Inconsistent Halpern Iteration

Source file

wc_inconsistent_halpern_iteration(n; verbose=true)

Problem statement

Compute a PEPit worst-case guarantee for wc_inconsistent_halpern_iteration.

Consider the fixed point problem

\[\mathrm{Find}\, x:\, x = Ax,\]

where $A$ is a non-expansive operator, that is a $L$-Lipschitz operator with $L=1$. When the solution of above problem, or fixed point, does not exist, behavior of the fixed-point iteration with A can be characterized with infimal displacement vector $v$.

Performance metric

This code computes a worst-case guarantee for the Halpern Iteration, when A is not necessarily consistent, i.e., does not necessarily have fixed point. That is, it computes the smallest possible $\tau(n)$ such that the guarantee

\[\|x_n - Ax_n - v\|^2 \leqslant \tau(n) \|x_0 - x_\star\|^2\]

is valid, where $x_n$ is the output of the Halpern iteration and $x_\star$ is the point where $v$ is attained, i.e.,

\[v = x_\star - Ax_\star\]

In short, for a given value of $n$, $\tau(n)$ is computed as the worst-case value of $\|x_n - Ax_n - v\|^2$ when $\|x_0 - x_\star\|^2 \leqslant 1$.

Algorithm

The Halpern iteration can be written as

\[x_{t+1} = \frac{1}{t + 2} x_0 + \left(1 - \frac{1}{t + 2}\right) Ax_t.\]

Theoretical guarantee

A worst-case guarantee for Halpern iteration can be found in [1, Theorem 8]:

\[\|x_n - Ax_n - v\|^2 \leqslant \left(\frac{\sqrt{Hn + 12} + 1}{n + 1}\right)^2 \|x_0 - x_\star\|^2.\]

References

The detailed approach is available in [1].

[1] J. Park, E. Ryu (2023). Accelerated Infeasibility Detection of Constrained Optimization and Fixed-Point Iterations. International Conference on Machine Learning.

Arguments

  • n: number of iterations.
  • verbose: print example and solver progress information when true.

Returns

  • pepit_tau: worst-case value
  • theoretical_tau: theoretical value

Julia usage

pepit_tau, theoretical_tau = wc_inconsistent_halpern_iteration(25; verbose=true)
## Returns approximately: (pepit_tau, theoretical_tau) = (0.025083, 0.036642)