Commit 663665a0 authored by Adam Fekete's avatar Adam Fekete

add readme and gitignore

parent 7bb81819
# use glob syntax.
syntax: glob
# eclipse conf file
# idea
# other scm
# switch to regexp syntax.
# syntax: regexp
# ^\.pc/
#emacs TAGS
# Content
## (1) a tool for complexity quantification for a given target system, based on error vs. database size curves.
Short description: complexity a given target system will be defined here as the data set size necessary to predict a property of a system up to a given accuracy threshold, a given fraction of the time. In GP regression this will be kernel dependent, while for a given kernel, the "relative learning complexity" of any two systems will be appreciable by this tool (for the first time as far as we know). We will provide a tutorial with examples of the above, e.g., showing the atomic force prediction error vs. database size for a given system as a function of temperature. The different behaviour of qualitatively different situations (e.g., crystalline vs. amorphous solids, or gap vs. gapless systems) will be also exemplified by the tutorial. This will hopefully counter any wrong assumption that predictive power of a force filed might be uniformly good whatever the environmental parameters of the run.
  • A suggestion: Maybe you should keep task 1 and 2 in separate repository. Or if the two tasks are independent but not completely then it might be better to have a common repository and two task specific repositories that use the common or library repository.

Please register or sign in to reply
## (2) a tool for force-field validation given a property database.
Short description: the tool will allow to carry out a direct validation of a force field. The force field will be uploadable by the user, while selected examples with modifiable parameters will be provided in tutorial form. The tool will use curated databases available from the NOMAD archive. Statistical analysis tools will e.g., compare the forces computed by the supplied/selected force field with the database ones, e.g., calculated by first-principles calculations. The user will also be able to generate plots of the relative force error distribution associated with the force field, to analyse the force field performance over the entire force intensity spectrum.
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment