ANL-CEEESA/MIPLearn
1
0
Fork 0
mirror of https://github.com/ANL-CEEESA/MIPLearn.git synced 2025-12-09 02:48:52 -06:00
Code Issues Packages Projects Releases Wiki Activity

Implement PMedianPerturber

Browse Source
This commit is contained in:
Alinson S. Xavier
2025-12-08 13:36:49 -06:00
parent 14e2fe331d
commit 427bd1d806
3 changed files with 104 additions and 99 deletions
Download Patch File Download Diff File Expand all files Collapse all files

84
docs/guide/problems.ipynb
View File

@@ -473,9 +473,10 @@
"\n",
"The class [PMedianGenerator][PMedianGenerator] can be used to generate random instances of this problem. First, it decides the number of customers and the parameter $p$ by sampling the provided `n` and `p` distributions, respectively. Then, for each customer $i$, the class builds its geographical location $(x_i, y_i)$ by sampling the provided `x` and `y` distributions. For each $i$, the demand for customer $i$ and the capacity of facility $i$ are decided by sampling the provided distributions `demands` and `capacities`, respectively. Finally, the costs $w_{ij}$ are set to the Euclidean distance between the locations of customers $i$ and $j$.\n",
"\n",
"If `fixed=True`, then the number of customers, their locations, the parameter $p$, the demands and the capacities are only sampled from their respective distributions exactly once, to build a reference instance which is then randomly perturbed. Specifically, in each perturbation, the distances, demands and capacities are multiplied by random scaling factors sampled from the distributions `distances_jitter`, `demands_jitter` and `capacities_jitter`, respectively. The result is a list of instances that have the same set of customers, but slightly different demands, capacities and distances.\n",
"To create multiple instances with the same customer locations but different values, you can use [PMedianPerturber][PMedianPerturber]. This class takes an existing PMedianData instance and generates new instances by applying randomization factors to the existing distances, demands, and capacities while keeping the graph structure and parameter p fixed. More specifically, the distances, demands and capacities are multiplied by random scaling factors sampled from the distributions `distances_jitter`, `demands_jitter` and `capacities_jitter`, respectively. The result is a list of instances that have the same set of customers, but slightly different demands, capacities and distances.\n",
"\n",
"[PMedianGenerator]: ../../api/problems/#miplearn.problems.pmedian.PMedianGenerator"
"[PMedianGenerator]: ../../api/problems/#miplearn.problems.pmedian.PMedianGenerator\n",
"[PMedianPerturber]: ../../api/problems/#miplearn.problems.pmedian.PMedianPerturber"
]
},
{
@@ -488,7 +489,7 @@
},
{
"cell_type": "code",
"execution_count": 3,
"execution_count": 7,
"id": "4e0e4223-b4e0-4962-a157-82a23a86e37d",
"metadata": {
"ExecuteTime": {
@@ -503,18 +504,18 @@
"text": [
"p = 5\n",
"distances =\n",
" [[ 0. 50.17 82.42 32.76 33.2 35.45 86.88 79.11 43.17 66.2 ]\n",
" [ 50.17 0. 72.64 72.51 17.06 80.25 39.92 68.93 43.41 42.96]\n",
" [ 82.42 72.64 0. 71.69 70.92 82.51 67.88 3.76 39.74 30.73]\n",
" [ 32.76 72.51 71.69 0. 56.56 11.03 101.35 69.39 42.09 68.58]\n",
" [ 33.2 17.06 70.92 56.56 0. 63.68 54.71 67.16 34.89 44.99]\n",
" [ 35.45 80.25 82.51 11.03 63.68 0. 111.04 80.29 52.78 79.36]\n",
" [ 86.88 39.92 67.88 101.35 54.71 111.04 0. 65.13 61.37 40.82]\n",
" [ 79.11 68.93 3.76 69.39 67.16 80.29 65.13 0. 36.26 27.24]\n",
" [ 43.17 43.41 39.74 42.09 34.89 52.78 61.37 36.26 0. 26.62]\n",
" [ 66.2 42.96 30.73 68.58 44.99 79.36 40.82 27.24 26.62 0. ]]\n",
"demands = [6.12 1.39 2.92 3.66 4.56 7.85 2. 5.14 5.92 0.46]\n",
"capacities = [151.89 42.63 16.26 237.22 241.41 202.1 76.15 24.42 171.06 110.04]\n",
" [[ 0. 54.88 80.58 34.54 35.61 32.75 78.74 75.78 45.82 65.11]\n",
" [ 54.88 0. 69.32 68.14 17.48 83.67 41.01 64.26 46.85 40.57]\n",
" [ 80.58 69.32 0. 64.6 68.52 86.81 65.36 4.08 38.29 28.39]\n",
" [ 34.54 68.14 64.6 0. 51.62 11.17 101.52 73.67 38.81 66.35]\n",
" [ 35.61 17.48 68.52 51.62 0. 67.13 59.17 68.95 32.99 48.98]\n",
" [ 32.75 83.67 86.81 11.17 67.13 0. 105.47 86.25 52.01 76.55]\n",
" [ 78.74 41.01 65.36 101.52 59.17 105.47 0. 69.09 65.27 40.97]\n",
" [ 75.78 64.26 4.08 73.67 68.95 86.25 69.09 0. 38.88 28.35]\n",
" [ 45.82 46.85 38.29 38.81 32.99 52.01 65.27 38.88 0. 25.89]\n",
" [ 65.11 40.57 28.39 66.35 48.98 76.55 40.97 28.35 25.89 0. ]]\n",
"demands = [6.69 1.32 2.92 3.51 4.36 7.12 2.04 5.14 5.39 0.44]\n",
"capacities = [164.29 40.41 15.11 236.72 264.86 191.67 78.77 25.7 162.08 115.06]\n",
"\n",
"Gurobi Optimizer version 13.0.0 build v13.0.0rc1 (linux64 - \"Ubuntu 22.04.5 LTS\")\n",
"\n",
@@ -522,15 +523,15 @@
"Thread count: 16 physical cores, 16 logical processors, using up to 16 threads\n",
"\n",
"Optimize a model with 21 rows, 110 columns and 220 nonzeros (Min)\n",
"Model fingerprint: 0x8d8d9346\n",
"Model fingerprint: 0x6d581c67\n",
"Model has 90 linear objective coefficients\n",
"Variable types: 0 continuous, 110 integer (110 binary)\n",
"Coefficient statistics:\n",
" Matrix range [5e-01, 2e+02]\n",
" Matrix range [4e-01, 3e+02]\n",
" Objective range [4e+00, 1e+02]\n",
" Bounds range [1e+00, 1e+00]\n",
" RHS range [1e+00, 5e+00]\n",
"Found heuristic solution: objective 368.7900000\n",
"Found heuristic solution: objective 375.3900000\n",
"Presolve time: 0.00s\n",
"Presolved: 21 rows, 110 columns, 220 nonzeros\n",
"Variable types: 0 continuous, 110 integer (110 binary)\n",
@@ -540,57 +541,64 @@
" Nodes | Current Node | Objective Bounds | Work\n",
" Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time\n",
"\n",
" 0 0 0.00000 0 6 368.79000 0.00000 100% - 0s\n",
"H 0 0 301.7200000 0.00000 100% - 0s\n",
"H 0 0 185.1900000 0.00000 100% - 0s\n",
"H 0 0 153.5000000 0.00000 100% - 0s\n",
"H 0 0 131.7700000 0.00000 100% - 0s\n",
" 0 0 17.14595 0 10 131.77000 17.14595 87.0% - 0s\n",
"H 0 0 122.8100000 17.14595 86.0% - 0s\n",
"H 0 0 98.3900000 17.14595 82.6% - 0s\n",
"H 0 0 92.2900000 64.28872 30.3% - 0s\n",
"H 0 0 91.6700000 64.28872 29.9% - 0s\n",
" 0 0 64.28872 0 15 91.67000 64.28872 29.9% - 0s\n",
"H 0 0 91.2300000 64.28872 29.5% - 0s\n",
" 0 0 0.00000 0 6 375.39000 0.00000 100% - 0s\n",
"H 0 0 299.2900000 0.00000 100% - 0s\n",
"H 0 0 181.8200000 0.00000 100% - 0s\n",
"H 0 0 151.3600000 0.00000 100% - 0s\n",
"H 0 0 129.7800000 0.00000 100% - 0s\n",
" 0 0 17.60837 0 10 129.78000 17.60837 86.4% - 0s\n",
"H 0 0 126.3600000 17.60837 86.1% - 0s\n",
"H 0 0 117.7900000 17.60837 85.1% - 0s\n",
"H 0 0 99.5900000 65.81544 33.9% - 0s\n",
" 0 0 65.81544 0 15 99.59000 65.81544 33.9% - 0s\n",
"H 0 0 94.0700000 65.81544 30.0% - 0s\n",
"H 0 0 93.1600000 85.53733 8.18% - 0s\n",
"H 0 0 91.3700000 85.53733 6.38% - 0s\n",
"\n",
"Cutting planes:\n",
" Cover: 16\n",
" MIR: 2\n",
" StrongCG: 1\n",
"\n",
"Explored 1 nodes (42 simplex iterations) in 0.01 seconds (0.00 work units)\n",
"Thread count was 16 (of 16 available processors)\n",
"\n",
"Solution count 10: 91.23 91.67 92.29 ... 368.79\n",
"Solution count 10: 91.37 93.16 94.07 ... 299.29\n",
"\n",
"Optimal solution found (tolerance 1.00e-04)\n",
"Best objective 9.123000000000e+01, best bound 9.123000000000e+01, gap 0.0000%\n",
"Best objective 9.137000000000e+01, best bound 9.137000000000e+01, gap 0.0000%\n",
"\n",
"User-callback calls 187, time in user-callback 0.00 sec\n"
"User-callback calls 250, time in user-callback 0.00 sec\n"
]
}
],
"source": [
"import numpy as np\n",
"from scipy.stats import uniform, randint\n",
"from miplearn.problems.pmedian import PMedianGenerator, build_pmedian_model_gurobipy\n",
"from miplearn.problems.pmedian import PMedianGenerator, PMedianPerturber, build_pmedian_model_gurobipy\n",
"\n",
"# Set random seed, to make example reproducible\n",
"np.random.seed(42)\n",
"\n",
"# Generate random instances with ten customers located in a\n",
"# Generate a reference instance with ten customers located in a\n",
"# 100x100 square, with demands in [0,10], capacities in [0, 250].\n",
"data = PMedianGenerator(\n",
"generator = PMedianGenerator(\n",
" x=uniform(loc=0.0, scale=100.0),\n",
" y=uniform(loc=0.0, scale=100.0),\n",
" n=randint(low=10, high=11),\n",
" p=randint(low=5, high=6),\n",
" demands=uniform(loc=0, scale=10),\n",
" capacities=uniform(loc=0, scale=250),\n",
")\n",
"reference_instance = generator.generate(1)[0]\n",
"\n",
"# Generate perturbed instances using the reference\n",
"perturber = PMedianPerturber(\n",
" distances_jitter=uniform(loc=0.9, scale=0.2),\n",
" demands_jitter=uniform(loc=0.9, scale=0.2),\n",
" capacities_jitter=uniform(loc=0.9, scale=0.2),\n",
" fixed=True,\n",
").generate(10)\n",
")\n",
"data = perturber.perturb(reference_instance, 10)\n",
"\n",
"# Print data for one of the instances\n",
"print(\"p =\", data[0].p)\n",