Fitting 4 parameters with constraints

7 views (last 30 days)
I am trying to fit a function to a data set in order to fit 4 parameters. I have been using lsqcurvefit because it is the only way that I know how to do a good easy fitting. But the problem is that I need to add constraints to my parameters. This is not possible using lsqcurvefit. I wonder if there is a function or a manual way of doing this.
where A1 D1 A2 D2 are unknown cosntants.
The constraints I need to have is that D1 and D2 are less than 0.001 (D1 and D2 < 0.0008). Any solution that includes either D1 or D2 greater than that is not of interest.
A solution that lsqcurvefit is finding is
A1 = 0.364
D1 = 0.001338
A2 = 0.610
D2 = 0.000110
*****Plot as a semilogY ****
D1 in this case is greater than my desired constraint. I would like for it to skip it and keep iterating.
I will include my data below if anybody could help me I appreciate it!
clear; clc; clf; close all;
D_0 = [0.5 0.00001 0.5 0.00001]; %initial guess if necessary for A1 D1 A2 D2 respectively
xdata = [9.85 32.05 66.83 114.44 174.55 247.57 333.00 431.44 542.19 666.04 802.12 951.39 1113.38 1287.47 1474.87 1674.29 1887.11 2600.14 2863.78 3139.17 3428.23 4043.41 4369.45 4709.33 5425.99 5802.67 6193.38 6595.39];
ydata = [1 0.9569 0.9528 0.8894 0.8387 0.8995 0.7911 0.773 0.7523 0.7155 0.7086 0.6478 0.6269 0.6175 0.574 0.551 0.4991 0.4559 0.4449 0.4314 0.4212 0.407 0.3856 0.3511 0.3526 0.303 0.3148 0.2912];

Accepted Answer

Star Strider
Star Strider on 30 Apr 2022
But the problem is that I need to add constraints to my parameters. This is not possible using lsqcurvefit.
Yes, it is, however only constraint bounds.
objfcn = @(b,x) b(1).*exp(-b(2).*x) + b(3).*exp(-b(4).*x);
D_0 = [0.5 0.00001 0.5 0.00001];
xdata = [9.85 32.05 66.83 114.44 174.55 247.57 333.00 431.44 542.19 666.04 802.12 951.39 1113.38 1287.47 1474.87 1674.29 1887.11 2600.14 2863.78 3139.17 3428.23 4043.41 4369.45 4709.33 5425.99 5802.67 6193.38 6595.39];
ydata = [1 0.9569 0.9528 0.8894 0.8387 0.8995 0.7911 0.773 0.7523 0.7155 0.7086 0.6478 0.6269 0.6175 0.574 0.551 0.4991 0.4559 0.4449 0.4314 0.4212 0.407 0.3856 0.3511 0.3526 0.303 0.3148 0.2912];
[B,rsdnrm] = lsqcurvefit(objfcn, D_0, xdata, ydata, -Inf(1,4), [Inf 0.001 Inf 0.001])
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.
B = 1×4
-18.9893 0.0001 19.7668 0.0001
rsdnrm = 0.1898
xv = linspace(min(xdata), max(xdata));
figure
plot(xdata, ydata, 'pg')
hold on
plot(xv, objfcn(B,xv), '-r')
hold off
grid
xlabel('x')
ylabel('y')
.
  2 Comments
Alfredo Scigliani
Alfredo Scigliani on 30 Apr 2022
Edited: Alfredo Scigliani on 30 Apr 2022
Thank you for your input! Nonetheless, I am trying to fit a multiexponential fucntion, what the fitting is doing is finding a solution as a monoexponential function. If both of the D parameters are the same, then you can add the two terms together and it will just be 1 exponential. D1 and D2 must be different from eachother as well.
Try plotting is as semilogy, if multiple exponentials are in play, the line will be curved.
Alex Sha
Alex Sha on 30 Apr 2022
if want D1 and D2 are all less than 0.001:
Sum Squared Error (SSE): 0.0116703699647364
Root of Mean Square Error (RMSE): 0.0204156539770838
Correlation Coef. (R): 0.995666057914124
R-Square: 0.991350898882251
Parameter Best Estimate
---------- -------------
a1 0.428678645876139
a2 0.534815808010459
d1 0.001
d2 8.58505177096869E-5
while, if want D1 and D2 are all less than 0.0008:
Sum Squared Error (SSE): 0.0140886399832964
Root of Mean Square Error (RMSE): 0.0224313555918754
Correlation Coef. (R): 0.994765584731217
R-Square: 0.989558568565641
Parameter Best Estimate
---------- -------------
a1 0.495993548397224
a2 0.458843635970715
d1 0.0008
d2 5.99343414652501E-5

Sign in to comment.

More Answers (0)

Products


Release

R2019b

Community Treasure Hunt

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

Start Hunting!