outputs from lsqlin parameter estimation

Hi

I am working on using 'lsqlin' for fitting my 3 parameters (a set of 3) in my model.

This is the command I used: k=lsqlin(stackedmatrix{:},ftderivativesofP,LB,UB)

Where k is output to store my estimated parameters. each of stackedmatrix{index} contains a square matrix (3X3) where index=1 to 189.

ftderivativesofP is a column vector (no.of rows=189). LB variable for setting lower bounds, I did: LB=0.00001*ones(3,1)

and UB for upper bounds. set as : UB=100000*ones(3,1)

Output I am getting is the following, with a bunch of warning messages:

Warning: Length of lower bounds is > length(x); ignoring extra bounds. > In checkbounds at 28 In lsqlin at 236 Warning: Length of upper bounds is > length(x); ignoring extra bounds. > In checkbounds at 42 In lsqlin at 236 Exiting due to infeasibility: 1 lower bound exceeds the corresponding upper bound.

k =

   1.0e+03 *

   -9.9275
   -9.9275
    9.9275
         0
    0.9996
    0.9996
   -0.9996
         0
         0
         0
   -0.0004
    0.0004

(My main) Question is : Why am I getting 12 estimated values, as there are only 3 parameters to be estimated.

Also, any guidance for the reasoning behind these warning messages would be helpful for me.

Thanks in advance

Answer 1

[Matt J]

You seem to be disregarding the documentation as to which order the input arguments are to be given in. If you look again at "doc lsqlin" you will see that LB and UB are supposed to be the 7th and 8th inputs, not the 3rd and 4th.

Aside from that

 lsqlin(stackedmatrix{:},arg1,arg2,...)

is equivalent to typing

 lsqlin(stackedmatrix{1}, stackedmatrix{2}, stackedmatrix{3},...,stackedmatrix{end}, arg1, arg2,...)

If this is really what you are doing, no doubt it is contributing to the problem.

[SreeHarish Muppirisetty]

Hi

Tried with the following in command line:

>> options = optimoptions('lsqlin','MaxIter',500000000);

>> lb=0.0000000000001*ones(3,1);

>> ub=100000*ones(3,1);

>> k=lsqlin(stackedmatrix{:},ftderivativesofP,[],[],[],[],lb,ub,options)

Warning: Length of lower bounds is > length(x); ignoring extra bounds. > In checkbounds at 28

In lsqlin at 236

Warning: Length of upper bounds is > length(x); ignoring extra bounds. > In checkbounds at 42 In lsqlin at 236

Exiting due to infeasibility: 1 lower bound exceeds the corresponding upper bound.

k =

   1.0e+03 *

   -9.9275
   -9.9275
    9.9275
         0
    0.9996
    0.9996
   -0.9996
         0
         0
         0
   -0.0004
    0.0004

Still getting similar warning messages.

The main question remains for me: I am getting 12 output values, where in I have only 3 parameters to be estimated.

[Matt J]

You are still inputting stackedmatrix{:}. If you are doing so intentionally, then I think you still don't understand what the {:} syntax means.

[SreeHarish Muppirisetty]

Hi Matt

I formed a matrix from my stacked cell array using cell2mat.

Now when I execute lsqlin as

k=lsqlin(stackedarray,ftderivativesofP,[],[],[],[],lb,ub,X0,options) where options is set as:

options = optimoptions('lsqlin','MaxIter',500000000); and

X0=[0.000000001;0.000000001;0.000000001];

I get the following Error:

Error using lsqlin (line 188) The number of rows in C must be equal to the length of d.

[Matt J]

It's because stackedarray has more rows than ftderivativesofP