Struct BenfordSampler 

pub struct BenfordSampler { /* private fields */ }

Sampler that produces amounts following Benford’s Law distribution.

The sampler derives the order-of-magnitude of each amount from the configured log-normal distribution (parameterised by lognormal_mu / lognormal_sigma on AmountDistributionConfig) and then overrides the first significant digit to match the requested Benford / anti-Benford distribution. This produces realistic per-amount magnitudes that match routine traffic, while preserving the first-digit statistical signature that fraud-detection consumers care about.

(Earlier versions sampled magnitude uniformly across [log10(min_amount), log10(max_amount)]. That over-represented the high tail and caused account-level balances to drift into numerically unsafe territory under audit-group + 12-month + hundred_k retail configs — see GitHub issue #185.)

Implementations

impl BenfordSampler

pub fn new(seed: u64, config: AmountDistributionConfig) -> Self

Create a new Benford sampler with the given seed and amount configuration.

pub fn sample(&mut self) -> Decimal

Sample an amount following Benford’s Law.

pub fn sample_with_first_digit(&mut self, first_digit: u8) -> Decimal

Sample an amount with a specific first digit.

The order-of-magnitude is drawn from the parent log-normal distribution (clamped to [min_amount, max_amount]), which matches the magnitude distribution of routine traffic. The first significant digit is then forced to first_digit, preserving the Benford / anti-Benford statistical signature that consumers expect. This avoids the uniform-magnitude over-representation of the high tail that previously inflated account-level balances into the 10^15+ range under audit-group + multi-period retail configs (issue #185).

pub fn reset(&mut self, seed: u64)

Reset the sampler with a new seed.