Implement quadratic functionality #802
Conversation
exaexa
force-pushed
the
sew-fba-tests-etc
branch
2 times, most recently
from
f0c6678 to
fe1ef35
Compare
|
/format |
|
✔️ Auto-formatting triggered by this comment succeeded, commited as e966427 |
exaexa
force-pushed
the
sew-pfba-tests-etc
branch
from
e966427 to
a985ffa
Compare
exaexa
force-pushed
the
sew-pfba-tests-etc
branch
from
00e2f41 to
0bc99a2
Compare
exaexa
force-pushed
the
sew-pfba-tests-etc
branch
from
0bc99a2 to
a2c121e
Compare
|
@stelmo if you could have a quick look now it would be great. I've set up proper intermediate (CT-based) functions for FBA/pFBA; same will follow with MOMA and FVA and likely later for others. These will also be used exclusively for the complex models. I split off the MOMA from QP tutorial. To be more realistic I assume we generally want to run pFBA instead of plain FBA to obtain a sane initial "reference" solution, right? |
| ctmodel; | ||
| objective = ctmodel.:l2objective.value, | ||
| optimizer = Clarabel.Optimizer, | ||
| sense = Minimal, |
There was a problem hiding this comment.
I don't like this Minimal. Since we are using JuMP's set_optimizer_attribute etc., we should use JuMP's Min and Max
There was a problem hiding this comment.
we're renaming JuMP's symbols to ours (people don't import them from JuMP!) so we shouldn't also force them to import parameters.
|
|
||
| vt = C.constraint_values(ctmodel, J.value.(opt_model[:x])) # ConstraintTrees.jl is called C in COBREXA | ||
|
|
||
| @test isapprox(vt.l2objective, ?; atol = QP_TEST_TOLERANCE) #src # TODO will break until mutable bounds |
There was a problem hiding this comment.
This is still commented out :D
|
|
||
| vt = C.constraint_values(ctmodel, J.value.(opt_model[:x])) # ConstraintTrees.jl is called C in COBREXA | ||
|
|
||
| @test isapprox(vt.l2objective, ?; atol = QP_TEST_TOLERANCE) #src # TODO will break until mutable bounds |
| Change the objective sense of optimization. Possible arguments are | ||
| `JuMP.MAX_SENSE` and `JuMP.MIN_SENSE`. | ||
| Change the objective sense of optimization. Accepted arguments include | ||
| [`Minimal`](@ref), [`Maximal`](@ref), and [`Feasible`](@ref). |
There was a problem hiding this comment.
What does Feasible do?
There was a problem hiding this comment.
|
|
||
| Same as `JuMP.FEASIBILITY_SENSE`. | ||
| """ | ||
| const Feasible = J.FEASIBILITY_SENSE |
There was a problem hiding this comment.
Did not know this exists
|
|
||
| """ | ||
| Minimal | ||
|
|
||
| Objective sense for finding the minimal value of the objective. | ||
|
|
||
| Same as `JuMP.MIN_SENSE`. | ||
| """ | ||
| const Minimal = J.MIN_SENSE | ||
|
|
||
| """ | ||
| Maximal | ||
|
|
||
| Objective sense for finding the maximal value of the objective. | ||
|
|
||
| Same as `JuMP.MAX_SENSE`. | ||
| """ | ||
| const Maximal = J.MAX_SENSE | ||
|
|
||
| """ | ||
| Maximal | ||
|
|
||
| Objective sense for finding the any feasible value of the objective. | ||
|
|
||
| Same as `JuMP.FEASIBILITY_SENSE`. | ||
| """ |
There was a problem hiding this comment.
My concern here is that it is just more stuff to remember. Can we not just re-export the JuMP constants?
There was a problem hiding this comment.
one can use the jump constants as well. I just don't want to 1] cause identifier collision 2] have folks to import jump all over again everytime they want to do something
Yep, it would make sense to use a pFBA solution for MOMA, but as long as its general i.e. the user can decide how sane they want to be, I don't mind |
I converted this to an admonition opportunity. |
No description provided.