using Gurobi,JuMP
include("../src/RDDP.jl")
N = 10
msro = RDDP.buildMultiStageRobustModel(
N_stage = N, # number of stages
optimizer = Gurobi.Optimizer, # choose an optimizer
MaxIteration = 100, # maximum epoches
MaxTime = 60, # maximum trainning time
Gap = 0.01, # optimal gap
use_maxmin_solver = false # using a vertex enumeration by default
) do ro::JuMP.Model,t::Int
# defining operator's strategy
@variable(ro,0 ≤ x ≤ 5,RDDP.State,initial_value = 0) # state variable (stock level)
@variable(ro,0 ≤ u ≤ 1) # control variable (supply)
@variable(ro,0 ≤ ξ[i = 1:2] ≤ 1,RDDP.Uncertain) # uncertain parameter (demand)
@constraint(ro,sum(ξ) ≤ 1) # constraint abount uncertain parameter
@constraint(ro, x.out == x.in - u + sum(ξ)) # state transition (temporal dependence of stock level)
@objective(ro,Min,x) # minimize the stock
end
RDDP.train(msro)
报错
ERROR: The objective function Main.RDDP.State{VariableRef}(x_in, x_out) is not supported by JuMP.
是什么原因，我该怎么修改
To fix this issue and create a more realistic inventory management model, we need to modify the objective function. Here's how we can do it:
Here's the modified code:
using Gurobi, JuMP
include("../src/RDDP.jl")
N = 10
msro = RDDP.buildMultiStageRobustModel(
N_stage = N, # number of stages
optimizer = Gurobi.Optimizer, # choose an optimizer
MaxIteration = 100, # maximum epoches
MaxTime = 60, # maximum trainning time
Gap = 0.01, # optimal gap
use_maxmin_solver = false # using a vertex enumeration by default
) do ro::JuMP.Model, t::Int
# defining operator's strategy
@variable(ro, 0 ≤ x ≤ 5, RDDP.State, initial_value = 0) # state variable (stock level)
@variable(ro, 0 ≤ u ≤ 1) # control variable (supply)
@variable(ro, 0 ≤ ξ[i = 1:2] ≤ 1, RDDP.Uncertain) # uncertain parameter (demand)
# Cost parameters
holding_cost = 0.1 # cost per unit of inventory held
ordering_cost = 0.5 # cost per unit ordered
@constraint(ro, sum(ξ) ≤ 1) # constraint about uncertain parameter
@constraint(ro, x.out == x.in - u + sum(ξ)) # state transition (temporal dependence of stock level)
# Objective: Minimize total cost (holding cost + ordering cost)
@objective(ro, Min, holding_cost * x.out + ordering_cost * u)
end
RDDP.train(msro)In this modified version:
holding_cost and ordering_cost parameters. You can adjust these values based on your specific problem.The objective function has been changed to:
@objective(ro, Min, holding_cost * x.out + ordering_cost * u)This minimizes the total cost, which is the sum of:
holding_cost * x.out (cost of holding inventory at the end of the period)ordering_cost * u (cost of ordering supplies)x.out in the objective function to represent the inventory level at the end of each period.This formulation should resolve the error you were encountering and provide a more realistic model for inventory management under uncertain demand. The model now balances the costs of holding inventory against the costs of ordering supplies, while dealing with uncertain demand.
If you encounter any further issues or need additional modifications, please let me know.
在 VS Code 中使用 Julia，有几种方法可以帮助你理解函数之间的依赖关系和执行顺序：
使用 VSCode 的 Outline 视图：
使用 Julia 的 @which 宏：
@which function_name(args...)使用 VSCode 的 "Go to Definition" 功能：
使用 Julia 的 @code_warntype 宏：
@code_warntype function_name(args...)使用 Julia 的 @trace 宏（需要 Traceur.jl 包）：
using Pkg; Pkg.add("Traceur")using Traceur; @trace function_name(args...)使用 VSCode 的调试器：
创建依赖图：
using GraphViz
function create_dependency_graph(functions)
g = DiGraph()
for (i, f) in enumerate(functions)
add_vertex!(g, f)
for j in 1:i-1
if depends_on(f, functions[j])
add_edge!(g, f, functions[j])
end
end
end
return g
end
# 假设的依赖检查函数
depends_on(f, g) = # 实现依赖检查逻辑
# 使用
functions = [func1, func2, func3, ...]
graph = create_dependency_graph(functions)
GraphViz.plot(graph)使用 Julia 的 Profile 工具：
@profile 宏和 Profile 模块来分析函数调用示例：
using Profile
@profile your_main_function()
Profile.print()这些方法可以帮助你更好地理解复杂的 Julia 代码结构和函数依赖关系。根据你的具体需求，可以选择最适合的方法或结合使用多种方法。