python - 将 scipy linprog 中的设置与 PuLP 匹配

Question

我在 SciPy linprog 中设置了一个相当简单的 LP 模型，但在 PuLP 中设置它时遇到了麻烦。

我计划增加更多的复杂性，这是 linprog 无法处理的，所以想在其中进行设置。虽然看起来不像 linprog 那样用户友好！

linprog 设置如下：

import pandas as pd
df = pd.read_excel('df.xlsx')

from scipy.optimize import linprog

#1d n-element array of ones, *-1 to minimize the sum of the variables to solve for: x1, x2 ... xn
constants=np.array(df['Ones'])*(-1)

# equations for the inequality constraints, simply a M x n array
A_ub=df.loc[:,'col_a':'col_n'].to_numpy().transpose()

# array of M zeros as bounds for above inequality constraint
b_ub = np.array((A_ub.shape[0])*[0])

# lower and upper bounds set for the elements
bounds=[(0,x) for x in list(df['bounds_max'])]

# the set up of the equation
main_result=linprog(constants, A_ub=A_ub, b_ub=b_ub, A_eq=None, b_eq=None, bounds=bounds, method='interior-point', callback=None, options=None, x0=None)

以上工作正常，但不知道如何在纸浆中设置相同？到目前为止我的尝试：

# importing the package
import pulp as pl

# defining the problem
prob = pl.LpProblem("df", pl.LpMinimize)

# defining the set of variables to solve for like above (x1, x2, ... xn) with the same constraints like in 'bounds' above. different 'x's have different upper bounds
x = pl.LpVariable.matrix('x', list(range(len(df['Helper']))), 0, [i for i in list(df['bounds_max'])])

# the equivalent to 'constant' above - minimize the sum of 'x1, x2, ... xn'
prob += pl.lpSum(x for j in range(len(list(df['Ones']))))

# the same equations for the inequality constraints
A_ub=df.loc[:,'col_a':'col_a'].to_numpy().transpose()

# here is where it breaks down, the multiplying each 'x' element be each element in each row of the array
prob += pl.lpSum(x*[a for a in A_ub]) <= 0

# Solve the problem
prob.solve()

从倒数第二行收到错误消息“TypeError: can't multiply sequence by non-int of type 'list'”。此外，上限的列表推导不像 linprog 那样工作。（TypeError：必须是实数，而不是生成器）

有谁知道正确的设置吗？

标签： pythonscipylinear-programmingpulp