Fit problem
in [Matlab]
|
From: Francesco on 1 Jul 2010 06:09 I'm using the following code to compute the King's Law coefficient (hot-wire probe calibration law). << start = [1;0.5;0.4]; options = fitoptions('Method','NonlinearLeastSquares','Display','iter','TolFun',10^-10,'TolX',10^-10,'StartPoint',start); f = fittype('sqrt(A+B*x^n)'); res1 = fit(U,E_corr_w1,f,options) >> The problem is that (after successfully using the code for a while), now the following error pops up: ??? Complex value computed by model function. Error in ==> fit at 437 errstr = handleerr( errid, errmsg, suppresserr ); while if using the 'cftool' with the same data set, graphically computing the coefficients everything works fine!! I can't really see where's the problem. Thanks in advance for any help. Best regards.
From: us on 1 Jul 2010 07:02 "Francesco " <francesco.baldani(a)rma.ac.be> wrote in message <i0hpg1$5ai$1(a)fred.mathworks.com>... > I'm using the following code to compute the King's Law coefficient (hot-wire probe calibration law). > > << > start = [1;0.5;0.4]; > > options = fitoptions('Method','NonlinearLeastSquares','Display','iter','TolFun',10^-10,'TolX',10^-10,'StartPoint',start); > > f = fittype('sqrt(A+B*x^n)'); > > res1 = fit(U,E_corr_w1,f,options) > >> > > The problem is that (after successfully using the code for a while), now the following error pops up: > > ??? Complex value computed by model function. > > Error in ==> fit at 437 > errstr = handleerr( errid, errmsg, suppresserr ); > > while if using the 'cftool' with the same data set, graphically computing the coefficients everything works fine!! > > I can't really see where's the problem. > > Thanks in advance for any help. > > Best regards. aree you sure you're using the very same options in both cases(?)... us
From: Francesco on 1 Jul 2010 07:29 if the default values for "Lower" and "Upper" (limits vectors) are -inf and inf then yes. Since I'm using also the same start values.... "us " <us(a)neurol.unizh.ch> wrote in message <i0hsjc$iju$1(a)fred.mathworks.com>... > "Francesco " <francesco.baldani(a)rma.ac.be> wrote in message <i0hpg1$5ai$1(a)fred.mathworks.com>... > > I'm using the following code to compute the King's Law coefficient (hot-wire probe calibration law). > > > > << > > start = [1;0.5;0.4]; > > > > options = fitoptions('Method','NonlinearLeastSquares','Display','iter','TolFun',10^-10,'TolX',10^-10,'StartPoint',start); > > > > f = fittype('sqrt(A+B*x^n)'); > > > > res1 = fit(U,E_corr_w1,f,options) > > >> > > > > The problem is that (after successfully using the code for a while), now the following error pops up: > > > > ??? Complex value computed by model function. > > > > Error in ==> fit at 437 > > errstr = handleerr( errid, errmsg, suppresserr ); > > > > while if using the 'cftool' with the same data set, graphically computing the coefficients everything works fine!! > > > > I can't really see where's the problem. > > > > Thanks in advance for any help. > > > > Best regards. > > aree you sure you're using the very same options in both cases(?)... > > us
From: Francesco on 1 Jul 2010 07:35 PROBLEM SOLVED up = [inf;inf;inf]; low = [-inf;-inf;-inf]; options = fitoptions('Method','NonlinearLeastSquares','Upper',up,'Lower',low,'Display','iter','TolFun',10^-10,'TolX',10^-10,'StartPoint',start); I added upper and lower limits and set them to +inf and -inf respectively and it works... Anyway I can't see why it was working before without specifying them and then it suddenly started giving troubles... thanks. "Francesco " <francesco.baldani(a)rma.ac.be> wrote in message <i0hu60$q74$1(a)fred.mathworks.com>... > if the default values for "Lower" and "Upper" (limits vectors) are -inf and inf then yes. Since I'm using also the same start values.... > > > > "us " <us(a)neurol.unizh.ch> wrote in message <i0hsjc$iju$1(a)fred.mathworks.com>... > > "Francesco " <francesco.baldani(a)rma.ac.be> wrote in message <i0hpg1$5ai$1(a)fred.mathworks.com>... > > > I'm using the following code to compute the King's Law coefficient (hot-wire probe calibration law). > > > > > > << > > > start = [1;0.5;0.4]; > > > > > > options = fitoptions('Method','NonlinearLeastSquares','Display','iter','TolFun',10^-10,'TolX',10^-10,'StartPoint',start); > > > > > > f = fittype('sqrt(A+B*x^n)'); > > > > > > res1 = fit(U,E_corr_w1,f,options) > > > >> > > > > > > The problem is that (after successfully using the code for a while), now the following error pops up: > > > > > > ??? Complex value computed by model function. > > > > > > Error in ==> fit at 437 > > > errstr = handleerr( errid, errmsg, suppresserr ); > > > > > > while if using the 'cftool' with the same data set, graphically computing the coefficients everything works fine!! > > > > > > I can't really see where's the problem. > > > > > > Thanks in advance for any help. > > > > > > Best regards. > > > > aree you sure you're using the very same options in both cases(?)... > > > > us
|
Pages: 1 Prev: Replacing a NaN by it nanmean Next: Calling .dll library: Method was not found |