Data generation from a computational perspective

Lührs, Anna

Talk (non-conference) (Invited)

FZJ-2018-05876

Data generation from a computational perspective

Lührs, A. (Corresponding author)FZJ*

2018

Neuroscience for ICT: Applications to computation and robotics, Berlin, Germany, 5 Jul 2018 - 6 Jul 2018

Please use a persistent id in citations: http://hdl.handle.net/2128/19817

Abstract: The generation of data is usually an early step in a research project. Several choices are made at this stage, for example where and in which format the data is stored, or which tools, methods or libraries are used to analyse and work with the data later on in the process. Such early choices also have an influence on other questions that are often only asked later on in the process, e.g. if and how the data can be re-used for future studies by the same lab, project partners or the community. This lecture will present a view on data generation from a computational perspective, i.e. what needs to be taken into account to make working with the data fast, easy and efficient. Different standard file formats will be presented, including their suitability for parallel access from a High-Performance Computing system. Another important topic will be the relation between data format and access patterns that – if not chosen wisely – can slow down the data access significantly.

Contributing Institute(s):

Research Program(s):

Appears in the scientific report 2018

Database coverage:
OpenAccess