Plaquette Archive
A Lattice QCD data sharing platform
A Lattice QCD data sharing platform
Plaquette Archive is a collection of git repositories with lattice QCD data attached via DataLad. Each dataset is version-controlled, reproducible, and can be cloned with a single command.
Data sets and analysis projects are organised under the
PlaquetteArchive GitLab group. Each project is a DataLad
dataset that can be cloned with datalad clone.
Public data sets and analysis projects. Openly accessible DataLad datasets that can be cloned by anyone.
igit.ific.uv.es/PlaquetteArchive/private →Private data sets and analysis projects. Access is restricted to collaboration members with appropriate credentials.
igit.ific.uv.es/PlaquetteArchive/private1 →
The archive is built on the following software packages.
Each can be cloned with git clone.
Julia package for error analysis of Monte Carlo data using automatic differentiation. Implements the Γ-method for autocorrelation-based error estimation with exact error propagation through arbitrary functions.
igit.ific.uv.es/alramos/aderrors.jl →Julia library for reading and writing BDIO files — a compact binary format used to store lattice QCD observables and configuration data.
igit.ific.uv.es/alramos/bdio.jl →Julia library to store lattice QCD observables in BDIO format, providing utilities to handle measurement results within the DataLad framework.
igit.ific.uv.es/alramos/alphaio →
Interactive analysis notebooks rendered with
Quarto.
These demonstrate how to use the archive's software stack for
lattice QCD data analysis. Source files live in
notebooks/ and can be re-rendered with
quarto render.
Creating observables, propagating errors, and combining independent runs.
Source (.qmd) →Introduces the Γ-method: an autocorrelation-based technique for estimating statistical errors of arbitrary functions of Monte Carlo observables, yielding more reliable error bars than traditional binning.
Extends the Γ-method using automatic differentiation, enabling exact error propagation through arbitrary — even iteratively defined — observables and fit parameters. ADerrors.jl is the Julia implementation of these ideas.