- JOB
- Czech Republic
Job Information
- Organisation/Company
- Czech Technical University in Prague
- Research Field
- Computer science » InformaticsMathematics » Algorithms
- Researcher Profile
- First Stage Researcher (R1)
- Country
- Czech Republic
- Application Deadline
- Type of Contract
- Temporary
- Job Status
- Full-time
- Hours Per Week
- 40
- Offer Starting Date
- Is the job funded through the EU Research Framework Programme?
- H2020
- Is the Job related to staff position within a Research Infrastructure?
- No
Offer Description
Today, Large Neighborhood Search (LNS) [PER] is the preferred heuristic algorithm in solvers like CP Optimizer, Google OR-Tools, or OptalCP, as it offers a better ability to escape local optima and higher diversity in the exploration than previously often used Local Search. At the same time, Failure-Directed Search (FDS) [VIL] is used to efficiently explore the whole search space using a Fail-First principle to prove that no solution or better solution than the one found so far exists.
Nevertheless, CP solvers are less efficient when a criterion is more complex than, e.g., makespan minimization in scheduling.
But any algorithm that quickly provides a good solution, even without optimality guarantees, can be used inside the CP solver in the same way as LNS. The objective values of found solutions can be used as bounds (hard constraints on objective value), limiting the search space for the FDS. Thus, the better the provided solution is, the faster FDS explores the whole solution space.
This opens up research areas like a hybridization of problem-oriented heuristics and CP solvers that can lead to better performance while sharing the objective value and no-goods among heuristics and CP workers. We expect to explore utilizing machine learning (ML) in search algorithms, such as guiding search branching. Due to our collaboration with Petr Vilím, OptalCP [OPT] developer, we expect to exchange more information between the heuristic and the solver during the algorithm run.
About the group: The Industrial Informatics Department, led by Zdenek Hanzalek, is oriented towards scheduling, combinatorial optimization, and automated cars. The group has strong collaboration with high-tech companies (Porsche Engineering Services, EATON, Skoda Auto, ST Microelectronics, CEZ, Volkswagen, PPL, …).
[PER] L. Perron, P. Shaw, V. Furnon, Propagation guided large neighborhood search, in: M. Wallace (Ed.), Principles and Practice of Constraint Programming – CP 2004, Springer Berlin Heidelberg, Berlin, Heidelberg, 2004, pp. 468–481.
[OPT] ScheduleOpt: OptalCP’s solver landing page (2023). URL https://scheduleopt.com/
[VIL ]P. Vilím, P. Laborie, P. Shaw, Failure-directed search for constraint-based scheduling, in: International Conference on Integration of Constraint Programming, Artificial Intelligence, and Operations Research,
Springer, 2015, pp. 437–453.
Where to apply
Requirements
- Research Field
- Computer science » Informatics
- Education Level
- Master Degree or equivalent
- Motivation to perform excellent research, become part of the world's research communities in your field, and publish in first-tier scientific conferences and journals,
- MSc degree or equivalent (awarded or to be completed soon) in informatics, operations research or a related field,
- Proficiency in spoken/written English (knowledge of the Czech language is not required).
- Good background in scheduling, combinatorial optimization and algorithm design/implementation.
- Languages
- ENGLISH
- Level
- Excellent
Additional Information
- An initial appointment for 1 year (with an extension of up to 4 years, based on annual performance evaluation),
- Full waiver of tuition fees,
- Net compensation of about 1,600 EUR monthly (includes salary and student stipends) plus publication bonuses; check the Numbeo database for the cost of living in Prague,
- Full social and health insurance.
- 30 days of paid annual leave.
- An informal and inclusive international working environment at the Industrial Informatics Department, CIIRC, CTU in Prague.
Interested candidates are invited to submit their applications at:
https://forms.gle/Q7Rxn4R4NZFyWZuY7 [using PhD Position ID: 01-PhD-Hanzalek]
Work Location(s)
- Number of offers available
- 1
- Company/Institute
- Czech Technical University in Prague
- Country
- Czech Republic
- State/Province
- Czech Republic
- City
- Prague 6
- Postal Code
- 16000
- Street
- Jugoslávských partyzánů 1580/3
- Geofield
Contact
- State/Province
- Czech Republic
- City
- Prague
- Website
- Street
- Jugoslávských partyzánů 1580/3
- Postal Code
- 16000
- zdenek.hanzalek@cvut.cz