Im testing rmpk using CPLEX on the below example data found on the online vignette:
set.seed(42)
n <- 10
returns <- matrix(
rnorm(n * 20,
mean = runif(n, 0.01, 0.03),
sd = runif(n, 0.1, 0.4)),
ncol = n) # 20 time periods
e <- colMeans(returns)
C <- cov(returns)
min_mu <- 0.02
solver <- ROI_optimizer("cplex")
model <- optimization_model(solver)
x <- model$add_variable("x", i = 1:n, lb = 0, ub = 1)
model$set_objective(sum_expr(2 C[i, j] x[i] x[j], i = 1:n, j = 1:n))
model$add_constraint(sum_expr(e[i] x[i], i = 1:n) >= min_mu)
model$add_constraint(sum_expr(x[i], i = 1:n) == 1)
solver$optimize()
But this does not return a solution. The value columns are NA.
I have also tested that (1) CPLEX is working using the ROI package, and (2) other solvers work for this example problem.
Debugging cplex within rmpk reveals that Q is not a symmetric matrix.
Hi @dirkschumacher ,
Im testing rmpk using CPLEX on the below example data found on the online vignette:
But this does not return a solution. The value columns are NA. I have also tested that (1) CPLEX is working using the ROI package, and (2) other solvers work for this example problem. Debugging cplex within rmpk reveals that Q is not a symmetric matrix.
Any ideas?