lumicks.pylake.channel.Slice¶

class Slice(data_source, labels=None, calibration=None)¶

A lazily evaluated slice of a timeline/HDF5 channel

Users will only ever get these as a result of slicing a timeline/HDF5 channel or slicing another slice (via this class’ __getitem__), i.e. the __init__ method will never be invoked by users.

Parameters:	data_source : Any A slice data source. Can be `Continuous`, `TimeSeries`, ‘TimeTags’, or any other source which conforms to the same interface. labels : Dict[str, str] Plot labels: “x”, “y”, “title”. calibration: ForceCalibration

__init__(data_source, labels=None, calibration=None)¶: Initialize self. See help(type(self)) for accurate signature.

Methods

`__init__`(data_source[, labels, calibration])	Initialize self.
`downsampled_by`(factor[, reduce])	Return a copy of this slice which is downsampled by `factor`
`downsampled_like`(other_slice[, reduce])	Downsample high frequency data analogously to a low frequency channel in the same way that Bluelake does it.
`downsampled_over`(range_list[, reduce, where])	Downsample channel data based on timestamp ranges.
`downsampled_to`(frequency[, reduce, where, …])	Return a copy of this slice downsampled to a specified frequency
`plot`([start])	A simple line plot to visualize the data over time
`range_selector`([show])

Attributes

`calibration`	Calibration data slicing is deferred until calibration is requested to avoid slicing values that may be needed.
`data`	The primary values of this channel slice
`sample_rate`	The data frequency for continuous data sources or `None` if it’s variable
`seconds`	Relative time (in seconds) that corresponds to the channel data
`start`	Starting timestamp of this time series in nanoseconds
`stop`	End timestamp of this time series in nanoseconds
`timestamps`	Absolute timestamps (since epoch) which correspond to the channel data

__getitem__(item)¶: All indexing is in timestamp units (ns)

downsampled_by(factor, reduce=<function mean>)¶

Return a copy of this slice which is downsampled by factor

Parameters:	factor : int The size and sample rate of the data will be divided by this factor. reduce : callable The `numpy` function which is going to reduce multiple samples into one. The default is `np.mean`, but `np.sum` could also be appropriate for some cases, e.g. photon counts.

downsampled_like(other_slice, reduce=<function mean>)¶

Downsample high frequency data analogously to a low frequency channel in the same way that Bluelake does it.

Note: some data required to reconstruct the first low frequency time point can actually occur before the starting timestamp of the marker and is therefore missing from the exported .h5 file. Therefore, it is not always possible to downsample to all of the data points in the low frequency other_slice. This function returns both the requested downsampled channel data and a copy of the input channel cropped such that both returned Slice objects have the same time points.

Parameters:	other_slice : Slice Timeline channel to downsample like. This should be a low frequency channel that provides the timestamps to downsample to. reduce : callable The `numpy` function which is going to reduce multiple samples into one. The default is `np.mean`, but `np.sum` could also be appropriate for some cases, e.g. photon counts.
Returns:	downsampled_slice : Slice This channel downsampled to the same timestamp ranges as `other_slice`. cropped_other_slice : Slice A copy of `other_slice` cropped such that the timestamps match those of `downsampled_slice`.

downsampled_over(range_list, reduce=<function mean>, where='center')¶

Downsample channel data based on timestamp ranges. The downsampling function (e.g. np.mean) is evaluated for the time between a start and end time of each block. A list is returned that contains the data corresponding to each block.

Parameters:

range_list : list of tuples: A list of (start, stop) tuples indicating over which ranges to apply the function. Start and stop have to be specified in nanoseconds.
reduce : callable: The numpy function which is going to reduce multiple samples into one. The default is np.mean, but np.sum could also be appropriate for some cases, e.g. photon counts.
where : str: Where to put the final time point. ‘center’ time point is put at (timestamps_subset[0] + timestamps_subset[-1]) / 2, where timestamps_subset are the timestamps corresponding to the samples being downsampled over. ‘left’ time point is put at start

Examples

from lumicks import pylake

file = pylake.File("example.h5")
stack = pylake.CorrelatedStack("example.tiff")
file.force1x.downsampled_over(stack.timestamps)

downsampled_to(frequency, reduce=<function mean>, where='center', method='safe')¶

Return a copy of this slice downsampled to a specified frequency

Parameters:

frequency : int

The desired downsampled frequency downsampled (Hz)

reduce : callable

The numpy function which is going to reduce multiple samples into one. The default is np.mean, but np.sum could also be appropriate for some cases, e.g. photon counts.

where : str

Where to put the final time point. ‘center’ time point is put at (timestamps_subset[0] + timestamps_subset[-1]) / 2, where timestamps_subset are the timestamps corresponding to the samples being downsampled over. ‘left’ time point is put at start

method : str

How to handle target sample times that are not exact multiples of the current sample time.

‘safe’ new sample time must be an exact multiple of the current sample time,: else an exception is raised.

‘ceil’ rounds the sample rate up to the nearest frequency which fulfills this condition ‘force’ downsample data with the target input frequency; this will result in variable

sample times and a variable number of sampling contributing to each target sample, but must be used for variable-frequency data

plot(start=None, **kwargs)¶

A simple line plot to visualize the data over time

Parameters:	start : int64 Origin timestamp. This can be used to plot two slices starting at different times on the same axis. **kwargs Forwarded to `matplotlib.pyplot.plot()`.

calibration¶: Calibration data slicing is deferred until calibration is requested to avoid slicing values that may be needed.

data¶: The primary values of this channel slice

sample_rate¶: The data frequency for continuous data sources or None if it’s variable

seconds¶: Relative time (in seconds) that corresponds to the channel data

start¶: Starting timestamp of this time series in nanoseconds

stop¶: End timestamp of this time series in nanoseconds

timestamps¶: Absolute timestamps (since epoch) which correspond to the channel data