在 Pyomo 中实现复杂约束

Question

我试图在 Pyomo 中实现一个相当复杂的约束并遇到错误。问题表述如下：

• 决策变量位于 C × A 矩阵中，包含在较大矩阵的右上象限中
• 对于复杂约束，必须迭代矩阵，并对每行左下象限（C x A 矩阵）中的值进行平均。这些值将用作复杂约束的 LHS，而 RHS 将是一个常数值。

当前的工作（最后的代码块）导致以下错误：

ERROR: Rule failed when generating expression for Constraint c4 with index A0:
KeyError: "Index '(0, 0)' is not valid for indexed component 'x'"
ERROR: Constructing component 'c4' from data=None failed: KeyError: "Index
'(0, 0)' is not valid for indexed component 'x'"

我不确定为什么在这种情况下会导致错误。想法？

def constructAndIterate(m):
    # Create the SuperMatrix including Decision Variables as Upper Right
    newTopRight = np.ndarray((len(C), len(A)))
    
    # Use the decision variables as matrix values
    for c, i in enumerate(C):
        for subA, j in enumerate(A):
            newTopRight[i,j] = m.x[c,subA]
 
    wNewTop = np.hstack((topLeft, newTopRight))

    nW = np.vstack((wNewTop, wBot))
        
    return iterateMatrix(nW)

model = pyo.ConcreteModel()

# Decision Variables
A = [ "A" + str(x) for x in range(10)]
C = [ "C" + str(x) for x in range(6)]

model.x = pyo.Var(C, A, domain = pyo.Reals)

@model.Objective(sense=pyo.minimize)
def cost(m):
    return sum([m.x[c,a]**2.0 for a in A for c in C])

# Sum Columns of decisions vars
@model.Constraint(A)
def c1(m, a):
    return sum([m.x[c,a] for c in C]) == 0

# Decision Var + W0  >= 0
@model.Constraint(C, A)
def c2(m, c, a):
    # Extracts index from a and c -> "A0" becomes 0a
    return m.x[c,a] + W0[int(c[1:])][int(a[1:])] >= 0


# Decision Var + W1 == 1
@model.Constraint(C,A)
def c3(m, c, a):
    return m.x[c,a] + W0[int(c[1:])][int(a[1:])] == 1


# Limiting Priorites for C >= 0

@model.Constraint(A)
def c4(m, a):
   
    _, _, _, _aVals = constructAndIterate(m)
    
    # Create appropriate constraints
    return _aVals[a] <= origPref

Answer 1

您正在翻转

enumerate()

的输出。

您所在的地方：

for c, i in enumerate(C):

你应该有

for i, c in enumerate(C):

你没有发布

enumerateMatrix

是什么/做什么，所以 YMMV。