Parameter fit fail

CADET Troubleshooting
1

Hi

Does anyone experience this error when using matlab parest example?

[Error: idasErrorHandler::196] In function ‘IDASolve’ of module ‘IDAS’, error code ‘IDA_CONV_FAIL’:
At t = 6 and h = 9.53674e-13, the corrector convergence failed repeatedly or with |h| = hmin.

Adjust relTol and absTol didn’t help. Any suggestions? Thanks!

params = cell(6, 1);
params{1}=makeSensitivity(0, {'MCL_KA'}, 0, -1, -1, 0, -1);
params{2}=makeSensitivity(0, {'MCL_KD'}, 0, -1, -1, 0, -1);
params{3}=makeSensitivity(0, {'MCL_QMAX'}, 0, -1, -1, 0, -1);
params{4}=makeSensitivity(0, {'MCL_KA'}, 1, -1, -1, 0, -1);
params{5}=makeSensitivity(0, {'MCL_KD'}, 1, -1, -1, 0, -1);
params{6}=makeSensitivity(0, {'MCL_QMAX'}, 1, -1, -1, 0, -1);
sim.setParameters(params, true(6, 1));

idxComp = cell(1, 1);
idxComp{1} = [1, 1];

pf.addExperiment(data, sim, [0], idxComp, [], [], [], [], 'Composition Profiles', {'Total Protein'});
initParams = [70, 1,3.6, 5.0.01,5];   
loBound = initParams.*0.1;
upBound = initParams.*10;

plotHd = [];
opts = optimset('TolFun', 1e-6, 'TolX', 1e-6, 'MaxIter', 100, 'Jacobian', 'on', 'Diagnostics', 'off', 'Display', 'iter', 'OutputFcn', @progressMonitor);

[optimalParams, optimalRes, ~, exitflag] = lsqnonlin(@residual, initParams, loBound, upBound, opts);
success = (exitflag > 0);
save('Set1-1.mat', 'optimalParams', 'optimalRes')
2

Hi Yiran,

Please also post the model setup code.

  • Is t = 6 at a section transition?
  • Does the model run (with the initial parameters) outside parameter estimation?
  • Does the problem occur in the first iteration of the optimization?
  • Does the model run without parameter sensitivities (outside parameter estimation)?

Best,
Sam

3

Hi Sam,
200gL_8min_simple.xlsx (30.1 KB) YWTwoCompExtendedLoading_FindOpt_simple.m (5.5 KB)

Thanks for your response. I attach the code and data here if that helps.

  • Is t = 6 at a section transition?
    It is the first time point of the calculation
  • Does the model run (with the initial parameters) outside parameter estimation?
    The model runs well with initial parameters.
  • Does the problem occur in the first iteration of the optimization?
    Yes
  • Does the model run without parameter sensitivities (outside parameter estimation)?
    Model works well without parameter estimation.

Thanks，
Yiran