Struct PhragmenModelConfig

pub struct PhragmenModelConfig<Backed> {
    pub weighted: bool,
    pub scale: Option<Backed>,
}

Configuration for the PhragmenModel plugin model

This struct allows fine-tuning of the Phragmen election algorithm to produce different variants of fair elections, including weighted, sequential, and scaled variants.

Each field modifies how candidates are evaluated and how backer loads are computed.

References:

• Phragmen voting - Wikipedia
• Enestrom, E. “Mathematical Theory of Proportional Representation”, 1896
• Thiele, T. N., “Proportional Representation in Elections”, 1895

Fields

weighted: bool

weighted: If true, the algorithm scales each backer’s contribution by their influence or stake.

• Behavior when true: Candidates supported by high-influence backers contribute more to the backers’ load. The algorithm prioritizes minimizing load while respecting weighted influence, meaning candidates with concentrated high-weight backers may be elected later to balance maximum loads.
• Behavior when false: All backers’ contributions are treated equally. Only the sum of contributions matters, ignoring stake or influence differences.

scale: Option<Backed>

scale: Optional multiplier applied to backer contributions when weighted is true.

• Behavior when Some(value): Each backer’s contribution is multiplied by scale. This allows amplifying or reducing the effective weight of all backers, controlling how heavily influence/stake impacts candidate ranking.
• Behavior when None: Contributions are used as-is, without additional scaling.

Trait Implementations

impl<Candidate, FairWeight, Backer, Backed, Input, Output> PurePluginModel<Input, PhragmenModelConfig<Backed>, Output> for PhragmenModel

where Input: IntoIterator<Item = (Candidate, FairWeight)> + Clone, Output: FromIterator<Candidate>, FairWeight: IntoIterator<Item = (Backer, Backed)> + Clone, Backed: Clone + Zero + Add<Output = Backed> + Ord + Div<usize, Output = Backed> + Mul<Backed, Output = Backed>, Candidate: Clone + Eq, Backer: Clone + Eq + Hash + Ord,

fn compute(&self, input: Input, ctx: &PhragmenModelConfig<Backed>) -> Output

Computes the Phragmen ranking of candidates imperatively.

Steps:

1. Initialize all backer loads to zero.
2. While candidates remain:
   • Compute maximum load increase for each candidate if elected.
   • Apply weighting/scaling if configured.
   • Deterministically select candidate with minimal max load (tie = first in list).
   • Update ranking and backer loads.
3. Return all candidates in ranked order.