array<real_T, 2U> &thisFit, real_T P)

{

::coder::array<real_T, 1U> terms;

::coder::array<int32_T, 1U> n;

::coder::array<boolean_T, 1U> b_terms;

real_T b_thisData[2];

real_T N;

int32_T i;

int32_T loop_ub;

// Chi-squared function is used to evaluate the goodness of fit.

// It is a measure of the difference between the observed and expected.

// allChis = zeros(1,numberOfContrasts);

// thisData = allData{i};

// thisFit = allFits{i};

b_thisData[0] = thisData.size(0);

b_thisData[1] = 1.0;

N = coder::internal::maximum(b_thisData);

if (N <= P) {

N = P + 1.0;

}

if (thisData.size(0) == 1) {

i = thisFit.size(0);

} else {

i = thisData.size(0);

}

if ((thisData.size(0) == thisFit.size(0)) && (i == thisData.size(0))) {

terms.set_size(thisData.size(0));

loop_ub = thisData.size(0);

for (i = 0; i < loop_ub; i++) {

real_T varargin_1;

varargin_1 = (thisData[i + thisData.size(0)] - thisFit[i + thisFit.size

(0)]) / thisData[i + thisData.size(0) * 2];

terms[i] = varargin_1 * varargin_1;

}

} else {

binary_expand_op(terms, thisData, thisFit);

}

b_terms.set_size(terms.size(0));

loop_ub = terms.size(0);

for (i = 0; i < loop_ub; i++) {

b_terms[i] = (terms[i] == rtInf);

}

coder::eml_find(b_terms, n);

if (n.size(0) != 0) {

loop_ub = n.size(0);

for (i = 0; i < loop_ub; i++) {

terms[n[i] - 1] = 0.0;

}

}

return 1.0 / (N - P) * coder::sum(terms);

// allChis(i) = chi2;