-
Notifications
You must be signed in to change notification settings - Fork 53
Conversation
Do we really need all of these One thing, though, is that you can never test that something |
I think we should have |
@jlperla Nevermind, this isn't true:
So yeah, we could definitely delete these. Will do from now on. |
Having |
I know for some with QuantEcon given a naive (in the lectures) and a high precision one (in the package) I think I needed |
That isn't true. It uses the machine epsilon, but comparing floating points adjusts for the scale. See https://docs.julialang.org/en/v1.0.0/base/math/#Base.isapprox Try to avoid them if possible and just add in extra numbers if necessary, but not a big deal. |
Sure, it can wait. But it also is only a problem if you use the julia> using LinearAlgebra
julia> I = 5
5
julia> I
5 |
OK, I stripped the tolerances from the tests and there were no issues. @Nosferican is this good to merge? |
Not much code in this one. Just verified that the three different projection schemes come out to the same result.