def log_feature():
"""Handles the edge cases of logDiff described in the paper."""
if f_cur <= f_prev:
return False
elif f_prev == 0:
return f_cur - f_prev > 1
else:
return safe_ratio_gt(log(f_cur - f_prev),
log(f_prev), 6.0 / 8.0)
def coarse_if_previously_fine(f_prev2, f_prev, f_cur):
"""Assumes the previous step taken was a fine step. Return true
if the next step should be a coarse step."""
if not safe_ratio_gt(f_cur, f_prev, 8.0 / 8.0):
# Coarse
return True
else:
# Fine
return False
def coarse_if_previously_coarse(f_prev2, f_prev, f_cur):
"""Assumes the previous step taken was a coarse step. Return true
if the next step should be a coarse step."""
def log_feature():
"""Handles the edge cases of logDiff described in the paper."""
if f_cur <= f_prev:
return False
elif f_prev == 0:
return f_cur - f_prev > 1
else:
return safe_ratio_gt(log(f_cur - f_prev),
log(f_prev), 6.0 / 8.0)
# Using 'if not' instead of 'if' to follow the layout of the algorithm
# as written in the paper.
# ratio(10, 8)
if not safe_ratio_gt(f_cur, f_prev, 10.0 / 8.0):
# downSlope(9, 8)
if not safe_ratio_gt(f_prev2 * f_cur, f_prev * f_prev, 9.0 / 8.0):
# Coarse
return True
else:
# ratioI(9, 8)
if not safe_ratio_gt(f_prev, f_cur, 9.0 / 8.0):
# logDiff(6, 8)
if not log_feature():
# upSlope(8, 4)
if not safe_ratio_gt(f_cur * f_prev2,
f_prev * f_prev, 8.0 / 4.0):
# Coarse
return True
else:
# Fine
return False
else:
# upTrend
if not (f_cur >= f_prev >= f_prev2):
# Fine
return False
else:
# Coarse
return True
else:
# Coarse
return True
else:
# downSlope(9, 8)
if not safe_ratio_gt(f_prev2 * f_cur, f_prev * f_prev, 9.0 / 8.0):
# ratio(11, 8)
if not safe_ratio_gt(f_cur, f_prev, 11.0 / 8.0):
# Coarse
return True
else:
# Fine
return False
else:
# Fine
return False
