Parameters estimation using lsqcurvefit

Hi I am trying to do parameter estimation following the code prepared by Star Strider ‘Igor_Moura'. I have added several differential equations and edited the code. But it does not run. I would really appreciate it if you would help me to solve my problem. Thank you.

Always I receive the following message : Failure in initial user-supplied objective function evaluation. LSQCURVEFIT cannot

continue.

Attached the script

Please help

[Mouna YAHYA]

up please help

Answer 1

[Star Strider]

I did not see this originally, or I would have responded to it several days ago.

The code needed some tweaks (that I will let you find, referring to the ‘Igor-Moura.m’ code will make this easier). I attached it here. It now runs, and produces:

Rate Constants:
		p(1) =  5.09562
		p(2) =  0.02642
		p(3) =  3.25429
		p(4) = -0.28403
		p(5) =  1.06602
		p(6) = 16.32329
		p(7) =  0.20617
		p(8) =  3.98434
		p(9) =  1.38866
		p(10) =  4.19445
		p(11) = -5.18517
		p(12) =  0.05000
		p(13) =  0.67375
		p(14) =  0.22527
		p(15) = -0.22388
		p(16) = 12.31547
		p(17) = 14.00000
		p(18) = -5.37072
		p(19) = 79.89868
		p(20) = -0.27531
		p(21) = 19.43315
		p(22) =  0.88034
		p(23) = 20.60852
		p(24) =  2.81029
		p(25) = -0.48154
		p(26) =  3.54883
		p(27) =  0.07324
		p(28) =  2.13279
		p(29) =  1.30150
		p(30) =  0.24330
		p(31) = 121.96400
		p(32) =  2.20000
		p(33) =  2.20000

Adding:

for k = 1:size(yexp,2)
    subplot(8,2,k)
    plot(t, yexp(:,k), 'p')
    hold on
    plot(tv, yfit(:,k), '-r')
    hold off
    title(sprintf('Column %2d',k))
end

after the ‘yfit’ assignment demonstrates that the fit is not close, however now that the code runs, you can work on fixing it so that it produces better results.

[Mouna YAHYA]

@Star Strider Thank you so much for your answer

unfortunately it didn't run and I didn't understand why. I got the following error

>> Mouna_YAHYA_kinetics

Warning: Failure at t=1.724809e+01. Unable to meet integration tolerances without reducing

the step size below the smallest value allowed (5.684342e-14) at time t.

> In ode15s at 669

In Mouna_YAHYA_kinetics>kinetics at 6

In lsqcurvefit>objective at 261

In snls at 338

In lsqncommon at 150

In lsqcurvefit at 253

In Mouna_YAHYA_kinetics at 118

Error using -

Matrix dimensions must agree.

Error in lsqcurvefit/objective (line 262)

F = F - YDATA;

Error in snls (line 338)

newfvec = feval(funfcn{3},xcurr,varargin{:});

Error in lsqncommon (line 150)

[xC,FVAL,LAMBDA,JACOB,EXITFLAG,OUTPUT,msgData]=...

Error in lsqcurvefit (line 253)

[xCurrent,Resnorm,FVAL,EXITFLAG,OUTPUT,LAMBDA,JACOB] = ...

Error in Mouna_YAHYA_kinetics (line 118)

[p,Rsdnrm,Rsd,ExFlg,OptmInfo,Lmda,Jmat]=lsqcurvefit(@kinetics,p0,t,yexp);

Could you please see it again

thank you

[Star Strider]

As always, my pleasure!

That is a Warning, not an error, and indicates that ode15s (necessary to run this) encountered a singularity at that point and could not continue. These are caused by what is essentially a division-by-0 or an exponential or similar function approaching ∞. You need to go through your code to understand where that occurred, and why. (Most times, this is related to an initial condition, or a coding error.) I do not understand what your code does, so my efforts were limited to getting it to run.

I will help as I can, however for situations such as this, you need to find the problem and then fix it.

[Mouna YAHYA]

okay I will see thank you for your answer

[Star Strider]

As always, my pleasure!