Matrix dimensions must agree

Hello,

So, I've tested this program with many other distributions and it worked. However, when I changed the probability density function 'custpdf' in the code below, I got

an error saying tha matrix dimentsions must agree.

x=xlsread('my_data.xlsx');
t=x(:,1);
cens=x(:,2);
x1=x(:,3);
x2=x(:,4);
custpdf = @(t,alpha,beta,b0,b1,b2)exp(b0+b1.*x1+b2.*x2).*(exp((alpha./-beta).*(exp(beta.*...
    t)-1))).^(exp(b0+b1.*x1+b2.*x2)-1).*alpha.*exp(beta.*t).*exp(-(alpha.*...
    (exp(beta.*t)-1))./beta);
custcdf = @(t,alpha,beta,b0,b1,b2) 1-(exp((alpha/-beta).*(exp(beta.*t)-1))).^exp(b0+b1.*...
    x1+b2.*x2);
[phat,pci] = mle(t,'pdf',custpdf,'cdf',custcdf,'start',[0.001,0.01,1,1,1],'Censoring',cens)

I get the error below:

Error using mlecustom (line 166)

Error evaluating the user-supplied pdf function

'@(t,alpha,beta,b0,b1,b2)exp(b0+b1.*x1+b2.*x2).*(exp((alpha./-beta).*(exp(beta.*t)-1))).^(exp(b0+b1.*x1+b2.*x2)-1).*alpha.*exp(beta.*t).*exp(-(alpha.*(exp(beta.*t)-1))./beta)'.

Error in mle (line 239)

[phat, pci] = mlecustom(data,varargin{:});

Error in mle_MGD_regression (line 22)

[phat,pci] = mle(t,'pdf',custpdf,'cdf',custcdf,'start',[0.001,0.01,1,1,1],'Censoring',cens)

Caused by:

Error using .^

Matrix dimensions must agree.

I can't seem to find the problem with the probability density function?

[Ameer Hamza]

Can you share the your data file. It will be easier to track down the error if we have data.

[tess hamd]

Yeah sure.

[Tommy]

The only difference between this code and previous code which worked is custpdf? When I run

custpdf(t,0.001,0.01,1,1,1)

using your data, t, and your start point, [0.001,0.01,1,1,1], I get all zeros. After a little more digging,

(exp((alpha./-beta).*(exp(beta.*t)-1))).^(exp(b0+b1.*x1+b2.*x2)-1)

returns the zeros, because it is essentially (numbers less than 1).^(insanely huge numbers). Is it possible that there are some parentheses out of place?

[Ameer Hamza]

tess, the custpdf function returns a vector

custpdf(t,0.001,0.01,1,1,1)
ans =
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0
     0

however, it should return a scalar. I think there is some issue in this formula of the cdf function. Can you share the mathematical formula of cdf and pdf.