The paper discusses the challenges in testing unconstrained optimization software, highlighting the limitations of current testing methods. It emphasizes the need for a larger set of test functions and guidelines to assess the reliability and robustness of optimization algorithms. The authors propose a collection of test functions and subroutines that allow for flexible testing of different problem areas, including systems of nonlinear equations, nonlinear least squares, and unconstrained minimization. These test functions are designed to evaluate how algorithms perform under various conditions, including different starting points and problem structures. The paper also describes specific test functions and their properties, along with examples of how they can be used to compare the performance of different optimization algorithms. The authors conclude that testing with a variety of starting points and problem types is essential for evaluating the effectiveness and reliability of optimization software.The paper discusses the challenges in testing unconstrained optimization software, highlighting the limitations of current testing methods. It emphasizes the need for a larger set of test functions and guidelines to assess the reliability and robustness of optimization algorithms. The authors propose a collection of test functions and subroutines that allow for flexible testing of different problem areas, including systems of nonlinear equations, nonlinear least squares, and unconstrained minimization. These test functions are designed to evaluate how algorithms perform under various conditions, including different starting points and problem structures. The paper also describes specific test functions and their properties, along with examples of how they can be used to compare the performance of different optimization algorithms. The authors conclude that testing with a variety of starting points and problem types is essential for evaluating the effectiveness and reliability of optimization software.