Our certification involves a rigorous evaluation process made by a referee under the supervision of an editor. The referee may not be a specialist of your research field but he or she is a specialists of the software used to generate your numerical results.
The referee is in charge of running the code, comparing the numerical results in the article with those resulting from the code, and identifying potential discrepancies. These potential differences may have different sources, for instance: errors when reporting the results or constructing figures, rounding problems, issued of numerical convergence, use of different versions of the same software, difference in system environment, etc. Depending on the nature of these differences, the referee suggests a rating, from RRR (perfect: all tables and figures can be replicated precisely) to DD (serious discrepancies) and provides the editor with an execution report.