Benchmark Tests for IV Fitting Algorithms

Clifford W. Hansen, Abigail R. Jones, Taos Transue, Marios Theristis

Abstract:

We propose a set of benchmark tests for current-voltage (IV) curve fitting algorithms. Benchmark tests enable
transparent and repeatable comparisons among algorithms,
allowing for measuring algorithm improvement over time. An
absence of such tests contributes to the proliferation of fitting
methods and inhibits achieving consensus on best practices.
Benchmarks include simulated curves with known parameter
solutions, with and without simulated measurement error. We
implement the reference tests on an automated scoring platform
and invite algorithm submissions in an open competition for
accurate and performant algorithms.