Are there specific approaches, methods or software that help in determening why an optimum is not found for a particular optimization problem?

For example, the solution landscape could be visualized for a toy problem with few parameters that is not converging to the global optimum as anticipated. One would plot each step of the optimization algorithm, for example gradient descent, and determine, aha, it is stuck in a particular local minimum because the step size seems not large enough to escape in this particular landscape, or possibly the gradient direction isn‘t helpful, because the optimiztion path goes in circles. Something along the line of this visualization: https://en.m.wikipedia.org/wiki/Rosenbrock_function#/media/File%3ARosenbrock.png

But, are there other methods, approaches or even „diagnostic“ software?


closed as off-topic by Jake Westfall, jbowman, mdewey, mkt, Michael Chernick Oct 13 '18 at 20:11

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "This question appears to be off-topic because EITHER it is not about statistics, machine learning, data analysis, data mining, or data visualization, OR it focuses on programming, debugging, or performing routine operations within a statistical computing platform. If the latter, you could try the support links we maintain." – Jake Westfall, jbowman, mdewey, mkt, Michael Chernick
If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    $\begingroup$ Diagnosing problems with numeric solvers is a very broad topic. Why do you ask it here, and in this way, on the statistics stackexchange? $\endgroup$ – Martijn Weterings Oct 13 '18 at 16:15