Skip to content

Commit d3fa758

Browse files
DvermettenDvermetten
committed
Add problems from Tailoring RSP workshop
1 parent 417f82f commit d3fa758

1 file changed

Lines changed: 370 additions & 0 deletions

File tree

‎problems.yaml‎

Lines changed: 370 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -900,3 +900,373 @@
900900
implementation: https://zenodo.org/records/8307853
901901
source (real-world/artificial): 'real-world'
902902
textual description: ''
903+
- name: MECHBench
904+
suite/generator/single: Problem Suite
905+
variable type: Continuous
906+
dimensionality: scalable'
907+
objectives: '1'
908+
constraints: Present
909+
dynamic: Not Present
910+
noise: Not Present
911+
multimodal: Present
912+
multi-fidelity: Not Present
913+
source (real-world/artificial): Real-World Application
914+
implementation: https://github.com/BayesOptApp/MECHBench
915+
textual description: This is a set of problems with inspiration from Structural
916+
Mechanics Design Optimization. The suite comprises three physical models, from
917+
which the user may define different kind of problems which impact the final design
918+
output.
919+
reference: https://arxiv.org/abs/2511.10821
920+
other info:
921+
name: null
922+
partial evaluations: Not Present
923+
full name: MECHBench
924+
constraint properties: Hard Constraints
925+
number of constraints: 1 or 2
926+
type of dynamicism: ''
927+
form of noise model: ''
928+
type of noise space: ''
929+
other noise properties: ''
930+
description of multimodality: Unstructured or non isotropic multimodality
931+
key challenges / characteristics: Embeds physical simulations and is flexible
932+
and modular
933+
scientific motivation: Bridge the black-box optimization techniques to a Mechanical
934+
Design Problem which require these kinds of algorithms
935+
limitations: The models do not include fracture or damage mechanics, just plasticity.
936+
implementation languages: Python
937+
links to implementations: https://github.com/BayesOptApp/MECHBench
938+
approximate evaluation time: Times -> from 1 minute to 7 minutes
939+
links to usage examples: ''
940+
general: ''
941+
- name: EXPObench
942+
suite/generator/single: Problem Suite
943+
variable type: Continuous, Integer, Categorical, Conditional
944+
dimensionality: 10 to 135
945+
objectives: '1'
946+
constraints: Present
947+
dynamic: Not Present
948+
noise: Present
949+
multimodal: Unknown
950+
multi-fidelity: Not Present
951+
source (real-world/artificial): Real-World Application
952+
implementation: https://github.com/AlgTUDelft/ExpensiveOptimBenchmark
953+
textual description: Wind farm layout optimization, gas filter design, pipe shape
954+
optimization, hyperparameter tuning, and hospital simulation
955+
reference: https://doi.org/10.1016/j.asoc.2023.110744
956+
other info:
957+
name: null
958+
partial evaluations: Not Present
959+
full name: EXPensive Optimization benchmark library
960+
constraint properties: Hard Constraints, Soft Constraints, Box Constraints, only
961+
box constraints implemented, others appear as penalty in objective
962+
number of constraints: 2 per variable (box), other constraints unknown (simulator
963+
fails)
964+
type of dynamicism: ''
965+
form of noise model: real-life (unknown)
966+
type of noise space: Observational
967+
other noise properties: ''
968+
description of multimodality: ''
969+
key challenges / characteristics: Expensive objectives
970+
scientific motivation: Address the lack of real-life expensive benchmarks
971+
limitations: single-objective only, constraints are handled naively (penalty in
972+
objective), no parallelization
973+
implementation languages: Python
974+
links to implementations: https://github.com/AlgTUDelft/ExpensiveOptimBenchmark
975+
approximate evaluation time: 2 to 80 seconds
976+
links to usage examples: ''
977+
general: ''
978+
- name: Gasoline direct injection engine design
979+
suite/generator/single: Single Problem
980+
variable type: Continuous, Ordinal
981+
dimensionality: '7'
982+
objectives: '2'
983+
constraints: Present
984+
dynamic: Not Present
985+
noise: Not Present
986+
multimodal: Unknown
987+
multi-fidelity: Present
988+
source (real-world/artificial): Real-World Application
989+
implementation: https://doi.org/10.1016/j.ejor.2022.08.032
990+
textual description: 'A multi-objective optimization problem seeking to minimize
991+
fuel consumption and NOx emissions over a two-minute dynamic duty cycle, subject
992+
to five constraints (turbine inlet temperature, number of knock occurrences, peak
993+
cylinder pressure, peak cylinder pressure rise, total work). Seven decision variables
994+
are defined: four define the hardware choices of cylinder compression ratio, turbo
995+
machinery and EGR cooler sizing; three relate to control variables that parameterise
996+
the engine control logic.'
997+
reference: ''
998+
other info:
999+
name: null
1000+
partial evaluations: Unknown
1001+
full name: ''
1002+
constraint properties: Hard Constraints, Soft Constraints
1003+
number of constraints: '5'
1004+
type of dynamicism: ''
1005+
form of noise model: ''
1006+
type of noise space: ''
1007+
other noise properties: ''
1008+
description of multimodality: ''
1009+
key challenges / characteristics: Expensive
1010+
scientific motivation: ''
1011+
limitations: Proprietary
1012+
implementation languages: Matlab Simulink and Wave RT co-simulation
1013+
links to implementations: ''
1014+
approximate evaluation time: ''
1015+
links to usage examples: ''
1016+
general: ''
1017+
- name: BEACON
1018+
suite/generator/single: Generator
1019+
variable type: Continuous
1020+
dimensionality: scalable
1021+
objectives: '2'
1022+
constraints: Not Present
1023+
dynamic: Not Present
1024+
noise: Not Present
1025+
multimodal: Present
1026+
multi-fidelity: Not Present
1027+
source (real-world/artificial): Artificially Generated
1028+
implementation: https://github.com/Stebbet/BEACON/
1029+
textual description: Generator for bi-objective benchmark problems with explicitly
1030+
controlled correlations in continuous spaces.
1031+
reference: https://dl.acm.org/doi/10.1145/3712255.3734303
1032+
other info:
1033+
name: null
1034+
partial evaluations: Not Present
1035+
full name: Continuous Bi-objective Benchmark problems with Explicit Adjustable
1036+
COrrelatioN control
1037+
constraint properties: Box Constraints
1038+
number of constraints: '0'
1039+
type of dynamicism: ''
1040+
form of noise model: ''
1041+
type of noise space: ''
1042+
other noise properties: ''
1043+
description of multimodality: Random
1044+
key challenges / characteristics: Multimodal, different correlations among objectives
1045+
scientific motivation: Controlled correlation among objectives
1046+
limitations: No analytical Pareto front, only bi-objective
1047+
implementation languages: Python
1048+
links to implementations: https://github.com/Stebbet/BEACON/tree/main
1049+
approximate evaluation time: Negligible
1050+
links to usage examples: ''
1051+
general: ''
1052+
- name: TulipaEnergy
1053+
suite/generator/single: Problem Suite
1054+
variable type: Continuous
1055+
dimensionality: scalable
1056+
objectives: '1'
1057+
constraints: Present
1058+
dynamic: Not Present
1059+
noise: Present
1060+
multimodal: Not Present
1061+
multi-fidelity: Present
1062+
source (real-world/artificial): Real-World Application
1063+
implementation: https://tulipaenergy.github.io/TulipaEnergyModel.jl/stable/
1064+
textual description: Determine the optimal investment and operation decisions for
1065+
different types of assets in the energy system (production, consumption, conversion,
1066+
storage, and transport), while minimizing loss of load.
1067+
reference: See https://tulipaenergy.github.io/TulipaEnergyModel.jl/stable/40-scientific-foundation/45-scientific-references
1068+
other info:
1069+
name: null
1070+
partial evaluations: Unknown
1071+
full name: TulipaEnergyModel.jl
1072+
constraint properties: Hard Constraints, Soft Constraints
1073+
number of constraints: millions
1074+
type of dynamicism: none
1075+
form of noise model: "depends on input \u2014 still working on stochastic inputs"
1076+
type of noise space: Parameter
1077+
other noise properties: ''
1078+
description of multimodality: ''
1079+
key challenges / characteristics: modeled as a potentially very large linear program,
1080+
different fidelities possible
1081+
scientific motivation: new techniques for solving large whitebox linear optimization
1082+
problems
1083+
limitations: not yet stochastic
1084+
implementation languages: Julia / JMP
1085+
links to implementations: https://github.com/TulipaEnergy/TulipaEnergyModel.jl
1086+
approximate evaluation time: from minutes to hours
1087+
links to usage examples: https://github.com/TulipaEnergy/Tulipa-OBZ-CaseStudy
1088+
general: ''
1089+
- name: ATO
1090+
suite/generator/single: Single Problem
1091+
variable type: Continuous
1092+
dimensionality: '10'
1093+
objectives: '2'
1094+
constraints: Not Present
1095+
dynamic: Not Present
1096+
noise: Not Present
1097+
multimodal: Not Present
1098+
multi-fidelity: Not Present
1099+
source (real-world/artificial): Real-World Application
1100+
implementation: '-'
1101+
textual description: Parameters of the Modules of the Automatic Train Operation
1102+
should be optimized. The parameters are continuous with different ranges. There
1103+
are two objectives (minimizing energy consumption, minimizing driving duration.
1104+
reference: ''
1105+
other info:
1106+
name: null
1107+
partial evaluations: Not Present
1108+
full name: ''
1109+
constraint properties: ''
1110+
number of constraints: ''
1111+
type of dynamicism: ''
1112+
form of noise model: ''
1113+
type of noise space: ''
1114+
other noise properties: ''
1115+
description of multimodality: ''
1116+
key challenges / characteristics: ''
1117+
scientific motivation: ''
1118+
limitations: ''
1119+
implementation languages: ''
1120+
links to implementations: ''
1121+
approximate evaluation time: ''
1122+
links to usage examples: ''
1123+
general: ''
1124+
- name: Brachytherapy treatment planning
1125+
suite/generator/single: Problem Suite
1126+
variable type: Continuous
1127+
dimensionality: 100-500
1128+
objectives: 2-3
1129+
constraints: Present
1130+
dynamic: Not Present
1131+
noise: Not Present
1132+
multimodal: Present
1133+
multi-fidelity: Present
1134+
source (real-world/artificial): Real-World Application
1135+
implementation: ''
1136+
textual description: Treatment planning for internal radiation therapy
1137+
reference: https://www.sciencedirect.com/science/article/pii/S1538472123016781
1138+
other info:
1139+
name: null
1140+
partial evaluations: Present
1141+
full name: Brachytherapy treatment planning
1142+
constraint properties: Hard Constraints
1143+
number of constraints: scalable
1144+
type of dynamicism: ''
1145+
form of noise model: ''
1146+
type of noise space: ''
1147+
other noise properties: ''
1148+
description of multimodality: ''
1149+
key challenges / characteristics: Multi-objective; aggregated objectives
1150+
scientific motivation: ''
1151+
limitations: No public source code
1152+
implementation languages: ''
1153+
links to implementations: ''
1154+
approximate evaluation time: ''
1155+
links to usage examples: ''
1156+
general: ''
1157+
- name: FleetOpt
1158+
suite/generator/single: Single Problem
1159+
variable type: Integer
1160+
dimensionality: 'Upper level: 54; lower level: 13208'
1161+
objectives: '1'
1162+
constraints: Present
1163+
dynamic: Not Present
1164+
noise: Not Present
1165+
multimodal: Unknown
1166+
multi-fidelity: Not Present
1167+
source (real-world/artificial): Real-World Application
1168+
implementation: 'Not public: was done for real client with their private data'
1169+
textual description: 'Healthcare organisation in the UK provided data about their
1170+
current fleet of vehicles to conduct non-emergency heathcare trips in the Argyll
1171+
and Bute region of Scotland, UK. They also provided historical data about the
1172+
trips the vehicles took and about the bases which the vehicles return to. The
1173+
aim is to reduce the existing fleet of vehicles while still ensuring all trips
1174+
can be covered. Moving a vehicle from one base to another to help cover trips
1175+
is OK as long as the original base can still cover its trips. Link to paper with
1176+
more details: https://dl.acm.org/doi/abs/10.1145/3638530.3664137'
1177+
reference: ''
1178+
other info:
1179+
name: null
1180+
partial evaluations: Present
1181+
full name: ''
1182+
constraint properties: ''
1183+
number of constraints: ''
1184+
type of dynamicism: ''
1185+
form of noise model: ''
1186+
type of noise space: ''
1187+
other noise properties: ''
1188+
description of multimodality: ''
1189+
key challenges / characteristics: ''
1190+
scientific motivation: ''
1191+
limitations: ''
1192+
implementation languages: ''
1193+
links to implementations: ''
1194+
approximate evaluation time: ''
1195+
links to usage examples: ''
1196+
general: ''
1197+
- name: Building spatial design
1198+
suite/generator/single: Single Problem
1199+
variable type: Continuous, Boolean
1200+
dimensionality: scalable depending on problem size (e.g. 90 for)
1201+
objectives: '2'
1202+
constraints: Present
1203+
dynamic: Not Present
1204+
noise: Not Present
1205+
multimodal: Unknown
1206+
multi-fidelity: Not Present
1207+
source (real-world/artificial): Real-World Application
1208+
implementation: https://github.com/TUe-excellent-buildings/BSO-toolbox
1209+
textual description: 'Optimise the spatial layout of a building to: minimise energy
1210+
consumption for climate control, and minimise the strain on the structure'
1211+
reference: https://hdl.handle.net/1887/81789
1212+
other info:
1213+
name: null
1214+
partial evaluations: Not Present
1215+
full name: Building spatial design
1216+
constraint properties: Hard Constraints, Box Constraints, Permutation Constraints
1217+
number of constraints: 2065 (as example, depends on problem size)
1218+
type of dynamicism: ''
1219+
form of noise model: ''
1220+
type of noise space: ''
1221+
other noise properties: ''
1222+
description of multimodality: ''
1223+
key challenges / characteristics: Many hard constraints (simulator cannot evaluate
1224+
the solution if these are violated); Mixed-variable search space (continuous
1225+
+ binary); Multiple objectives; (Somewhat) expensive solution evaluations
1226+
scientific motivation: ''
1227+
limitations: ''
1228+
implementation languages: C++
1229+
links to implementations: https://github.com/TUe-excellent-buildings/BSO-toolbox
1230+
approximate evaluation time: Roughly 1 second per evaluation for the smallest
1231+
considered design, and roughly 40 seconds for the larger designs we considered.
1232+
(Even the larger designs we considered are still relatively small for the considered
1233+
problem.)
1234+
links to usage examples: ''
1235+
general: ''
1236+
- name: Electric Motor Design Optimization
1237+
suite/generator/single: Single Problem
1238+
variable type: Continuous, Integer
1239+
dimensionality: '13'
1240+
objectives: '1'
1241+
constraints: Present
1242+
dynamic: Not Present
1243+
noise: Present
1244+
multimodal: Present
1245+
multi-fidelity: Not Present
1246+
source (real-world/artificial): Real-World Application
1247+
implementation: Implementation not freely available
1248+
textual description: The goal is to find a design of a synchronous electric motor
1249+
for power steering systems that minimizes costs and satisfies all constraints.
1250+
reference: https://dis.ijs.si/tea/Publications/Tusar23Multistep.pdf (paper in Slovene)
1251+
other info:
1252+
name: null
1253+
partial evaluations: Not Present
1254+
full name: Electric Motor Design Optimization
1255+
constraint properties: Hard Constraints, Soft Constraints, Box Constraints
1256+
number of constraints: '12'
1257+
type of dynamicism: ''
1258+
form of noise model: ''
1259+
type of noise space: ''
1260+
other noise properties: ''
1261+
description of multimodality: Constraints are multimodal
1262+
key challenges / characteristics: Time-consuming solution evaluation, highly-constrained
1263+
problem
1264+
scientific motivation: Challenging to find good solutions in a limited time
1265+
limitations: 'Unavailability, even if available, it wouldn''t be helpful to use
1266+
for benchmarking due taking a long time to evaluate a single solution '
1267+
implementation languages: Python
1268+
links to implementations: Implementation not freely available
1269+
approximate evaluation time: 8 minutes
1270+
links to usage examples: ''
1271+
general: This is not an available problem, but could be interesting to show to
1272+
researchers which difficulties appear in real-world problems

0 commit comments

Comments
 (0)