| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 430044 | Journal of Computational Science | 2016 | 7 Pages |
•We present technical and organizational challenges in modern computational sciences.•We illustrate two models for solving these challenges: workshop and partnership.•The efficiency of the workshop model is illustrated by a case study.•A design of a partnership model providing longer term solution is presented.
Research in computational science (CS) poses several challenges, stemming both from the scientific domain being studied and the sophisticated IT infrastructures used by the CS researchers. IT infrastructures typically consist of the latest and most powerful supercomputers, high-performance networks and high-capacity data storage services that are provided, developed and operated by IT experts who do not have in-depth understanding of the CS problems being studied. This “standard service approach” has been a very useful tool for streamlining the interaction between IT and application domain users. However, in many cases it has started to become a limiting factor that may prevent realizing the full potential of the IT services when tackling the most challenging CS research topics. The partnership initiative πCS established at the Leibniz Supercomputing Centre (LRZ) aims at transforming the relationship between Computational Scientists and IT service providers from a service-centered approach to an integrated partnership. The partnership between CS researchers and IT service provider allows IT services to better address the requirements of CS in an optimized and efficient manner. In addition to describing this model and its background and results achieved so far, the paper sheds some light on future development and assessment activities.
