| Author |
 Topic  |
|
|
guardian_angel
5 Posts |
 Posted - 03/05/2012 : 6:09:01 PM
|
Origin Ver. and Service Release (Select Help-->About Origin): 8.0724
Operating System: Vista x64
I'd like to fit this function:
y=510/sqrt(510^2+(x*a-1/0,000000001*x)^2)
Are there any errors in this equation? I have these settings:
Function Type: User-Defined
Independent variables: x
Dependent variables: y
Parameter names: a
Function Form: Origin C
Compilation is okay.
After saving when I try to fit this function, there is nothing going on, I receive empty table.
"Fit till converge" says "Fit did not converge - reason unknown."
How am I supposed to fit this equation?
|
|
|
Hideo Fujii
USA
1582 Posts |
Posted - 03/06/2012 : 10:00:24 AM
|
Hi,
> I'd like to fit this function:
> y=510/sqrt(510^2+(x*a-1/0,000000001*x)^2)
> Function Type: User-Defined
> Independent variables: x
> Dependent variables: y
> Parameter names: a
> Function Form: Origin C
> Compilation is okay.
I suggest you to check the following:
1) Because this is in OriginC type, use the 510.0(real) instead of 510(integer).
2) As possibly comma(,) as a decimal point has not been set, try 0.000000001 instead of 0,000000001 .
3) For user-defined function as it doesn't have the parameter initialization routine, always make sure that before you start fitting, any parameter (in your case only a) shouldn't be a missing value, and you must give an appropriate initial value.
Hope this helps.
--Hideo Fujii
OriginLab |
 |
|
|
Drbobshepherd
USA
Posts |
Posted - 03/06/2012 : 10:26:10 AM
|
Check your function. I do not think it is calculating the way you think it is. The rules of algebreic hierarchy dictate that the inner-most parenthetical expression will be calculated first,
(x*a-1/0,000000001*x).
The products and ratios will be evaluated from left to right, then the difference will be taken. So your inner-most expression will be
(xa-(1,0E9)x).
I suspect this is not the result you intended as 1/0,000000001 is an odd way of writing (1,0E9). As it stands, unless a is close to a billion, your formula is dominated by the second term in the inner-most expression.
DrBobShepherd |
|
|
|
guardian_angel
5 Posts |
Posted - 03/06/2012 : 12:29:39 PM
|
Thanks for quick replies but any of these hints work 
This is for my physics report. I need to find amplitude-frequency chracteristics for series RLC circuit.
There is no problem at all with graph, it looks like in the picture
There is transmission coefficient (alfa) on y-axis, and frequency on x-axis. Now I need to do the fitting:
alfa = R/sqrt(R^2+(wL-1/wc)^2)
R = 510 ohm
C = 1*10-9 F
w = frequency (x)
L is parameter which I want to evaluate by fitting the function.
How would you do this? |
Edited by - guardian_angel on 03/06/2012 12:31:18 PM |
|
|
|
Hideo Fujii
USA
1582 Posts |
Posted - 03/06/2012 : 2:43:13 PM
|
Hi,
It seems the parameter L only controls ONLY the peak position with the function:
alfa = R/sqrt(R^2+(w*L-(1/(w*c)))^2)
It seems that you need at least one more parameter to control the width of the peak. (Screenshot with L=0.10294)
--Hideo Fujii
OriginLab |
|
|
|
guardian_angel
5 Posts |
Posted - 03/06/2012 : 4:02:57 PM
|
| So maybe there is some function that will only estimate the peak? This would be enough to calculate inductance L |
|
|
|
Hideo Fujii
USA
1582 Posts |
Posted - 03/06/2012 : 5:55:38 PM
|
such as fitting by Gauss with the bias term(y0)=0 fixed after converting to log values of W(X)??
--Hideo Fujii
OriginLab |
|
|
| |
Topic |
|
Problem with defined function