Allow eom_bounds to be set to arrays

[Peter230655]

In #529, I believe you refer to bounds not to eom_bounds
Same feature would be nice for them, too.

[moorepants]

Please make a useful title for issues.

[Peter230655]

What I meant was this:

• currently we can set variable bounds for $x^L \leq x \leq x^U$ for x a state variable, using bound = {...}

If I understood what you said in #546 correctly, we can also set variable bouund $g^L \leq g(x) \leq g^U$, using eom_bounds.
If this is the case, this issue can be closed.

[moorepants]

No, I don't think you can use opty's eom_bound kwarg to set the bounds to anything other than a single value per equation. In #546 I was discussing what is possible with IPOPT because you seemed to be requesting have the bounds change dynamically as a function of the state value. The free values change during each IPOPT iteration. That is not possible with IPOPT and thus not possible with opty.

[Peter230655]

Do I understand correctly: opty could do it in principle, as IPOPT can, but currently not possible?

[moorepants]

It is not so clear to me what you have opened this issue about. You refer to #529 but that is a discussion about updating the bounds array after Problem instantiation. That is not possible, for bounds= or eom_bounds, due to IPOPT's design. The title of this issue is also not entirely clear. Do you want to adjust bounds? Or set them initially?

I merged #532 into opty which allows you to set bounds to arrays instead of simply floats at Problem instantiation. This can also be done for eom_bounds, in mostly the same way.

You also request something else in #546, which still isn't entirely clear. Are you desiring bounds or eom_bounds to change during solve iterations as a function of what the free variables are?

[Peter230655]

I used the wrong reference here,, #532 would be correct.
Here ( in issue 530) I asked that the eom_bounds could have same variable (arrays) limits as bounds have. This could be done as you exlapned above.
In #546 I asked that the variable bounds be functions of the sates, which you explained is not possible due to IPOPT's limitations. Therefore I closed #546 as being answered.

[moorepants]

Ok, I've changed the title of this issue. Hopefully that now reflects what you are requesting.

[Peter230655]

Yes, exactly what I request.