tag:joss.theoj.org,2005:/papers/by/James%20QuinlanJournal of Open Source Software2023-03-15T00:11:38ZJournal of Open Source Softwarehttps://joss.theoj.orgtag:joss.theoj.org,2005:Paper/41582023-03-15T00:11:38Z2023-03-16T00:04:32ZUniversal Numbers Library: Multi-format Variable Precision Arithmetic Libraryacceptedv3.66.12023-01-09 23:04:06 UTC832023-03-15 00:11:38 UTC820235072E.Theodore L.OmtzigtStillwater Supercomputing, Inc, USA0000-0003-0194-951XJamesQuinlanSchool of Mathematical and Physical Sciences, University of New England, USA0000-0002-2628-165110.21105/joss.05072https://doi.org/10.5281/zenodo.7735084C++, Chttps://joss.theoj.org/papers/10.21105/joss.05072.pdfnumerical analysis, deep learning, machine learning, floating-point representations, mixed-precision