Did anyone use "x = lsqnonneg(C,d,options)" function mathematically?

I have trouble with use "x = lsqnonneg(C,d,options)" mathematically especially "options" ... options = optimset('param1',value1,'param2',value2,...)

Thanks in advance

Riyadh

Answer 1

[Walter Roberson]

The only options accepted for lsqnonneg are listed at http://www.mathworks.com/help/matlab/ref/optimset.html as Display and TolX. Display has no effect on the calculation. TolX is described http://www.mathworks.com/help/matlab/math/setting-options.html#bt00l89-1

"TolX is a lower bound on the size of a step, meaning the norm of (xi – xi+1). If the solver attempts to take a step that is smaller than TolX, the iterations end."

Mathematics might be infinitely precise in theory, but reaching infinite precision can take an infinite amount of time, and the user might not need that precision. Besides, double precision cannot distinguish more than one part in 2^53; see http://www.mathworks.com/help/matlab/ref/eps.html

[Riyadh Muttaleb]

Thank you for your reply, My problem is that I have vector [10X1] all its values are known and matrix [10X3] has 3 unknown [d1, d2, d3] for each row, so how can I find these values using (lsqnonneg)?

Best Regards

[Walter Roberson]

Example:

C = rand(10,3); d = rand(10,1);
d123 = lsqnonneg(C,d);
d1 = d123(1); d2 = d123(2); d3 = d123(3);

This presumes that the "3 unknown in each row" is a linear relationship, C(k,1) * d1 + C(k,2) * d2 + C(k,3) * d3 = d(k) . If that is not the case then lsqnonneg() is probably not the correct routine for the situation.

[Riyadh Muttaleb]

Thank you so much!

[Riyadh Muttaleb]

Hello Walter, I would like to find the confidence interval for the results of "lsqnonneg", could you please help me?

Thanks in advance Riyadh

[Walter Roberson]

You appear to be discussing this with John over in https://www.mathworks.com/matlabcentral/answers/338575-how-can-i-calculate-the-confidence-interval-for-lsqnonneg-regression