MIP using R
Answered
Hi, I was solving an MIP using R Gurobi, this was my original code and it can solve it. Then I changed the inputs from Xi being (0,1) to Zi being(-1,0,1), I added a quadratic constraint Sum of Zi^<=m and tried to solve the MIP again, and it always gave me the infeasible solution, even when I set the quadratic constraint to be redundant, could you please take a look and tell me if I'm using something wrong here, Thank you! Attached is the main code for both parts.
Original was:
n <- nrow(W)
nedge <- rowSums(W)
nedge.total <- sum(nedge)
delta <- qnorm(alpha)*sqrt(sum(nedge^2))
model <- list()
params <- list(MIPGap=0.01, TimeLimit=300)
A <- matrix(1, 1, n)
A <- rbind(A, nedge)
A <- rbind(A, nedge)
model$A <- A
model$rhs <- c(ifelse(n%%2,(n+1)/2, n/2), nedge.total/2+delta/2, nedge.total/2-delta/2)
model$sense <- c("=","<=",">=")# set up quadratic objective
model$Q <- W*obj.sign
model$obj <- -rowSums(W)*obj.signmodel$vtype <- rep('B', n)
result <- gurobi(model, params)
And then this is the updated code:
n <- nrow(W)
nedge <- rowSums(W)
nedge.total <- sum(nedge)
delta <- qnorm(alpha)*sqrt(sum(nedge^2))
model <- list()
A <- matrix(1, 1, n)
A <- rbind(A, nedge)
A <- rbind(A, -nedge)
model$A <- A
model$rhs <- c(ifelse(n%%2,(n+1)/2, n/2), nedge.total/2 + delta/2, nedge.total/2 - delta/2)
model$sense <- c("=", "<=", ">=")
model$Q <- W * obj.sign
model$obj <- -rowSums(W) * obj.sign
# Set the variable types and bounds for Zi variables
model$vtype <- rep('I', n)
model$lb <- rep(-1, n)
model$ub <- rep(1, n)
Qc <- sparseMatrix(i = 1:n, j = 1:n, x = 1) # Diagonal elements set to 1 for Zi^2
# Define the quadratic constraint for the sum of Zi^2 being less than m
model$quadcon <- list(list(Qc = Qc, sense = "<", rhs = m))
params <- list(MIPGap = 0.01, TimeLimit = 300, BarConvTol = 1e-8, BarQCPConvTol = 1e-8)
result <- gurobi(model, params)
print(result)
-
Official comment
-
Gurobi Staff
This post is more than two years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum, or try Gurobot, our chatbot interface offering instant, expert-level support. -
-
-
Please refer to the Knowledge Base article How do I determine why my model is infeasible? It is probably best to compute an IIS. In R you can compute an IIS for your infeasible model via the gurobi_iis() function.
It might also make sense to use the gurobi_write() function to write your model to a human-readable LP file and inspect this file by hand to try to find a mistake in the formulation.
Best regards,
Jaromił0 -
Post is closed for comments.
Comments
2 comments