| Author |
 Topic  |
|
|
nejo
Japan
6 Posts |
 Posted - 01/17/2007 : 02:45:08 AM
|
Origin Version (Select Help-->About Origin): 7.5
Operating System:WindowsXP
A correlation is obtained on a single spctrum line using for example FFT. It is required on the otherhand, to get correlation among several spectra lines which have diffent frequency chratcteristics. Could anyone suggest how I can get this kind of correlation usng Orign?
Hitoshi Nejo |
|
|
Deanna
China
Posts |
Posted - 01/17/2007 : 03:26:30 AM
|
Hi Hitoshi Nejo.
Origin 7.5 should be able to perform correlation to any two columns in a worksheet. I am not very sure about what your problem is. Would you please further explain?
Deanna
OriginLab Technical Services |
 |
|
|
nejo
Japan
6 Posts |
Posted - 01/17/2007 : 7:18:58 PM
|
Hi Deanna,
Thank you for your reply.
If you mean single line by two columns, then yes, I could already do somthing. A single line is drawn using two columns for two axes as a matter of course. If my understanding of the two clomuns is correct, then what I would like to perform is on more than three columns. Namely several lines are drawn in a single graph sheet and the correlation among those several lines should be gotten.
Hope to clary my proble encounterd.
Hitoshi
|
|
|
|
Deanna
China
Posts |
Posted - 01/17/2007 : 10:08:52 PM
|
Thank you for your explanation.
What I mean by "two columns" is actually "two Y columns", which means that you can plot two lines with these two columns.
Usually, correlation is used for detecting similarities between two signals, which can be expressed by two Y columns. That's why the correlation tool in Origin is designed to operate on two Y columns.
It seems that you have more than two signals. However, you can only calculate correlation between two of them each time you use the correlation tool in Origin. I am sorry if this design causes any inconvenience.
Deanna
OriginLab Technical Services |
|
|
|
nejo
Japan
6 Posts |
Posted - 01/18/2007 : 11:53:43 PM
|
Thank you for your instructions.
Although I have tried to make correlation between two columns ( be sure that in addition to x column) through , the Analysis -> Correlation menue, the prgrams issue an error message "select two colomns".
If I choose Analysis->FFT, then it gives a result. What is the probelem and also what is the difference between Correlation and FFT is this repect?
Thanks in advance. Hitoshi
|
|
|
|
nejo
Japan
6 Posts |
Posted - 01/22/2007 : 03:19:50 AM
|
Hi Kitkat,
Thank you.
You description explaines very well what I want to know.
Now coming back to my original question of getting a correlation between more than three lines.
Deanna already replied that the current Origin does not have capability of that.
But is there any mathematics treating in that direction? Also is there any idea for implimenting such process using a macro or somthing? Except rigorouse treatment, if some can give a flavour, that is fine for the moment. Thanks in advance.
Hitoshi
|
|
|
|
nejo
Japan
6 Posts |
Posted - 01/24/2007 : 12:58:59 AM
|
Hi Kitkit,
Thank you for your suggestion and description.
I will try.
Another question on algorithm of the correlation implemented in Origin;
Does Origin take into account the vertical valuse for getting correlation? In my understanding, FFT requires only the phase information not amplitude. If Origin uses the FFT technique for getting a correlation, then it does not use the information of amplitude? The reason I ask you this is that I may make correlations among rather linear -like lines not oscillating lines. In other word, does Origin give a correct answer of a correlation for liner behavior lines?
THanks in advance.
Hitoshi
|
|
|
|
Deanna
China
Posts |
Posted - 01/24/2007 : 01:31:10 AM
|
quote:
In my understanding, FFT requires only the phase information not amplitude. If Origin uses the FFT technique for getting a correlation, then it does not use the information of amplitude?
Hi. Actually, I believe it is not like what you think. Amplitude is very useful for calculating correlation.
The correlation algorithm we used is based on the Correlation Theorem, which states that the correlation of two signals, f & g, is equal to:
ifft(fft(f).fft(g))
where ifft() denotes the inverse Fourier transform while fft() dentoes the forward Fourier transform.
You may also review the FFT CORRELATION.OPJ project located in your Origin \SAMPLES\ANALYSIS\FFT folder.
Deanna
OriginLab Technical Services |
|
|
| |
Topic |
|
Correlation