Objective function is returning Inf or NaN values at initial point. lsqcurvefit cannot continue.

Hello together,

I have a diffusion model for porous solids but I get the error message " Objective function is returning Inf or NaN values at initial point. lsqcurvefit cannot continue.". However, I dont understand it since I can calculate the values and wont get Nan or Inf values if I use a normal calculator. Is my code wrong or is it because of numerical problems? Thanks in advance

A = 0.0019634954; 
R = 8.314; 
T = 298.15; 
N_A = 6.02214086*(10^23); 
p = 100000; 
k_B = 1.38064852E-23; 
M = 31.999E-3; 
dc = 40; 

s_Kunststoff = 23E-06; 



p_1 = [1; 0.1343; 0.0173; 0; 0; 0]; 
p_2 = [0; 0.5089; 0.4445; 0.4606; 0.3222; 0.632];  
p_3 = [0; 0.0465; 0.069; 0.061; 0.067; 0.076];  
p_4 = [0; 0.2407; 0.3855; 0.3641; 0.4883; 0.7211]; 

r_m = 15E-09; 
d = 0.06E-09; 

D_Knudsen = ((2/3)*r_m*((8*R*T)/(pi*M))^0.5);
D_Molekularsieb = D_Knudsen/15

D_Molekularsieb = 2.9610e-07

x = [0 5E-09 9E-09 14E-09 18E-09 25E-09]';  
y = [33.20312499864E-09 19.57947530784E-09 4.079861110944E-09 1.59143518512E-09 1.456404320928E-09 1.18152006168E-09]';  

modelFun = @(b,x) A.*dc./((1./(((D_Knudsen.*p_2)./(b(5).*x))+((D_Molekularsieb.*p_3)./(b(6).*x))+((b(3).*exp(-b(4)./(R.*T)))./x)))+(s_Kunststoff./(b(1).*exp(-b(2)./(R.*T)))));
b0 = [1e+03 8e+04 1e+03 8e+04 1 1]

b0 = 1×6

1000 80000 1000 80000 1 1

lb = [0 0 0 0 1 1];
ub = [1e+20 1e+20 1e+20 1e+20 30 30];
b = lsqcurvefit(modelFun,b0,x,y,lb,ub);

Error using lsqncommon
Objective function is returning Inf or NaN values at initial point. lsqcurvefit cannot continue.

Error in lsqcurvefit (line 274)
lsqncommon(funfcn,xCurrent,lb,ub,options,defaultopt,optimgetFlag,caller,...

times = linspace(x(1),x(end));
plot(x,y,'ko',times,modelFun(b,times),'b-')
legend('Data','Fitted exponential')
title('Data and Fitted Curve')

Answer 1

[Matt]

If you try this :

modelFun(b0,x)

the output has a nan. So you need to rework your function modelFun() in order to avoid this.

[Amir Mirmoini]

Thank you very much! That helped me to find the issue.

For anyone interested: I have a zero in my x values, so this is where the NaN comes from

[Matt J]

@Amir Mirmoini

If your issue is now resolved, please Accept-click @Matt's answer.