Persistent committee size

Non-persistent committee size

Total persistent stake

Total non-persistent stake

Split a stake distrubution into persistent and non-persistent committee seats according to the weighted Fait-Accompli scheme.

This function returns: * number of persistent seats granted via the weighted Fait-Accompli scheme * number of non-persistent seats expected to vote via local sortition * total persistent stake * total non-persistent stake

Evaluate whether a voter with the given stake and relative position in the stake distribution can be granted a persistent seat in the voting committee.

Seat index in the voting committee

Number of pools with positive stake in the underlying stake distribution

Total stake in the underlying stake distribution

Errors that can occur when trying to split the stake distribution into persistent and seats via weighted Fait-Accompli.

Tiebreaker for voters with the same stake in the cumulative stake.

This is needed to ensure that the cumulative stake distribution is fair with respect to the edge case where there are multiple voters with the same stake around the persistent seat threshold, e.g.:

| seat index | stake | selection outcome | |------------|-------|-------------------| | 0 | 50 | persistent | | 1 | 30 | persistent | | 2 | 20 | persistent | | 3 | 20 | non-persistent | | 4 | 20 | non-persistent | | 5 | 10 | non-persistent | | ... | ... | ... |

In the case above, the pools with seat index 2, 3 and 4 have the same stake, but (under some hypothetical parameterization) only the one with seat index 2 can be granted a persistent seat according to the weighted Fait-Accompli scheme. Then, the job of this tiebreaker is to ensure that the seat index 2 is fairly distributed among the pools with the same stake.

One possible implementation of this tiebreaker is to sort the pools with the same stake according to the hash of the epoch nonce and the pool ID. This way the tiebreaker would be deterministic and resilient to manipulation since an adversary would not be able to predict the epoch nonce in advance (see wFATiebreakerWithEpochNonce below).

Fair weighted Fait-Accompli tiebreaker.

For this, we throw the current epoch nonce into the mix to avoid giving an adversary an edge to manipulate the tiebreaking in their favor, as they cannot predict the epoch nonce in advance.

NOTE: this implementation uses BLS-based hashing, but could be replaced by any other cryptographic hash function with similar properties.

Extended cumulative stake distribution.

Stake distribution in descending order with precomputed right-cumulative stake, i.e., the total stake of voters with smaller or equal stake than the current one (including the current one itself). In addition, this wrapper also allows the inclusion of an arbitrary payload of type a. This is useful to keep track of anything else we might need to know about the voters in the committee selection scheme (e.g. their public keys) in a single place.

E.g.: given the following stake distribution:

  PoolId 1 -> (50, PK#1)
  PoolId 2 -> (15, PK#2)
  PoolId 3 -> (10, PK#3)
  PoolId 4 -> (20, PK#4)
  PoolId 5 -> (5,  PK#5)

We would have the following cumulative stake distribution:

  Array.listArray
    (SeatIndex 0, SeatIndex 4)
    [ (PoolId 1, PK#1, LedgerStake 50, CumulativeStake 100)
    , (PoolId 4, PK#4, LedgerStake 20, CumulativeStake 50)
    , (PoolId 2, PK#2, LedgerStake 15, CumulativeStake 30)
    , (PoolId 3, PK#3, LedgerStake 10, CumulativeStake 15)
    , (PoolId 5, PK#5, LedgerStake 5,  CumulativeStake 5)
    ]

NOTE: this wrapper exists to allow us to share the same cumulative stake distribution across multiple committee selection instances derived from the same underlying stake distribution (e.g. Leios and Peras voting committees for the same epoch).

Construct an extended cumulative stake distribution.

Returns an error if the underlying stake distribution is empty.

Retrieve the candidate information associated to a given seat index, if the seat index is within bounds in the stake distribution.

Same as getCandidateIfSeatWithinBounds, but assumming that the seat index is within bounds in the stake distribution.