search

jump-dev / Pajarito.jl

A solver for mixed-integer convex optimization
Mozilla Public License 2.0
130 stars 22 forks source link

Pajarito.jl/examples/cardls.jl，conic_solver = MosekSolver(LOG=0)，MosekSolver not defined #438

Closed duyanran closed 2 years ago

duyanran commented 3 years ago

when i try to run exmaple given from the official file, using MosekTools conic_solver =MosekSolver()

i‘ve got this image

how can i fix it？ thanks

blegat commented 3 years ago

MosekTools contains the MathOptInterface version of the interface, the MathProgBase version is in Mosek.jl but not the latest version

duyanran commented 3 years ago

MosekTools contains the MathOptInterface version of the interface, the MathProgBase version is in Mosek.jl but not the latest version

thanks for your explaination, do you mean i should update my MathProgBase to the latest version or should i update mosek? packages' version always drives me nuts, really appreciate your reply!

blegat commented 3 years ago

MathProgBase is deprecated so the latest version of Mosek.jl does not export the MathProgBase wrapper anymore. You need to use an older one. See https://github.com/MOSEK/Mosek.jl/issues/183

duyanran commented 3 years ago

不推荐使用MathProgBase，因此最新版本的Mosek.jl不再导出MathProgBase包装器。您需要使用较旧的。参见MOSEK / Mosek.jl＃183

hi, thanks for ur reply. i try to medel with convex and the following error comes:

ERROR: The providedoptimizer_constructoris invalid. It must be callable with zero arguments. For example, "Ipopt.Optimizer" or "() -> ECOS.Optimizer()". It should not be an instantiated optimizer like "Ipopt.Optimizer()" or "ECOS.Optimizer()". (Note the difference in parentheses!)

i guess still something wrong with my package version....my english isn't really good, so it's a bit hard for me to understand the link u enloped....sorry.....

could u tell me which version of mosek, jump and convex i should add ? i think scs can solve misdp too, so i am about to try it first and if it's not working maybe i'll go with mosek@0.8 . (i to run this code : Pajarito.jl/examples/gatesizing.jl)

many thanks!!

(BTW, i really hope every example in any pkg could specify their package version, for a green hand this bothered me many times......

blegat commented 3 years ago

Normally you should always use the latest versions. Pajarito is an exception as it hasn't been update for a while, see https://github.com/JuliaOpt/Pajarito.jl/issues/430. You should use JuMP v0.18 You could try https://github.com/lanl-ansi/Juniper.jl for MISDP. It's doing branch and bound and does not assume any particular structure for the model so it should in principle work with MISDP although I've never tried.

duyanran commented 3 years ago

通常，您应该始终使用最新版本。Pajarito是一个例外，因为它已经有一段时间没有更新了，请参阅＃430。 您应该使用JuMP v0.18。 您可以尝试使用MISDP的https://github.com/lanl-ansi/Juniper.jl。它正在分支和绑定，并且不为模型采用任何特定的结构，因此尽管我从未尝试过，但它原则上应与MISDP一起使用。

thanks for ur advise, i tried JuMP@0.18 and it worked, but my model involved complex variable, i'd like to build the model with convex, but it seems it can not solve complex MISDP?

btw, i tried to add a lower version of Mosek, and i got this ERROR: Unsatisfiable requirements detected for package Mosek [6405355b]: Mosek [6405355b] log: ├─possible versions are: [0.9.1-0.9.11, 1.0.2-1.0.4, 1.1.0-1.1.3] or uninstalled └─restricted to versions 0.10.2 by an explicit requirement — no versions left image and v1.0.2 is the lowest ver of Mosek, but still i got not defined image i am really sad but this

odow commented 2 years ago

The current version is effectively unmaintained. We are working on a replacement. Closing in favor of #411.