Algorithm parameters

Description

ProxAL.AlgParams — Type

AlgParams

Specifies ProxAL's algorithmic parameters.

Parameter	Description	Default value
`decompCtgs::Bool`	if true: decompose across contingencies (along with time)	false
`jacobi::Bool`	if true: do Jacobi updates, else do Gauss-Siedel updates	true
`num_sweeps::Int`	number of jacobi/gauss-seidel sweeps per primal update step	1
`iterlim::Int`	maximum number of ProxAL iterations	100
`nlpiterlim::Int`	maximum number of NLP subproblem iterations	100
`tol::Float64`	tolerance used for ProxAL termination	1.0e-4
`zero::Float64`	tolerance below which is regarded as zero	1.0e-8
`θ_t::Float64`	see Formulation	1.0
`θ_c::Float64`	see Formulation	1.0
`ρ_t::Float64`	AL penalty weight for ramp constraints	1.0
`ρ_c::Float64`	AL penalty weight for ctgs constraints	1.0
`updateρ_t::Bool`	if true: dynamically update `ρ_t`	true
`updateρ_c::Bool`	if true: dynamically update `ρ_c`	true
`τ::Float64`	Proximal weight parameter	3.0
`updateτ::Bool`	if true: dynamically update `τ`	true
`verbose::Int`	level of output: 0 (none), 1 (stdout)	0
`mode::Symbol`	computation mode `∈ [:nondecomposed, :coldstart, :lyapunov_bound]`	`:coldstart`
`optimizer::Any`	NLP solver	`nothing`
`gpu_optimizer::Any`	GPU-compatible NLP solver	`nothing`
`nr_tol::Float64`	Tolerance of the Newton-Raphson algorithm (used only in `ExaPFBackend()` model)	1e-10
`init_opt::Bool`	if true: initialize block OPFs with base OPF solution	false
`device::TargetDevice`	Target device to deport the resolution of the optimization problem	CPU
`verbose_inner::Int`	Verbose level for `ExaTronBackend()`	0
`tron_rho_pq::Float64`	Parameter for `ExaTronBackend()`	4e2
`tron_rho_va::Float64`	Parameter for `ExaTronBackend()`	4e4
`tron_inner_iterlim::Int`	Parameter for `ExaTronBackend()`	800
`tron_outer_iterlim::Int`	Parameter for `ExaTronBackend()`	20
`tron_outer_eps::Float64`	Parameter for `ExaTronBackend()`	1e-4

source