StochasticOptim/dtu02435a3_1.jl at master · jiedxu/StochasticOptim · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
# DTU02435 Decision Making under Uncertainty
# Assignment 3, Task 2
# Edward J. Xu
# May 5th, 2019
push!(LOAD_PATH, "$(homedir())/Desktop/StochasticAirplaneRent")
cd("$(homedir())/Desktop/StochasticAirplaneRent")
using JuMP
# using GLPKMathProgInterface
# using Gurobi
using CPLEX
using CSV


function processData()
    datf_numMax = CSV.read("Data/MaxPlanes.csv", header = ["p", "num"], delim = ',')
    datf_cost = CSV.read("Data/OperationCosts.csv", header = true, delim = ',')
    datf_cap = CSV.read("Data/PlaneCapacity.csv", header = ["p", "cap"], delim = ',')
    datf_route = CSV.read("Data/DemandsRoute.csv", header = true, delim = ',')
    datf_costAdmin = CSV.read("Data/CostAdquisition.csv", header = ["p", "cost"], delim = ',')
    datf_prob = CSV.read("Data/Probability.csv", header = ["prob"], delim = ',')
    datf_demand = CSV.read("Data/Scenario.csv", header = true, delim = ',')
    datf_expect = CSV.read("Data/ExpectedValue.csv", header = ["value"], delim = ',')
    # Calculate the demand matrix
    mat4_demand = zeros(4, 10, 12, 10)
    for i = 1: 4
        for j = 1: 10
            for w = 1: 10
                for t = 1: 12
                    mat4_demand[i, j, t, w] = 1000 * datf_demand[t, w] * datf_route[i, j+1]
                end
            end
        end
    end
    mat3_demandExpect = zeros(4, 10, 12)
    for i = 1: 4
        for j = 1: 10
            for t = 1: 12
                mat3_demandExpect[i, j, t] = 1000 * datf_expect.value[t] * datf_route[i, j+1]
            end
        end
    end
    return (datf_numMax.num, datf_cost, datf_cap.cap, datf_costAdmin.cost, mat4_demand,
            datf_prob.prob, mat3_demandExpect)
end


function solvePrintResult(model, variable_y, variable_m)
    println("---------------------------- 1/2. Begin Optimization ---------------------------")
    timeStart = time()
    solve(model)
    println("Elapsed time: $(time()-timeStart) seconds.")
    ## 2,  Print the Result
    objResult = getobjectivevalue(model)
    vec_yResult = getvalue(variable_y)
    vec_mResult = getvalue(variable_m)
    println("Result: obj = $(objResult)\n",
            "Result: vec_y = $(vec_yResult)\n",
            "Result: total unfulfilled passenger demand = $(sum(vec_mResult))\n",
            "---------------------------------- 2/2. Ending ---------------------------------")
    return vec_yResult
end


function optim_determin(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat3_demandExpect)
    ## 1,  Begin Optimization
    model = Model(solver = CplexSolver(CPX_PARAM_SCRIND = 0))
    # model = Model(solver = GurobiSolver(Presolve=0))
    # model = Model(solver = GLPKSolverMIP())
    @variable(model, vec_y[1: 10] >= 0, Int)
    @variable(model, mat4_z[1: 4, 1: 10, 1: 10, 1: 12] >= 0, Int)
    @variable(model, mat3_m[1: 4, 1: 10, 1: 12] >= 0)
    @objective(model, Min, sum(vec_y[p] * vec_costAdmin[p] for p = 1: 10) +
        sum(datf_cost.cost[.&((datf_cost[:i] .== i), (datf_cost[:j] .== j), (datf_cost[:p] .== p))][1] *
        mat4_z[i, j, p, t] for p = 1: 10, i = 1: 4, j = 1: 10, t = 1: 12) +
        sum(mat3_m[i, j, t] * 1000 for i = 1: 4, j = 1: 10, t = 1: 12))
    @constraint(model, [p = 1: 10], vec_y[p] <= vec_numMax[p])
    @constraint(model, [i = 1: 4, j = 1: 10, t = 1: 12], mat3_demandExpect[i, j, t] -
        sum([mat4_z[i, j, p, t] * vec_cap[p]] for p = 1: 10)[1] <= mat3_m[i, j, t])
    @constraint(model, [p = 1: 10, t = 1: 12], sum(mat4_z[i, j, p, t] for i = 1: 4, j = 1: 10) <= vec_y[p])
    vec_yResult = solvePrintResult(model, vec_y, mat3_m)
    return vec_yResult
end


function optim(vec_y, datf_cost, vec_cap, vec_costAdmin, mat_demand)
    ## Most basic function to optimize the arrangement of plane for any month in any scenario, with fixed fleet.
    model = Model(solver = CplexSolver(CPX_PARAM_SCRIND = 0))
    # model = Model(solver = GurobiSolver(Presolve=0))
    # model = Model(solver = GLPKSolverMIP())
    @variable(model, mat3_z[1:4, 1:10, 1:10] >= 0, Int)
    @variable(model, mat_m[1:4, 1:10] >= 0)
    @objective(model, Min, sum(datf_cost.cost[.&((datf_cost[:i] .== i), (datf_cost[:j] .== j),
        (datf_cost[:p] .== p))][1] * mat3_z[i, j, p] for p = 1:10, i = 1:4, j = 1:10) + sum(mat_m) * 1000)
    @constraint(model, [i = 1:4, j = 1:10], mat_demand[i, j] - sum(mat3_z[i, j, p] * vec_cap[p] for p = 1:10)
        <= mat_m[i, j])
    @constraint(model, [p = 1:10], sum(mat3_z[i, j, p] for i = 1:4, j = 1:10) <= vec_y[p])
    solve(model)
    objResult = getobjectivevalue(model)
    vec_mResult = getvalue(mat_m)
    return (objResult, sum(vec_mResult))
end


function optim_determinStochas(vec_yResult, datf_cost, vec_cap, vec_costAdmin, mat4_demand, vec_prob)
    # Function to calculate the obj when first stage variable is fixed by solutions from deterministic program, using
    # stochastic model. The program is decomposed into programs for different months.
    mat_objResult = zeros(10, 12)
    vec_objResult = zeros(10)
    mat_mResult = zeros(10, 12)
    vec_mResult = zeros(10)
    ## 1,  Begin Optimization
    println("---------------------------- 1/2. Begin Calculation ----------------------------\n")
    timeStart = time()
    for w = 1:10
        for t = 1:12
            (mat_objResult[w, t], mat_mResult[w, t]) = optim(vec_yResult, datf_cost, vec_cap, vec_costAdmin,
                mat4_demand[:, :, t, w])
        end
        vec_objResult[w] = sum(mat_objResult[w, tt] for tt = 1:12) +
            sum(vec_yResult[p] * vec_costAdmin[p] for p = 1: 10)
        vec_mResult[w] = sum(mat_mResult[w, tt] for tt = 1:12)
    end
    objResult = sum(vec_objResult[w] * vec_prob[w] for w = 1:10)
    mResult = sum(vec_mResult)
    println("Elapsed time: $(time()-timeStart) seconds.\n",
            "Result: obj = $(objResult)\n",
            "Result: vec_y = $(vec_yResult)\n",
            "Result: total unfulfilled passenger demand = $(mResult)\n",
            "---------------------------------- 2/2. Ending ---------------------------------")
end


function optim_stochas(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat4_demand, vec_prob)
    ## 1,  Begin Optimization
    model = Model(solver = CplexSolver(CPX_PARAM_SCRIND = 0))
    # model = Model(solver = GurobiSolver(Presolve=0))
    # model = Model(solver = GLPKSolverMIP())
    @variable(model, vec_y[1: 10] >= 0, Int)
    @variable(model, mat5_z[1: 4, 1: 10, 1: 10, 1: 12, 1: 10] >= 0, Int)
    @variable(model, mat4_m[1: 4, 1: 10, 1: 12, 1: 10] >= 0)
    @objective(model, Min, sum(vec_y[p] * vec_costAdmin[p] for p = 1: 10) +
        sum((datf_cost.cost[.&((datf_cost[:i] .== i), (datf_cost[:j] .== j), (datf_cost[:p] .== p))][1] *
        mat5_z[i, j, p, t, w] + mat4_m[i, j, t, w] * 1000) * vec_prob[w] for w = 1: 10, p = 1: 10, i = 1: 4,
        j = 1: 10, t = 1: 12))
    @constraint(model, [p = 1: 10], vec_y[p] <= vec_numMax[p])
    @constraint(model, [i = 1: 4, j = 1: 10, t = 1: 12, w = 1: 10], mat4_demand[i, j, t, w] -
        sum([mat5_z[i, j, p, t, w] * vec_cap[p]] for p = 1: 10)[1] <= mat4_m[i, j, t, w])
    @constraint(model, [p = 1: 10, t = 1: 12, w = 1: 10], sum(mat5_z[i, j, p, t, w] for i = 1: 4, j = 1: 10) <=
        vec_y[p])
    solvePrintResult(model, vec_y, mat4_m)
end


# function optim_stochas_2(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat4_demand, vec_prob)
#     ## 1,  Begin Optimization
#     model = Model(solver = CplexSolver(CPX_PARAM_SCRIND = 0))
#     # model = Model(solver = GurobiSolver(Presolve=0))
#     # model = Model(solver = GLPKSolverMIP())
#     @variable(model, mat_y[1: 10, 1: 10] >= 0, Int)
#     @variable(model, mat5_z[1: 4, 1: 10, 1: 10, 1: 12, 1: 10] >= 0, Int)
#     @variable(model, mat4_m[1: 4, 1: 10, 1: 12, 1: 10] >= 0)
#     @objective(model, Min, sum(mat_y[p, w] * vec_costAdmin[p] +
#         (datf_cost.cost[.&((datf_cost[:i] .== i), (datf_cost[:j] .== j), (datf_cost[:p] .== p))][1] *
#         mat5_z[i, j, p, t, w] + mat4_m[i, j, t, w] * 1000) * vec_prob[w] for w = 1: 10, p = 1: 10, i = 1: 4,
#         j = 1: 10, t = 1: 12))
#     @constraint(model, [p = 1:10, w = 1:10], mat_y[p, w] <= vec_numMax[p])
#     @constraint(model, [i = 1:4, j = 1:10, t = 1:12, w = 1:10], mat4_demand[i, j, t, w] -
#         sum([mat5_z[i, j, p, t, w] * vec_cap[p]] for p = 1:10)[1] <= mat4_m[i, j, t, w])
#     @constraint(model, [p = 1:10, t = 1: 12, w = 1: 10], sum(mat5_z[i, j, p, t, w] for i = 1: 4, j = 1: 10) <=
#         mat_y[p, w])
#     @constraint(model, [p = 1:10, w = 2:10], mat_y[p, 1] == mat_y[p, w])
#     solvePrintResult(model, mat_y, mat4_m)
# end


function optim_waitSeeSub(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat3_demand)
    model = Model(solver = CplexSolver(CPX_PARAM_SCRIND = 0))
    # model = Model(solver = GurobiSolver(Presolve=0))
    # model = Model(solver = GLPKSolverMIP())
    @variable(model, vec_y[1:10] >= 0, Int)
    @variable(model, mat4_z[1:4, 1:10, 1:10, 1:12] >= 0, Int)
    @variable(model, mat3_m[1:4, 1:10, 1:12] >= 0)
    @objective(model, Min, sum(vec_y[p] * vec_costAdmin[p] for p = 1:10) +
        sum(datf_cost.cost[.&((datf_cost[:i] .== i), (datf_cost[:j] .== j), (datf_cost[:p] .== p))][1] *
        mat4_z[i, j, p, t] + mat3_m[i, j, t] * 1000 for p = 1: 10, i = 1: 4, j = 1: 10, t = 1: 12))
    @constraint(model, [p = 1:10], vec_y[p] <= vec_numMax[p])
    @constraint(model, [i = 1:4, j = 1:10, t = 1:12], mat3_demand[i, j, t] -
        sum(mat4_z[i, j, p, t] * vec_cap[p] for p = 1:10) <= mat3_m[i, j, t])
    @constraint(model, [p = 1: 10, t = 1: 12], sum(mat4_z[i, j, p, t] for i = 1: 4, j = 1: 10) <= vec_y[p])
    solve(model)
    objResult = getobjectivevalue(model)
    vec_yResult = getvalue(vec_y)
    vec_mResult = getvalue(mat3_m)
    println("Result: obj = $(objResult)\n",
            "Result: vec_y = $(vec_yResult)\n",
            "Result: unfulfilled passenger demand = $(sum(vec_mResult))")
    return objResult
end


function optim_waitSee(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat4_demand, vec_prob)
    ## 1,  Begin Optimization
    vec_objResult = zeros(10)
    for w = 1:10
        println("----- $(w)-th Sub-Problem -----")
        vec_objResult[w] = optim_waitSeeSub(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat4_demand[:,:,:,w])
    end
    objWeightedSum = sum(vec_prob[i] .* vec_objResult[i] for i = 1:10)
    println("----- Final Result -----\n",
            "Result: weighted sum of obj = $(objWeightedSum)")
end


function optim_robust(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat4_demand)
    ## 1,  Begin Optimization
    model = Model(solver = CplexSolver(CPX_PARAM_SCRIND = 0))
    # model = Model(solver = GurobiSolver(Presolve=0))
    # model = Model(solver = GLPKSolverMIP())
    @variable(model, vec_y[1: 10] >= 0, Int)
    @variable(model, mat5_z[1: 4, 1: 10, 1: 10, 1: 12, 1: 10] >= 0, Int)
    @variable(model, mat4_m[1: 4, 1: 10, 1: 12, 1: 10] >= 0)
    @variable(model, alpha >= 0)
    @objective(model, Min, sum(vec_y[p] * vec_costAdmin[p] for p = 1: 10) + alpha)
    @constraint(model, [w = 1: 10],
        sum((datf_cost.cost[.&((datf_cost[:i] .== i), (datf_cost[:j] .== j), (datf_cost[:p] .== p))][1] *
        mat5_z[i, j, p, t, w] + mat4_m[i, j, t, w] * 1000) for p = 1: 10, i = 1: 4, j = 1: 10, t = 1: 12) <= alpha)
    @constraint(model, [p = 1: 10], vec_y[p] <= vec_numMax[p])
    @constraint(model, [i = 1: 4, j = 1: 10, t = 1: 12, w = 1: 10], mat4_demand[i, j, t, w] -
        sum([mat5_z[i, j, p, t, w] * vec_cap[p]] for p = 1: 10)[1] <= mat4_m[i, j, t, w])
    @constraint(model, [p = 1: 10, t = 1: 12, w = 1: 10], sum(mat5_z[i, j, p, t, w] for i = 1: 4, j = 1: 10) <=
        vec_y[p])
    solvePrintResult(model, vec_y, mat4_m)
end


function main()
    (vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat4_demand, vec_prob, mat3_demandExpect) = processData()
    println("################################################################################\n",
            "######################## 1/5, Deterministic Programming ########################\n",
            "################################################################################")
    vec_yResult = optim_determin(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat3_demandExpect)
    # vec_yResult = [0.0, 1.0, 2.0, 2.0, 0.0, 12.0, 5.0, 4.0, 8.0, 8.0]
    # optim_determinStochas(vec_yResult, datf_cost, vec_cap, vec_costAdmin, mat4_demand, vec_prob)
    # println("################################################################################\n",
    #         "########################## 2/5, Stochastic Programming #########################\n",
    #         "################################################################################")
    # optim_stochas(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat4_demand, vec_prob)
    # println("################################################################################\n",
    #         "########################### 3/5, Wait-See Programming ##########################\n",
    #         "################################################################################")
    # optim_waitSee(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat4_demand, vec_prob)
    # println("################################################################################\n",
    #         "########################### 4/5, Robust Programming ############################\n",
    #         "################################################################################")
    # optim_robust(vec_numMax, datf_cost, vec_cap, vec_costAdmin, mat4_demand)
    println("################################################################################\n",
            "################################### 5/5, End ###################################\n",
            "################################################################################")
end


main()