问题

我正在尝试使用CP-SAT求解器重新创建Google OR-Tools的Stigler Diet示例，而不是使用线性求解器，但结果显示状态为MODEL_INVALID。我不知道如何获取详细的错误消息或获取有关为什么模型无效的任何其他详细信息。

查看Python源代码时，在MODEL_INVALID定义旁边的代码注释中说：“给定的CpModelProto未通过验证步骤。您可以通过调用ValidateCpModel(model_proto)来获取详细的错误信息。” 但是，我无法弄清楚如何访问ValidateCpModel()，它在文档或or-tools存储库中的任何示例中都没有被引用。

供参考，这是我正在尝试运行的程序：

from ortools.sat.python import cp_model
import sys

# 每种营养素的最小和最大值
nutritional_requirements = [
    [1, 2],  # 维生素A
    [1, 2],  # 维生素B
    [1, 2],  # 维生素C
]

# 每种食物的营养价值
foods = [
  # 维生素
  #  A  B  C
    [1, 0, 0],  # 食物A
    [0, 1, 0],  # 食物B
    [0, 0, 1],  # 食物C
]

model = cp_model.CpModel()

quantity_of_food = [model.NewIntVar(0, sys.maxsize, str(i)) for i in range(len(foods))]

for i, nutrient in enumerate(nutritional_requirements):
    model.Add(
        sum([food[i] * quantity_of_food[i] for food in foods]) >= nutrient[0]
    )
    model.Add(
        sum([food[i] * quantity_of_food[i] for food in foods]) <= nutrient[1]
    )

model.Minimize(sum(quantity_of_food))

solver = cp_model.CpSolver()
status = solver.Solve(model)

outcomes = [
    "UNKNOWN",
    "MODEL_INVALID",
    "FEASIBLE",
    "INFEASIBLE",
    "OPTIMAL",
]

print(outcomes[status])  # 输出 "MODEL_INVALID"

对我来说，这个程序看起来相当简单。我如何找到详细的错误消息，解释为什么模型无效？

from ortools.sat.python import cp_model
import sys


# Minimum and maximum values for each nutrient
nutritional_requirements = [  
    [1, 2],  # Vitamin A
    [1, 2],  # Vitamin B
    [1, 2],  # Vitamin C
]

# Nutritional value for each food
foods = [
  # Vitamins
  #  A  B  C
    [1, 0, 0],  # Food A
    [0, 1, 0],  # Food B
    [0, 0, 1],  # Food C
]

model = cp_model.CpModel()

quantity_of_food = [model.NewIntVar(0, sys.maxsize, str(i)) for i in range(len(foods))]

for i, nutrient in enumerate(nutritional_requirements):
    model.Add(
        sum([food[i] * quantity_of_food[i] for food in foods]) >= nutrient[0]
    )
    model.Add(
        sum([food[i] * quantity_of_food[i] for food in foods]) < nutrient[1]
    )

model.Minimize(sum(quantity_of_food))

solver = cp_model.CpSolver()
status = solver.Solve(model)

outcomes = [
    "UNKNOWN",
    "MODEL_INVALID",
    "FEASIBLE",
    "INFEASIBLE",
    "OPTIMAL",
]

print(outcomes[status])  # Prints "MODEL_INVALID"

答案1

得分: 3

有两个选项：
- solver.parameters.log_search_progress = True

或

- print(model.Validate())

输出

```plaintext
无效模型：变量＃0的域不在[kint64min + 2，kint64max - 1]范围内。名称：“0”域：0域：9223372036854775807

CP-SAT求解器拒绝可能导致整数算术溢出的模型。
特别是，变量的域应远离int64max。

将变量的域尽量减小是一个良好的实践。这样可以简化之后的求解过程。

<details>
<summary>英文:</summary>

There are two options:
  - solver.parameters.log_search_progress = True

or

  - print(model.Validate())



outputs 

Invalid model: var #0 domain do not fall in [kint64min + 2, kint64max - 1]. name: "0" domain: 0 domain: 9223372036854775807

CP-SAT solvers rejects models that can overflow integer arithmetic.
In particular, domains of variables should stay away from int64max.

It is good practice to reduce the domain of variables as much as possible. It simplifies solving afterwards.

</details>