Error using integral; First input argument must be a function handle.

3 vues (au cours des 30 derniers jours)
Bibek Dhami
Bibek Dhami le 28 Juin 2022
Commenté : Torsten le 30 Juin 2022
I am trying to fit my equation using the modelE which is the sum of two functions exciton and continum. Continum is the integral of function cont. I am having error while using matlab inbuilt integral function. Can someone guide me where I am messing up on this code?
A=xlsread('Data');
xdata=1240./A(:,1);
ydata=A(:,2);
plot(xdata,ydata,'r.','markersize',30)
box on
energy=xdata;
Absorption=ydata;
options =optimoptions(@lsqcurvefit,'Algorithm','trust-region-reflective',...
'StepTolerance',1e-19,'MaxFunctionEvaluations',1e10, ...
'MaxIterations',1e3)
options =
lsqcurvefit options: Options used by current Algorithm ('trust-region-reflective'): (Other available algorithms: 'levenberg-marquardt') Set properties: Algorithm: 'trust-region-reflective' MaxFunctionEvaluations: 1.0000e+10 MaxIterations: 1000 StepTolerance: 1.0000e-19 Default properties: CheckGradients: 0 Display: 'final' FiniteDifferenceStepSize: 'sqrt(eps)' FiniteDifferenceType: 'forward' FunctionTolerance: 1.0000e-06 JacobianMultiplyFcn: [] OptimalityTolerance: 1.0000e-06 OutputFcn: [] PlotFcn: [] SpecifyObjectiveGradient: 0 SubproblemAlgorithm: 'factorization' TypicalX: 'ones(numberOfVariables,1)' UseParallel: 0
var=[0.0529 2.3926 0.0307 1.002];
MI=lsqcurvefit(@modelE,var,energy,Absorption,[],[],options);
Local minimum possible. lsqcurvefit stopped because the final change in the sum of squares relative to its initial value is less than the value of the function tolerance.
fitdata=modelE(MI,energy);
hold on
plot(energy,modelE(var,energy),'co')
Warning: Imaginary parts of complex X and/or Y arguments ignored.
plot(energy,exciton(var,energy),'m.','markersize',30)
plot(energy, continum(var,energy),'k.')
Error using integral
First input argument must be a function handle.

Error in solution>continum (line 40)
z = integral(cont(var,energy),2.39,inf);
function alpha = modelE(var,energy)
alpha = cont(var,energy) +var(2).* exciton(var,energy);
end
function x= exciton(var,energy)
x= 2.*var(1).*sech((energy-var(2)+var(1))./var(3))+...
0.125.*var(1).*sech((energy-var(2)+(var(1)./8))./var(3))+...
(2/27).*var(1).*sech((energy-var(2)+(var(1)./27))./var(3));
end
function y= cont(var,energy)
y=(sech((energy-var(2))./var(3))).*(1./(1-exp(-2.*pi.*sqrt(var(1)./(energy-var(2)))))).*(1./(1-(var(4).*energy-var(2))));
end
function z= continum(var,energy)
z = integral(cont(var,energy),2.39,inf);
end
  2 commentaires
Bibek Dhami
Bibek Dhami le 30 Juin 2022
@Torsten I had typo in my integral, so I was getting the constant value. Actually the integral should be not over energy.
But now I am having some issue with the curve fitting. I am using lower and upper bound to restrict the values but I am getting following error.
Lower and upper bounds not supported with complex-valued initial
A=xlsread('Data');
xdata=1240./A(:,1);
ydata=A(:,2);
plot(xdata,ydata,'r.','markersize',30)
box on
energy=xdata;
Absorption=ydata;
options =optimoptions(@lsqcurvefit,'Algorithm','levenberg-marquardt',...
'StepTolerance',1e-19,'MaxFunctionEvaluations',1e10, ...
'MaxIterations',1e3);
lb=[0.03, 2.35, 0.02, 2];
ub=[0.09, 2.40, 0.05, 5];
var=[0.0429 2.3776 0.0207 2.902];
MI=lsqcurvefit(@modelE,var,energy,Absorption,lb,ub,options);
fitdata=modelE(MI,energy);
hold on
% plot(energy,modelE(var,energy),'co')
% plot(energy,exciton(var,energy),'m.','markersize',30)
% plot(energy, continum(var,energy),'k.')
plot(energy,fitdata,'ro')
plot(energy,exciton(MI,energy),'g.','markersize',30)
plot(energy, continum(MI,energy),'b.')
function alpha = modelE(var,energy)
alpha = continum(var,energy) + exciton(var,energy);
end
function x= exciton(var,energy)
x= 2.*var(1).*sech((energy-var(2)+var(1))./var(3))+...
0.125.*var(1).*sech((energy-var(2)+(var(1)./8))./var(3))+...
(2/27).*var(1).*sech((energy-var(2)+(var(1)./27))./var(3));
end
function y= cont(var,energy,x)
y=(sech((energy-x)./var(3))).*(1./(1-exp(-2.*pi.*sqrt(var(1)./(x-var(2)))))).*(1./(1-(var(4).*(x-var(2)))));
end
function z= continum(var,energy)
z = integral(@(x)cont(var,energy,x),2.25,200,'ArrayValued',true);
end
Torsten
Torsten le 30 Juin 2022
During the integration in "continuum", the expression
var(1)./(x-var(2))
becomes or is negative and thus
sqrt(var(1)./(x-var(2)))
becomes complex-valued.
This makes MATLAB's "lsqcurvefit" throw an error and stop.

Connectez-vous pour commenter.

Réponses (1)

Torsten
Torsten le 28 Juin 2022
function z= continum(var,energy)
z = integral(@(x)cont(var,x),2.39,inf);
end
  2 commentaires
Bibek Dhami
Bibek Dhami le 28 Juin 2022
Modifié(e) : Bibek Dhami le 28 Juin 2022
@Torsten Thank you so much it works. But I am having another issue. The plot of this function should give us something like this but it is just giving a line parallel to x axis.
Torsten
Torsten le 28 Juin 2022
An integral gives a single scalar value, not a curve.

Connectez-vous pour commenter.

Catégories

En savoir plus sur Linear and Nonlinear Regression dans Help Center et File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by