
>> sample-mean and unit sample-variance.
>>
>> --
>> Oli
>
> yes that is correct.
> my problem is for phase noise correction algorithm in ofdm system, i
> want to introduce phase noise as wiener process. for example if we set
> variance(using randn in matlab) as .1 that var is changing from symbol
> to symbol by + or - .02. this small change will affect the estimation
> badly. becoz the ofdm is more sensitive to phase noise
Not sure I understand. randn() should generate a stationary sequence,
i.e. the variance of each sample will be the same. What do you mean
when you say that the "var is changing from symbol to symbol"? How are
you measuring this?
--
Oli
Reply by charanchar●December 29, 20082008-12-29
On Dec 29, 10:43�pm, Oli Charlesworth <ca...@olifilth.co.uk> wrote:
> charanchar wrote:
> > On Dec 29, 10:22 pm, Tim Wescott <t...@seemywebsite.com> wrote:
> >> charanchar wrote:
> >>> On Dec 29, 10:13 pm, Greg Berchin <gberc...@comicast.net.invalid>
> >>> wrote:
> >>>>> have u got any solution for getting gaussian noise(100 samples) with
> >>>>> exactly zero mean and variance=1 in matlab?
> >>>> help randn
> >>> it will not give the random noise with mean=0 and var=1. mean and var
> >>> will change as seed value changes. they are not fixed
> >> I think you are confusing a random process with a sample thereof.
>
> >> If you take a Gaussian random process with mean = 0 and variance = 1,
> >> and take 100 samples, you will _not_ get sample sets with mean = 0 and
> >> variance = 1. �What you'll get is samples whose means and variances are
> >> themselves random variables, with _expected_ values of 0 and 1,
> >> respectively.
>
> >> A Gaussian process that is guaranteed to generate a mean = 0 and
> >> variance = 1 for every 100 sample set would not be i.i.r., if it were
> >> even possible in general.
>
> >> --
>
> >> Tim Wescott
> >> Wescott Design Serviceshttp://www.wescottdesign.com
>
> >> Do you need to implement control loops in software?
> >> "Applied Control Theory for Embedded Systems" gives you just what it says.
> >> See details athttp://www.wescottdesign.com/actfes/actfes.html
> > thank u Tim, that is if we run the process several times, then the
> > average of all means will be zero and var will be 1. am i correct?
>
> No. �You can only guarantee a particular mean and variance by generating
> an infinite number of samples.
>
> However, given a set of 100 samples, you can normalise them to have zero
> sample-mean and unit sample-variance.
>
> --
> Oli
yes that is correct.
my problem is for phase noise correction algorithm in ofdm system, i
want to introduce phase noise as wiener process. for example if we set
variance(using randn in matlab) as .1 that var is changing from symbol
to symbol by + or - .02. this small change will affect the estimation
badly. becoz the ofdm is more sensitive to phase noise
Reply by Jerry Avins●December 29, 20082008-12-29
charanchar wrote:
> On Dec 29, 10:22 pm, Tim Wescott <t...@seemywebsite.com> wrote:
>> charanchar wrote:
>>> On Dec 29, 10:13 pm, Greg Berchin <gberc...@comicast.net.invalid>
>>> wrote:
>>>>> have u got any solution for getting gaussian noise(100 samples) with
>>>>> exactly zero mean and variance=1 in matlab?
>>>> help randn
>>> it will not give the random noise with mean=0 and var=1. mean and var
>>> will change as seed value changes. they are not fixed
>> I think you are confusing a random process with a sample thereof.
>>
>> If you take a Gaussian random process with mean = 0 and variance = 1,
>> and take 100 samples, you will _not_ get sample sets with mean = 0 and
>> variance = 1. What you'll get is samples whose means and variances are
>> themselves random variables, with _expected_ values of 0 and 1,
>> respectively.
>>
>> A Gaussian process that is guaranteed to generate a mean = 0 and
>> variance = 1 for every 100 sample set would not be i.i.r., if it were
>> even possible in general.
>>
>> --
>>
>> Tim Wescott
>> Wescott Design Serviceshttp://www.wescottdesign.com
>>
>> Do you need to implement control loops in software?
>> "Applied Control Theory for Embedded Systems" gives you just what it says.
>> See details athttp://www.wescottdesign.com/actfes/actfes.html
> thank u Tim, that is if we run the process several times, then the
> average of all means will be zero and var will be 1. am i correct?
No. The average of the means and vars will tend toward 0 and 1 respectively.
Jerry
--
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������
Reply by Oli Charlesworth●December 29, 20082008-12-29
charanchar wrote:
> On Dec 29, 10:22 pm, Tim Wescott <t...@seemywebsite.com> wrote:
>> charanchar wrote:
>>> On Dec 29, 10:13 pm, Greg Berchin <gberc...@comicast.net.invalid>
>>> wrote:
>>>>> have u got any solution for getting gaussian noise(100 samples) with
>>>>> exactly zero mean and variance=1 in matlab?
>>>> help randn
>>> it will not give the random noise with mean=0 and var=1. mean and var
>>> will change as seed value changes. they are not fixed
>> I think you are confusing a random process with a sample thereof.
>>
>> If you take a Gaussian random process with mean = 0 and variance = 1,
>> and take 100 samples, you will _not_ get sample sets with mean = 0 and
>> variance = 1. What you'll get is samples whose means and variances are
>> themselves random variables, with _expected_ values of 0 and 1,
>> respectively.
>>
>> A Gaussian process that is guaranteed to generate a mean = 0 and
>> variance = 1 for every 100 sample set would not be i.i.r., if it were
>> even possible in general.
>>
>> --
>>
>> Tim Wescott
>> Wescott Design Serviceshttp://www.wescottdesign.com
>>
>> Do you need to implement control loops in software?
>> "Applied Control Theory for Embedded Systems" gives you just what it says.
>> See details athttp://www.wescottdesign.com/actfes/actfes.html
> thank u Tim, that is if we run the process several times, then the
> average of all means will be zero and var will be 1. am i correct?
No. You can only guarantee a particular mean and variance by generating
an infinite number of samples.
However, given a set of 100 samples, you can normalise them to have zero
sample-mean and unit sample-variance.
--
Oli
Reply by charanchar●December 29, 20082008-12-29
On Dec 29, 10:22�pm, Tim Wescott <t...@seemywebsite.com> wrote:
> charanchar wrote:
> > On Dec 29, 10:13 pm, Greg Berchin <gberc...@comicast.net.invalid>
> > wrote:
> >>> have u got any solution for getting gaussian noise(100 samples) with
> >>> exactly zero mean and variance=1 in matlab?
> >> help randn
>
> > it will not give the random noise with mean=0 and var=1. mean and var
> > will change as seed value changes. they are not fixed
>
> I think you are confusing a random process with a sample thereof.
>
> If you take a Gaussian random process with mean = 0 and variance = 1,
> and take 100 samples, you will _not_ get sample sets with mean = 0 and
> variance = 1. �What you'll get is samples whose means and variances are
> themselves random variables, with _expected_ values of 0 and 1,
> respectively.
>
> A Gaussian process that is guaranteed to generate a mean = 0 and
> variance = 1 for every 100 sample set would not be i.i.r., if it were
> even possible in general.
>
> --
>
> Tim Wescott
> Wescott Design Serviceshttp://www.wescottdesign.com
>
> Do you need to implement control loops in software?
> "Applied Control Theory for Embedded Systems" gives you just what it says.
> See details athttp://www.wescottdesign.com/actfes/actfes.html
thank u Tim, that is if we run the process several times, then the
average of all means will be zero and var will be 1. am i correct?
Reply by Tim Wescott●December 29, 20082008-12-29
charanchar wrote:
> On Dec 29, 10:13 pm, Greg Berchin <gberc...@comicast.net.invalid>
> wrote:
>>> have u got any solution for getting gaussian noise(100 samples) with
>>> exactly zero mean and variance=1 in matlab?
>> help randn
>
> it will not give the random noise with mean=0 and var=1. mean and var
> will change as seed value changes. they are not fixed
I think you are confusing a random process with a sample thereof.
If you take a Gaussian random process with mean = 0 and variance = 1,
and take 100 samples, you will _not_ get sample sets with mean = 0 and
variance = 1. What you'll get is samples whose means and variances are
themselves random variables, with _expected_ values of 0 and 1,
respectively.
A Gaussian process that is guaranteed to generate a mean = 0 and
variance = 1 for every 100 sample set would not be i.i.r., if it were
even possible in general.
--
Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by Oli Charlesworth●December 29, 20082008-12-29
charanchar wrote:
> On Dec 29, 10:13 pm, Greg Berchin <gberc...@comicast.net.invalid>
> wrote:
>>> have u got any solution for getting gaussian noise(100 samples) with
>>> exactly zero mean and variance=1 in matlab?
>> help randn
>
> it will not give the random noise with mean=0 and var=1. mean and var
> will change as seed value changes. they are not fixed
The seed doesn't affect the statistics.
--
Oli
Reply by charanchar●December 29, 20082008-12-29
On Dec 29, 10:13�pm, Greg Berchin <gberc...@comicast.net.invalid>
wrote:
> >have u got any solution for getting gaussian noise(100 samples) with
> >exactly zero mean and variance=1 in matlab?
>
> help randn
it will not give the random noise with mean=0 and var=1. mean and var
will change as seed value changes. they are not fixed