Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

add perf_ceiling in rating computation #2410

Draft
wants to merge 1 commit into
base: master
Choose a base branch
from
Draft

add perf_ceiling in rating computation #2410

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 13 additions & 7 deletions judge/ratings.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
from bisect import bisect
from math import pi, sqrt, tanh
from math import inf, pi, sqrt, tanh
from operator import attrgetter, itemgetter

from django.db import transaction
Expand All @@ -19,6 +19,7 @@
VAR_LIM = (sqrt(VAR_PER_CONTEST**2 + 4 * BETA2 * VAR_PER_CONTEST) - VAR_PER_CONTEST) / 2
SD_LIM = sqrt(VAR_LIM)
TANH_C = sqrt(3) / pi
PERF_CEILING_INCREMENT = 400


def tie_ranker(iterable, key=attrgetter('points')):
Expand Down Expand Up @@ -77,17 +78,19 @@ def get_var(times_ranked, cache=[VAR_INIT]):
return cache[times_ranked]


def recalculate_ratings(ranking, old_mean, times_ranked, historical_p):
def recalculate_ratings(ranking, old_mean, times_ranked, historical_p, perf_ceiling):
n = len(ranking)
new_p = [0.] * n
new_mean = [0.] * n

updated_bounds = [VALID_RANGE[0], min(VALID_RANGE[1], perf_ceiling)]

# Note: pre-multiply delta by TANH_C to improve efficiency.
delta = [TANH_C * sqrt(get_var(t) + VAR_PER_CONTEST + BETA2) for t in times_ranked]
p_tanh_terms = [(m, d, 1) for m, d in zip(old_mean, delta)]

# Calculate performance at index i.
def solve_idx(i, bounds=VALID_RANGE):
def solve_idx(i, bounds):
r = ranking[i]
y_tg = 0
for d, s in zip(delta, ranking):
Expand All @@ -111,8 +114,8 @@ def divconq(i, j):
new_mean = list(old_mean)
else:
# Calculate performance.
solve_idx(0)
solve_idx(n - 1)
solve_idx(0, updated_bounds)
solve_idx(n - 1, updated_bounds)
divconq(0, n - 1)

# Calculate mean.
Expand All @@ -131,7 +134,7 @@ def divconq(i, j):
w_sum += w / BETA2
w0 = 1. / get_var(times_ranked[i] + 1) - w_sum
p0 = eval_tanhs(tanh_terms[1:], old_mean[i]) / w0 + old_mean[i]
new_mean[i] = solve(tanh_terms, w0 * p0, lin_factor=w0)
new_mean[i] = solve(tanh_terms, w0 * p0, lin_factor=w0, bounds=updated_bounds)

# Display a slightly lower rating to incentivize participation.
# As times_ranked increases, new_rating converges to new_mean.
Expand Down Expand Up @@ -160,6 +163,9 @@ def rate_contest(contest):
users = users.exclude(last_rating__lt=contest.rating_floor)
if contest.rating_ceiling is not None:
users = users.exclude(last_rating__gt=contest.rating_ceiling)
perf_ceiling = contest.rating_ceiling + PERF_CEILING_INCREMENT
else:
perf_ceiling = inf

users = list(users)
participation_ids = list(map(itemgetter('id'), users))
Expand All @@ -176,7 +182,7 @@ def rate_contest(contest):
idx = user_id_to_idx[h['user_id']]
historical_p[idx].append(h['performance'])

rating, mean, performance = recalculate_ratings(ranking, old_mean, times_ranked, historical_p)
rating, mean, performance = recalculate_ratings(ranking, old_mean, times_ranked, historical_p, perf_ceiling)

now = timezone.now()
ratings = [Rating(user_id=i, contest=contest, rating=r, mean=m, performance=perf,
Expand Down