Brainstorming 1

From ALPS
Jump to: navigation, search

Question: Can you separate the process of reading / writing from their meaning. We should try to find a common format to read/write data. The meaning and what should be done with the data is left to the devloper. We will restrict the case to MC simulations.

Suggestion: What tools can we use to make the link to legacy codes (fortran). Which tools do we want to have rather than how should the format look like in detail?

Question: Is the idea worth the effort? When/why do you need a common data format? The past shows that the motivation is usually quite low (physicists being busy with physics). The motivation for a common format/tools etc. has to be high enough for successful, wide adoption. How do we make the whole machinery 'fun' enough so that people will use it?

On the other hand, public sponsored research may have (at least some) rights to impose how things are done, such as which format is being used. The 'quality of service' is not guaranteed for MC simulations (people who still do not calculate correct error bars etc.). Standardized tools would give them easily applicable means for better analysis quality.

The need for an accessible data storage is also a question of culture. Some groups share data, some don't. Driving things for a common data format: Easy analysis for other people. This really makes life easier.


Motivation (not complete/accurate; see separate document): - Simplicity and portability - Efficiency - Ability to validate the quality of code - Shared analysis and data mining tool (simplify analysis and raise quality) - Data analysis


How do we transform information to knowledge? Current tools are just not suited for the (growing) requirements today.