Full-Range Cohypomonotone Inclusions
Descriptions may be stale. Consult the linked Lean sources.
Full-Range Cohypomonotone Inclusions
We seek a zero of
In terms of a point
Our assumptions:
Assumption 1 (Operator inequalities). For all
Assumption 2 (Initial distance). There exists a
solution
Assumption 3 (Stochastic evaluations and resolvent
access). At call
After the allotted calls, the output
Results
In a finite-dimensional real inner-product space,
| Result | Oracle complexity | Novelty | Limitations |
|---|---|---|---|
| (1) FullRangeCohypomonotoneMlmc
(Ours) |
|||
| (2) Inexact Halpern (Alacaoglu et al. 2026, Cor. 4.1) | (1): For fixed |
(1) bounds the mean norm.
(2) controls the second moment of its resolvent
residual for maximally |
|
| (3) RAIN |
(1) extends
(3) from unconstrained saddle-gradient operators ( |
||
| (4) Regularized prox-extragradient (LongTails 2026b) | (1) extends |
(1) has a seventh-power logarithm involving noise. (4) has a fourth-power logarithm independent of noise. | |
| (5) Multistep anchored method (LongTails 2026a) | (1) allows
finite-variance samples where (5) requires exact |
With exact |
Abbr: