Hi,

I use pulp module to solve my some optimization problems. With integer numbers, it works like a charm. I wonder if it is possible for pulp to work with real numbers.

Here is a sample optimization problem I build for my project.

prob = pulp.LpProblem("example", pulp.LpMinimize)

    # Variable represent number of times device i is used
    n1 = pulp.LpVariable("n1", 1, 5, pulp.LpInteger)
    n2 = pulp.LpVariable("n2", 0, max_exp, pulp.LpInteger)
    n3 = pulp.LpVariable("n3", 0, max_exp, pulp.LpInteger)
    n4 = pulp.LpVariable("n4", 0, max_exp, pulp.LpInteger)
    n5 = pulp.LpVariable("n5", 0, max_exp, pulp.LpInteger)


    # The objective function that we want to minimize: the total cost
    prob += n1 * Device1[-1] + n2 * Device2[-1] + n3 * Device3[-1] + n4 * Device4[-1] + n5 * Device5[-1], "Minimize total cost"

    # Constraint that we use no more than max. devices per controller device
    prob += n2 + n3 + n4 + n5 <= n1*max_exp

    #Constraint (total_UI >= (target_AI + target_BI))
    prob += n1 * Device1[0] + n2 * Device2[0] + n3 * Device3[0] + n4 * Device4[0] + n5 * Device5[0] >= (Target[0]+Target[2])

    #Constraint ((total_UO + total_BO) >= target_BO)
    prob += n1 * Device1[1] + n2 * Device2[1] + n3 * Device3[1] + n4 * Device4[1] + n5 * Device5[1] + n1 * Device1[3] + n2 * Device2[3] + n3 * Device3[3] + n4 * Device4[3] + n5 * Device5[3]>= (Target[3])

    #Constraint [total_UO + total_AO + total_BO - target_BO] >= target_AO
    prob += n1 * Device1[1] + n2 * Device2[1] + n3 * Device3[1] + n4 * Device4[1] + n5 * Device5[1] + n1 * Device1[2] + n2 * Device2[2] + n3 * Device3[2] + n4 * Device4[2] + n5 * Device5[2] + n1 * Device1[3] + n2 * Device2[3] + n3 * Device3[3] + n4 * Device4[3] + n5 * Device5[3] >= (Target[1]+Target[3])        
    
    # Actually solve the problem, this calls GLPK so you need it installed
    prob.solve()

Lets assume DeviceX[-1] be a real number. Could you make the required changes?

Thanks in advance.

If you would multiply all values by constant, say 1000000, and divide solution accordingly, maybe it could work out to transform the problem to integer realm.

If you would multiply all values by constant, say 1000000, and divide solution accordingly, maybe it could work out to transform the problem to integer realm.

This is not a good idea. LP problems are much easier to solve with real numbers than with integers, so yes I think pulp can work with real numbers, but since I don't use pulp, I can't make the required changes to your code. The solution is in the pulp documentation https://www.coin-or.org/PuLP/ .

This article has been dead for over six months. Start a new discussion instead.