Skip to contents

Run anomaly detection pipeline over detector network

pipe_net.Rd

This function applies the single-detector pipe() function across a network of detectors (e.g., H1 and L1), managing parallel execution, results updating, and coincidence analysis.

Usage

pipe_net(
  batch_net,
  prev_batch,
  res.net,
  coinc.lis,
  arch_params,
  update_model = TRUE,
  verb = TRUE,
  debug = FALSE
)

Arguments

batch_net

A named list of current batches per detector. Each element is a ts object.

prev_batch

A named list of previous batches per detector (used for continuity in AR/MA).

res.net

A named list of results (updated per detector).

coinc.lis

A list of previous coincidence results (will be appended).

arch_params

A list of pipeline configuration parameters; see config_pipe.

update_model

Logical. Whether to update internal statistics across batches.

verb

Logical. If TRUE, print lambda diagnostics.

debug

Logical. If TRUE, runs sequentially for debugging (no parallelism).

Value

This function updates the following in the parent environment:

  • res.net: Updated detection results per detector.

  • prev_batch: Updated previous batch list for the next call.

  • coinc.lis: Appended coincidence analysis result.

These objects are expected to be initialized beforehand (e.g., via init_pipe).

Details

For each detector, the function calls the internal pipe function with the current batch, previous batch, and accumulated results. Parallel execution is performed via foreach + dopar unless debug = TRUE, in which case it runs sequentially.

Coincidence analysis is conducted using coincide_P0 when all detectors have valid results.

Examples

# Run detection pipeline over current batch
if (FALSE) { # \dontrun{
pipe_net(batch_net, prev_batch, res.net, coinc.lis, arch_params)
} # }