"All Zero guesses not acceptable" in 'optim-1.0.4/leasqr.m'

[Sergei Steshenko]

Hello,

I think "All Zero guesses not acceptable" in 'optim-1.0.4/leasqr.m' is a 
wrong limitation.

The guesses, I guess, are related to the parameters to be calculated by the
function.

Suppose we have the following function:

y = p * x

where 'y', 'p', 'x' are scalars.

Suppose there are two observed values according to the following table:

x y
1 1
2 2
.

So, if I guess p = 1, it's the correct solution.

Suppose I rewrite the function differently:

y = (1 + p) * x

, and observed values are the same, as well as observation points.

Obviously for such a case the correct value of 'p' is 0.

So, why can't I supply the absolutely correct guess of 0 in this case ?

Thanks,
  Sergei.