Database for developing, testing, understanding, and launching vdW-DF-method progress
||Database for developing, testing, understanding, and launching vdW-DF-method progress|
||SNIC Medium Storage|
||Per Hyldgaard <email@example.com>|
||Chalmers tekniska högskola|
||2022-01-01 – 2023-01-01|
||10304 10302 10407|
This application concerns the building and use of a database for asserting performance of regular and new hybrid vdW-DF versions in benchmark suites that reflect challenges found in general materials, from molecules to bulk. The database is, for example, needed for our goal of crafting a general-purpose materials-theory tool. For a more complete description of the underlying database AND computational Material science please see description
of the SNIC LARGE Computational application that we have also just submitted
(it has been ID'ed as SNIC 2021/3-18).
The application also reflects that our Chalmers vdW-DF program is not only expanding but also changing in the character of our immediate research focus. Thanks to changes in internal and external funding, we are significantly increasing the participation.
We are moving to a focus on more bulk and surface testing and these foci require even more temporary storage (as there are many k-point-wavefunction combinations during calculations). That requires more than what we used for molecular benchmarks of the gMTKN55. We simply need now to hold a lot of info while the test suites are actually being performed (and that is true for any given combination of a benchmark suit and functional).
On top of that comes the need to hold these final results (energies and density files) for all system-functional combinations so as to make best use of the computational investments made. This is something we can cannot do now (in the present database) due to size limitation of previous storage allocations. However, we need it to be able to react fast when collaborators (or other functional users) ask us "why does this or that work for this or that system?" We therefore apply to have storage large enough so that we can hold a full database of simple energy results and a partial database of corresponding density files. The latter means that we will be able to quickly start additional analysis.