MATLAB 'fminunc' not converging with custom objective function that includes conditional logic
I'm integrating two systems and This might be a silly question, but I'm working on a personal project and I'm attempting to optimize a custom objective function using `fminunc` in MATLAB R2023a, but I'm running into convergence issues..... My function has conditional logic to handle different ranges of input parameters, and it seems to be causing `fminunc` to struggle. When I run the optimizer, it outputs `Warning: Objective function incomplete. Using the default objective function value.` This happens even though I verified the objective function returns a numeric value under normal conditions. Here's a simplified version of the function I'm using: ```matlab function fval = myObjective(x) if x(1) < 0 fval = 100 + x(1)^2 + x(2)^2; else fval = 10 + (x(1) - 1)^2 + (x(2) + 2)^2; end end ``` And I'm calling `fminunc` like this: ```matlab options = optimoptions('fminunc', 'Display', 'iter', 'Algorithm', 'quasi-newton'); [x_opt, fval_opt] = fminunc(@myObjective, [0, 0], options); ``` When I provide an initial guess of `[0, 0]`, it seems to get exploring and doesn't proceed with the iterations. I've tried changing the initial guess to `[1, -1]`, but the behavior is similar. I've also ensured that the function outputs a scalar value, and Iām returning `fval` correctly. Is there something specific about how `fminunc` handles discontinuities or conditional logic in the objective function that I might be missing? Is there a better way to structure my function to avoid this convergence scenario? This is part of a larger API I'm building. I'd really appreciate any guidance on this. Am I missing something obvious? I'm open to any suggestions. My development environment is Linux. Any ideas what could be causing this?