election-methods@mailman.electorama.com

Technical discussion of election methods

View all threads

Re: [EM] Schulze STV question

MS
Markus Schulze
Sun, Jan 15, 2017 5:17 PM

Hallo,

49 seconds to count 366 votes is pretty rough. Even if
all the explosive combinatorics is in Candidates and
Seats and that just scales linearly with additional
votes, anything more than a small town starts to take
all day to compute.

Example A90 is an extreme example, because the result is
very close and there are no strong candidates.

For 60 of the 66 examples in Tideman's database, the runtime
to calculate the Schulze STV winners is below 0.1 s. For one
example (A17), the runtime is between 0.1 s and 1.0 s. For
four examples (A03, A05, A35, A67), the runtime is between
1.0 s and 10.0 s. Only for one example (A90), the runtime is
above 10.0 s.

Markus Schulze

Hallo, > 49 seconds to count 366 votes is pretty rough. Even if > all the explosive combinatorics is in Candidates and > Seats and that just scales linearly with additional > votes, anything more than a small town starts to take > all day to compute. Example A90 is an extreme example, because the result is very close and there are no strong candidates. For 60 of the 66 examples in Tideman's database, the runtime to calculate the Schulze STV winners is below 0.1 s. For one example (A17), the runtime is between 0.1 s and 1.0 s. For four examples (A03, A05, A35, A67), the runtime is between 1.0 s and 10.0 s. Only for one example (A90), the runtime is above 10.0 s. Markus Schulze