| Author |
 Topic  |
|
|
hgibhar
Germany
12 Posts |
 Posted - 09/13/2013 : 11:07:26 AM
|
Hello,
I try to do global fitting with shared parameter on a huge amount of datasets (40 at one time). Using this in the normal dialog does the work, but is slow. Hence, I tried to do Labtalk programming, letting the computer the calculations for the night. This becomes more and more slowly. Hence, I tried to take OriginC. A test with a 3 datasets shows, that this will be quite quickly and might work over night, too.
THe question now is: How can I produce a result report worksheet exactly similar to the normal fitting dialogbox? Everything is done there: the graphs, parameters and errors, statistics. Very well presented form. It is unnecessary, to invent this anew!
Here som code snippets:
Programmatically getting the input data:
---------------
XYRange drInput;
int i;
for(i=1;i<4;i++){
drInput.Add(wks, 0, "X");
drInput.Add(wks, i, "Y");
if(i<3) drInput.Add();
}
vector vX1[3], vY1[3];
for(i=0;i<3;i++){
drInput.GetData(vY1[i], vX1[i], NULL, i);
}
---------------
NLFitSession nlfSession;
// Set Function: a self generated one (in OriginC)
if ( !nlfSession.SetFunction("vgt1") )
ERR_RETURN("Fail to set function!");
for(i=0;i<2;i++){
// Set Data and only need to set the last dataset with global fit mode, as to allow share parameters
if ( !nlfSession.SetData(vY1[i], vX1[i], NULL, i, 3) )
ERR_RETURN("Fail to set data for the first dataset!")
}
if ( !nlfSession.SetData(vY1[2], vX1[2], NULL, 2, 3, DATA_MODE_GLOBAL) )
ERR_RETURN("Fail to set data for the last dataset!")
// Parameter initialization
if ( !nlfSession.ParamsInitValues() )
ERR_RETURN("Fail to init parameters values!")
// Share xc parameter
int nSharedParamIndex = 2; // 1, the index of xc in function, must be DATA_MODE_GLOBAL datamode, or else will fail to share parameters
nlfSession.SetParamShare(nSharedParamIndex);
----------Fitting
bool bRet = nlfSession.Fit(&nFitOutcome); //if success
------------------
Now, there should be something like: "Make Results Report like Origin"
Thank you very much for any hint!
Holger
PS: It seems, that the functions
GetFitResultsStats(RegStats* pRegStats, NLSFFitInfo* pFitInfo, bool bPreDataset = false, int nSetIndex = 0, int nDepIndex = 0)
and
GetFitResultsParams(FitParameter* pFitParameter, RegStats *pstRegStats = NULL, int nDataset = 0, bool bIncDerivedParams = false)
do not work with global fitting. THe Compiler always says "member function NLFitSession::GetFitResultsParams cannot be found".
The function GetFitResultsParams(vector& vParams, vector& vErrors) works.
Origin Ver. 9.0.0 32 bit Operating System: Win 7 |
|
|
Penn
China
644 Posts |
|
|
hgibhar
Germany
12 Posts |
Posted - 09/30/2013 : 06:31:22 AM
|
Hello Penn,
thank you very much for your reply. I have just read the links you sent. The main question was, if there is a simple way to produce the same report as Origin. This seems not to be the case with respect to your mail.
A second remark concerned GetFitResultsStats and GetFitResultsParams. I wrote, that with global fitting the compiler produces an error using GetFitResultsParams(FitParameter* pFitParameter, RegStats *pstRegStats = NULL, int nDataset = 0, bool bIncDerivedParams = false). Instead of using this, the other constructor GetFitResultsParams(vector& vParams, vector& vErrors) works. THis is not mentioned in the OriginC help. And one looks for the mistake, which is in the code (and will not find it).
Thank you for your reply.
Holger |
 |
|
|
Penn
China
644 Posts |
Posted - 10/11/2013 : 05:37:08 AM
|
Hi,
I am afraid there is no simple way to generate the same report as Origin in Origin C.
About the methods you mentioned, could you please try the example in this page to see whether it works or not?
Penn |
|
|
| |
Topic |
|
|
|
Global fitting