Convex.jl
Convex.jl copied to clipboard
1x1 SDP variables incorrectly sized?
Using 1x1 SDP variables produces some wonky behavior. The problem
using Convex
m = 2
k = 1
AX = rand(m)
AVV = rand(m, k^2)
alpha = Variable()
S = Semidefinite(k)
problem = minimize(sum(alpha*AX + AVV*vec(S)))
c,A,b,cones = conic_problem(problem)
cones
produces a zero cone with an empty index range 2:1
:
3-element Array{(Symbol,UnitRange{Int64}),1}:
(:SDP,1:1)
(:Zero,2:1)
(:Zero,2:2)
Setting k=2
(or greater) fixes the problem.