I've assigned Psi1,2 to be obtained by subtraction, and therefore Psi1,1 and 1,3 appear in the Design Matrix. I fix 1,1 to zero, and then constrain 1,3 according to covariates. Obviously, Psi1,2 does not appear in the DM, but it is still being "constrained" in the same manner as Psi1,3?
If I'm following...
3 states. Conditional on being alive at (i+1), psi(1,1)+psi(1,2)+psi(1,3)=1. If psi(1,1)=0, then psi(1,2)+psi(1,3)=1, meaning, psi(1,2)=1-psi(1,3), and so on. Meaning, if you constrain one estimate to be a linear function of a covariate, then you're constraining the other one as well.
But, be careful. Since the parameters are linear functions of one another, then a constraint on one parameter can have unintended effects on the other. For example, suppose you constrain psi(1,2) to increase over time. Then, since psi(1,3)=1-psi(1,2), then psi(1,3) will be constrained to decrease over time, which may not be what you want.
This same sort of problem holds for 'Pradel models' - if you (say) constrain lambda to be constant over time, then if recruitment goes up, survival must go down. Which may not, in fact, make any sense.
Moral: you need to be very careful applying constraints to models where parameters or sets of parameters are linear functions of each other. This linear structure means that whatever constraints you apply to one parameter may have logical implications for other parameters - whether you want them to or not. Remember, you're subjecting parameters to a given constraint, subject to an overriding logical constraint.