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

Jump to bottom

update time_slice documentation, add additional option for the exact … #4658

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

update time_slice documentation, add additional option for the exact … #4658

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
4f43519
update time_slice documentation, add additional option for the exact …
ahnitz Feb 28, 2024
b76368a
cc
ahnitz Mar 12, 2024
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
60 changes: 48 additions & 12 deletions pycbc/types/timeseries.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -175,31 +175,67 @@ def get_sample_rate(self):
sample_rate = property(get_sample_rate,
doc="The sample rate of the time series.")

def time_slice(self, start, end, mode='floor'):
def time_slice(self, start, end,
mode='floor',
exact_indices_threshold=1e-3):
"""Return the slice of the time series that contains the time range
in GPS seconds.

Note on indices.

Parameters
----------
start: float
The start time
end: float
The end time
mode: str, 'floor'
The rounding mode to be used when times are requested between samples.
The options are 'floor' which preferences choosing rounded down
index and 'nearest' which preferences whichever index is closest.
exact_indices_threshold: float, 1e-3
If the start or end time are intended to be at exact indices then
we need to identify this by testing if they are within an absolute
error introduced by numerical error accumulation.
"""
# (0) check if times are contained in the time series
if start < self.start_time:
raise ValueError('Time series does not contain a time as early as %s' % start)

if end > self.end_time:
raise ValueError('Time series does not contain a time as late as %s' % end)

# (1) Determine the start and end index corresponding to the requested
# times. This is calculated to subsample precision as start / end time
# are not restricted to the sample values.
start_idx = float(start - self.start_time) * self.sample_rate
end_idx = float(end - self.start_time) * self.sample_rate

if _numpy.isclose(start_idx, round(start_idx), rtol=0, atol=1E-3):
start_idx = round(start_idx)

if _numpy.isclose(end_idx, round(end_idx), rtol=0, atol=1E-3):
end_idx = round(end_idx)

if mode == 'floor':
# When dealing with time series of different size or transforming
# between different sample rates, floating point precision calculations
# can result in start / end times that should be at exact index locations
# to be some epsilon different in floating point value.
# (2) identify if we intended the start / end to be at an exact sample
# based on identifying only a trivial difference.
if exact_indices_threshold:
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Here's a further explainer why this is needed. One might ask, why not just use the 'nearest' mode. Is that not the same? This block is primarily useful when doing the 'floor' rounding. You don't want it to accidentally round down say a value of 3.9999999 to the previous sample if the difference was just floating point error. If you do then depending on the situation you might get a different length in the same situation. This is intended to make it more robust against this scenario.

if _numpy.isclose(start_idx, round(start_idx),
rtol=0, atol=exact_indices_threshold):
start_idx = round(start_idx)

if _numpy.isclose(end_idx, round(end_idx),
rtol=0, atol=exact_indices_threshold):
end_idx = round(end_idx)

# (3) Choose how index selection is done if the start or end time
# is actually between samples of the time series.
# We choose the index that is rounded down
if mode == 'floor':
start_idx = int(start_idx)
end_idx = int(end_idx)
elif mode == 'nearest':
start_idx = int(round(start_idx))
end_idx = int(round(end_idx))
# We choose the index that is nearest to the
# time requested.
elif mode == 'nearest':
start_idx = round(start_idx)
end_idx = round(end_idx)
else:
raise ValueError("Invalid mode: {}".format(mode))

Expand Down
Loading