How can I add a constrain to a OLS estimation?

I need to estimate the parameters of a linear regression with OLS and the sum of those parameters has to be equal to 1. How can I add this condition?

The regression is P=H*Z where H is a matrix; P and Z are two vectors

I have to estimate Z with OLS subject to the sum of the elements of Z is 1

Answer 1

[Matt J]

Rewrite Z as

 Z=A*x+b

where x=Z(1:n-1) are the n-1 unknowns and

 A=[eye(n-1);-ones(1,n-1)];
 b=[zeros(n-1,1);+1];

Now, the regression equation in terms of x

 P=(H*A)*x+(H*b)

is unconstrained.