Convergence of solving an OP $ minlimits_{x, alpha} sum_i alpha_i f_i(x) + g(alpha)$

Consider the optimization problem
$$ minlimits_{x, alpha} sum_i alpha_i f_i(x) + g(alpha). $$
I solved the above problem by alternating fixing $x$ solve for $alpha$ and vice versa. When $x$ is fixed, the corresponding optimization with variables $alpha$ is convex and has a closed form solution. When $alpha$ is fixed, the problem is non-convex and it is solved by a gradient descent algorithm (called this problem OP2).

My question is does the above alternative scheme converge (a local optimum) given than OP2 converges to a local optimum. If it is not do you know in which conditions does alternative scheme converge?


MathOverflow Asked by user263322 on January 1, 2021

0 Answers

Add your own answers!

Related Questions

Odd Steinhaus problem for finite sets

0  Asked on November 30, 2020 by domotorp


Measurable total order

1  Asked on November 5, 2020 by aryeh-kontorovich


Geodesics and potential function

0  Asked on October 29, 2020 by bruno-peixoto


Multivariate monotonic function

2  Asked on October 25, 2020 by kurisuto-asutora


Ask a Question

Get help from others!

© 2022 All rights reserved.