Hi all,

When trying to implement the SOCP in JuMP and Mosek, the Mosek solver stalls and SLOW_PROGRESS JuMP status show up. After checking the results, I notice that it is not feasible, and the upper bounds are depressed.

Post: https://discourse.julialang.org/t/second-order-cone-constraint-in-jump-0-19-with-gurobi-solver/21813/4?u=chengda_ji has successfully addressed this problem by adding upper_bounds when defining the variable.

Unfortunately, in my case, the upper bounds are coupled by the decision variable, as showed in the code below.

```
@variable(m, x[1:5, 1])
@variable(m, 0<=y[1:5, 1]<=1)
for i=1:4
@constraint(x[i]==y[i]+sum(x[j,1] for j in i+1:5))
@constraint(m, [0.5*10, 10, x[Bus,1],y[Bus,1]] in RotatedSecondOrderCone());
end
```

Really appericate it if anyone can help me with that.