DOcplex复现CPLEX ILOG目标函数遇IndexError问题求助

问题背景 #

尝试在DOcplex中复现CPLEX ILOG的目标函数，但运行时持续报错，因不明确type_of_next()函数与索引的配合逻辑，调试受阻。目标是构建可添加至模型的目标函数表达式，后续将执行DARP_cp.add(DARP_cp.minimize(totDistance))。

CPLEX ILOG原代码 #

dexpr int totDistance =
(sum(j in Jobs, v in Vehicles) Dist[j.id][typeOfNext(seqVeh[v], itvJ2V[j][v], j.id, j.id)]) ;

DOcplex复现代码 #

totDistance = DARP_cp.sum(distance_matrix[i][DARP_cp.type_of_next(x[k], zeta[(i,k)], i, i)]
for i in range(nrows) for k in range(1, no_vehicles + 1))

错误信息 #

Traceback (most recent call last):
File "C:\Users\bsraf3\OneDrive - Loughborough University\CPLEX TUTORIALS\DOcplex Tutorials\DARP\DARP_CP.py", line 115, in 
for i in range(nrows) for k in range(1, no_vehicles + 1))
File "C:\ProgramData\Anaconda3\lib\site-packages\docplex\cp\modeler.py", line 567, in sum_of
arr = build_cpo_expr(x)
File "C:\ProgramData\Anaconda3\lib\site-packages\docplex\cp\expression.py", line 2390, in build_cpo_expr
return build_cpo_expr_array(val)
File "C:\ProgramData\Anaconda3\lib\site-packages\docplex\cp\expression.py", line 2420, in build_cpo_expr_array
cpval = _CPO_VALUES_FROM_PYTHON.get_or_create('array', val, normalize_value, _create_cpo_array_expr)
File "C:\ProgramData\Anaconda3\lib\site-packages\docplex\cp\expression.py", line 2300, in get_or_create
kval = kbldr(pexpr)
File "C:\ProgramData\Anaconda3\lib\site-packages\docplex\cp\expression.py", line 2409, in normalize_value
return _CacheKeyTuple(val)
File "C:\Users\bsraf3\OneDrive - Loughborough University\CPLEX TUTORIALS\DOcplex Tutorials\DARP\DARP_CP.py", line 115, in 
for i in range(nrows) for k in range(1, no_vehicles + 1))

IndexError: only integers, slices (:), ellipsis (...), numpy.newaxis (None) and integer or boolean arrays are valid indices

错误原因排查与修正建议 #

• 索引类型不匹配：DARP_cp.type_of_next()返回的是CPO决策变量表达式，而非Python整数。直接用它作为distance_matrix的索引会触发错误，因为Python列表/数组仅接受整数类索引。
• ILOG与DOcplex逻辑差异：ILOG OPL会自动处理决策变量表达式与数组索引的绑定，但DOcplex中需要显式使用element()函数关联决策变量索引和数组值。
• 代码修正示例：

  totDistance = DARP_cp.sum(
      DARP_cp.element(distance_matrix[i], DARP_cp.type_of_next(x[k], zeta[(i,k)], i, i))
      for i in range(nrows) for k in range(1, no_vehicles + 1)
  )
  

• 参数验证：确保zeta[(i,k)]是对应任务的区间变量（与ILOG中itvJ2V[j][v]语义一致），x[k]是车辆的序列变量（对应ILOG的seqVeh[v]），且type_of_next()的参数完全符合type_of_next(sequence, interval, type, default)的签名要求。

内容的提问来源于stack exchange，提问作者Richlove Frimpong