runSA: Stochastic approximation for parameter fitting

In ELAplus: Energy Landscape Analysis for Ecological Communities

Stochastic approximation for parameter fitting

Description

Estimate species interaction parameters and (optionally) environmental preference parameters by stochastic approximation (SA). The function runs multiple SA optimizations in parallel (when threads > 1) using foreach.

Usage

runSA(
  ocmat,
  enmat = NULL,
  rep = 128,
  threads = 1,
  we = 0.001,
  totalit = 1000,
  lambda = 0.01,
  intv = 100,
  runadamW = TRUE,
  maxlr = 0.005,
  sparse = TRUE,
  getall = FALSE,
  reporting = TRUE
)

Arguments

ocmat	Numeric matrix (0/1). Rows are samples and columns are species.
enmat	Optional numeric matrix. Environmental variables for each sample. Should be normalized/standardized beforehand.
rep	Integer. Number of parallel runs (replicates) for optimization.
threads	Integer. Number of cores used when running in parallel.
we	Numeric. Weight decay parameter for AdamW.
totalit	Integer. Total number of optimization iterations.
lambda	Numeric. Sparsity penalty parameter (used when sparse = TRUE).
intv	Integer. Interval of iterations at which intermediate results are recorded.
runadamW	Logical. If TRUE, use AdamW-style optimization.
maxlr	Numeric. Maximum learning rate.
sparse	Logical. If TRUE, apply sparsity constraint.
getall	Logical. If TRUE, return all replicate trajectories.
reporting	Logical. If TRUE, print progress messages.

Value

If getall = TRUE, a list of SA trajectories (length rep). Otherwise, a list with two elements, c(params, enmat) is returned. params is the concatenated matrix of the estimated parameter set:

h

Implicit environmental effects on species occurrence

J

Species-species interaction matrix

g

(If enmat is not NULL) Explicit effects of the environmental factors.

ELAplus documentation built on May 28, 2026, 5:09 p.m.