Descriptions may be stale. Consult the linked Lean sources.
Stochastic Regular Equality
Constraints
We consider minimizing a differentiable function subject to
differentiable equalities with . The gradient is Lipschitz, the
constraint Jacobian is bounded and Lipschitz, and its adjoint is
uniformly bounded below. The objective is lower bounded, with controlled
initial gap and constraint residual. The oracle returns unbiased
estimates of the gradient, constraints, and adjoint Jacobian, with
bounded variance and mean-square Lipschitz dependence; consecutive call
pairs share independent samples. For , the goal is
satisfying ,
where expectation includes oracle and independent algorithm
randomness.
The optimization problem is
Here . Vector norms
are Euclidean, and matrix norms are Frobenius
norms. Write for the matrix of the
adjoint derivative of . The
algorithm receives ,
where , , and all listed scalar
parameters other than and are nonnegative.
Our assumptions:
Assumption 1 (Smoothness and regularity). The
functions and are differentiable everywhere. For all
and ,
Assumption 2 (Initial bounds). There exists such that
Assumption 3 (Stochastic oracle). Let be a probability law on a measurable
sample space . The jointly
measurable maps are Bochner integrable in the sample argument at
every fixed point. For and
every , Define the nonnegative random quantities For every , these quantities are integrable and
satisfy Let be measurable, mutually
independent samples, each with law , independent of the algorithm’s
randomness. At call , a query
returns Thus calls and use the same sample, including
across the three components of each reply.
Queries may be made at any point of . The output must be
measurable, and its squared stationarity and feasibility residual must
be integrable and satisfy where is the joint law of the oracle samples
and the independent algorithm randomness.