Error in nlmefit function vectorizeSinglePhiGroupX

Hi, I am trying to use nlmefit to do a linear random effects model of some data using the following command:

beta = nlmefit(a,b,grp,[],@(PHI,X)(X*PHI),[1 0]);

where

a=[0 1 2 0 1 2 0 1 0 1]';   b=[6.4512 5.9866 4.8052 7.0858 6.2563 5.5702 7.0976 7.7617 5.8512 7.1175]';   grp=['ABC' 'ABC' 'ABC' 'ADW' 'ADW' 'ADW' 'AEC' 'AEC' 'AEL' 'AEL'];

a,b, and grp have been shortened for brevity, they are actually 118x1 vectors.

I get the following error that I do not understand:

In an assignment  A(I) = B, the number of elements in B and I must be the same.

Error in nlmefit>getFcns/vectorizeSinglePhiGroupX (line 1367)
        fval(k) = model(phi(i,:),X(k,:),V(i,:));

Error in nlmefit>getFcns/computeFAllAndCount (line 1450)
    f = modelVectorizedAllGroups(phi, X, VVectorized);

Error in nlmefit>@(Beta,b)computeFAllAndCount(phiAllGroupsFcn(Beta,b)) (line 1459)
computeFAllGroupsFcn = @(Beta,b) computeFAllAndCount(phiAllGroupsFcn(Beta,b));

Error in nlmefit>@(Beta)computeFAllGroupsFcn(Beta,bDummy) (line 478)
    model = @(Beta) computeFAllGroupsFcn(Beta, bDummy);

Error in nlmefit>LMfit (line 1752)
yfit = model(beta);

Error in nlmefit (line 481)
    Beta_hat = LMfit(Y,Beta_hat,LMfitOptions,model,jacobian);

Thanks in advance for any clues!

Erin

Answer 1

[the cyclist]

The proximate cause of your error is that your function specification

@(PHI,X)(X*PHI)

has only one parameter, PHI, but you have specified a vector with two initial values, [1 0].

The following syntax will execute

beta = nlmefit(a,b,grp,[],@(PHI,X)(X*PHI),[1]);

although I don't know if that is really the function you are trying to solve.

[Erin MacMillan]

Thank you so much! You are right that the syntax fixed the error but was not actually the function I was trying to solve. If I want my model to be b = Xa + c, where c is some constant, then should the function be

@(PHI,X)(X*PHI(:,1)+PHI(:,2)

The nlmefit function executed without error with this model but I'm not totally sure that this is the correct model. Thanks for your help!