Validate Edge Weights and Per-Source Totals — validate_edge

Inspect a weighted edge list before it reaches the PageRank solver. The report identifies negative and non-finite weights, degenerate sources whose outgoing weights are all zero, and (optionally) source totals that do not match an expected probability total.

Usage

validate_edge_weights(
  edge_list_df,
  weight_col = "weight",
  from_col = "from",
  expected_total = NULL,
  tolerance = sqrt(.Machine$double.eps),
  action = c("error", "warning", "none")
)

Arguments

edge_list_df: A data frame containing source and weight columns.
weight_col: Name of the numeric edge-weight column.
from_col: Name of the source-node column used to define outgoing choice sets.
expected_total: Optional finite, non-negative total expected for each source. Use `1` to validate an already-normalized transition-probability column. `NULL` (default) reports totals without enforcing a target.
tolerance: Non-negative numeric tolerance for `expected_total`.
action: How validation failures are handled: `"error"` (default), `"warning"`, or `"none"`. The report is returned in every mode.

Value

A data frame with one row per source and columns describing edge count, weight total, invalid-value counts, all-zero status, optional total agreement, and overall validity.

Examples

edges <- data.frame(
  from = c("A", "A", "B"),
  to = c("B", "C", "C"),
  probability = c(0.25, 0.75, 1)
)
validate_edge_weights(
  edges,
  weight_col = "probability",
  expected_total = 1
)
#>   source n_edges total n_negative n_na n_nan n_infinite all_zero total_ok valid
#> 1      A       2     1          0    0     0          0    FALSE     TRUE  TRUE
#> 2      B       1     1          0    0     0          0    FALSE     TRUE  TRUE