从AMPL转换为Pyomo

Question

我正在尝试将AMPL模型转换为Pyomo（我没有使用过的经验）。我发现语法难以适应，尤其是约束和客观部分。我已经将我的计算机与python，anaconda，Pyomo和GLPK联系在一起，只需要将实际代码关闭。我是初学者编码器，请原谅我，如果我的代码写得不好。仍然试图掌握这一点！

以下是AMPL代码中的数据：

set PROD := 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30;

set PROD1:= 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30;

ProdCost    414 3   46  519 876 146 827 996 922 308 568 176 58  13  20  974 121 751 130 844 280 123 275 843 717 694 72  413 65  631

HoldingCost 606 308 757 851 148 867 336 44  364 960 69  428 778 485 285 938 980 932 199 175 625 513 536 965 366 950 632 88  698 744

Demand  105 70  135 67  102 25  147 69  23  84  32  41  81  133 180 22  174 80  24  156 28  125 23  137 180 151 39  138 196 69

这是模型：

set PROD;  # set of production amounts
set PROD1; # set of holding amounts

param ProdCost {PROD} >= 0;     # parameter set of production costs

param Demand {PROD} >= 0;     # parameter set of demand at each time

param HoldingCost {PROD} >= 0;     # parameter set of holding costs

var Inventory {PROD1} >= 0;     # variable that sets inventory amount at each time

var Make {p in PROD} >= 0;  # variable of amount produced at each time

minimize Total_Cost: sum {p in PROD} ((ProdCost[p] * Make[p]) + (Inventory[p] * HoldingCost[p]));

               # Objective: minimize total cost from all production and holding cost

subject to InventoryConstraint {p in PROD}: Inventory[p] = Inventory[p-1] + Make[p] - Demand[p];

                # excess production transfers to inventory

subject to MeetDemandConstraint {p in PROD}: Make[p] >= Demand[p] - Inventory[p-1];

               # Constraint: holding and production must exceed demand

subject to InitialInventoryConstraint: Inventory[0] = 0;

                # Constraint: Inventory must start at 0

这是我到目前为止所拥有的。不确定它是否正确：

from pyomo.environ import *

demand=[105,70,135,67,102,25,147,69,23,84,32,41,81,133,180,22,174,80,24,156,28,125,23,137,180,151,39,138,196,69]

holdingcost=[606,308,757,851,148,867,336,44,364,960,69,428,778,485,285,938,980,932,199,175,625,513,536,965,366,950,632,88,698,744]

productioncost=[414,3,46,519,876,146,827,996,922,308,568,176,58,13,20,974,121,751,130,844,280,123,275,843,717,694,72,413,65,631]

model=ConcreteModel()

model.I=RangeSet(1,30)
model.J=RangeSet(0,30)
model.x=Var(model.I, within=NonNegativeIntegers)
model.y=Var(model.J, within=NonNegativeIntegers)

model.obj = Objective(expr = sum(model.x[i]*productioncost[i]+model.y[i]*holdingcost[i] for i in model.I))

def InventoryConstraint(model, i):
    return model.y[i-1] + model.x[i] - demand[i] <= model.y[i]
InvCont = Constraint(model, rule=InventoryConstraint)

def MeetDemandConstraint(model, i):
    return model.x[i] >= demand[i] - model.y[i-1]
DemCont = Constraint(model, rule=MeetDemandConstraint)

def Initial(model):
    return model.y[0] == 0
model.Init = Constraint(rule=Initial)

opt = SolverFactory('glpk')
results = opt.solve(model,load_solutions=True)
model.solutions.store_to(results)
results.write()

谢谢！

Answer 1

我看到的唯一问题是你的一些约束声明。您需要将约束附加到模型，并且传入的第一个参数应该是索引集（我假设应该是model.I）。

def InventoryConstraint(model, i):
    return model.y[i-1] + model.x[i] - demand[i] <= model.y[i]
model.InvCont = Constraint(model.I, rule=InventoryConstraint)

def MeetDemandConstraint(model, i):
    return model.x[i] >= demand[i] - model.y[i-1]
model.DemCont = Constraint(model.I, rule=MeetDemandConstraint)

您用来解决模型的语法有点过时但应该有效。另一种选择是：

opt = SolverFactory('glpk')
opt.solve(model,tee=True) # The 'tee' option prints the solver output to the screen
model.display() # This will print a summary of the model solution

另一个对调试有用的命令是model.pprint()。这将显示整个模型，包括约束和目标的表达式。