reformat solver interface table

[mlubin]

• The JuMP column doesn't transmit much information since all solvers are now supported.
• Incorporate Convex.jl into the table. All LP/SOCP solvers are supported so maybe again not too useful to make a new column?
• Merge MILP/MISOCP/MINLP into a single "integer variables" column?

[madeleineudell]

• It's not clear whether JuMP and Convex are columns or rows. Convex connects to some kinds of solvers to solve some kinds of problems, eg it connects to Mosek but cannot solve MINLPs with Mosek.
• Is it true that if a solver solves problems with integer variables, and it solves problems of type X, then it solves MIXs? This is not true for Convex or Mosek, which can solve SDPs and can solve MISOCPs but can't solve MISDPs (though I've not tried Mosek yet; please correct me if that's false).

[mlubin]

I like the idea of putting Convex and JuMP in as (differently styled) rows. We can use footnotes for the rare (and hopefully transitional) cases where a solver supports the problem class but not through MPB/JuMP/Convex.

It's probably not true that all solvers that support MIPs and SDPs also support MISDPs. We can either handle this with footnotes or some other restructuring.

[tkelman]

And is the Mosek NLP/MINLP solver only capable of solving convex problems or am I imagining seeing that elsewhere?

[mlubin]

Yeah, that deserves a footnote. On Jan 11, 2015 9:00 AM, "Tony Kelman" notifications@github.com wrote:

And is the Mosek NLP/MINLP solver only capable of solving convex problems or am I imagining seeing that elsewhere?

— Reply to this email directly or view it on GitHub https://github.com/JuliaOpt/juliaopt.org/issues/12#issuecomment-69499264 .