This API reference is not for the latest stable Salvus version.

`salvus.project.tools.processing.block_processing.resample`

Resampling modelled after https://ccrma.stanford.edu/~jos/resample/.

Functions

`resample()`

def resample(
    data: numpy.ndarray,
    t_in: numpy.ndarray,
    t_out: numpy.ndarray,
    window_params: salvus.project.tools.processing.block_processing.resample.ResamplingFilterParams = ResamplingFilterParams(
        num_zeros=32, precision=9, window=numpy.blackman, rolloff=0.945
    ),
    n_threads: int = 1,
) -> numpy.ndarray:
    ...

Resample a receiver array from one time axis to another.

Uses a fast band-limited resampling algorithm described in https://ccrma.stanford.edu/~jos/resample/ and influenced via the open-source implementations listed therein.

Parameters

data numpy.ndarray — The data to resample. Will be checked to ensure that it is of the correct shape (n_rec, n_cmp, n_samples).
t_in numpy.ndarray — The time axis of the input samples. Must be strictly increasing.
t_out numpy.ndarray — The desired time axis of the output samples. Must be strictly increasing, and must be fully contained within the input time axis.
window_params salvus.project.tools.processing.block_processing.resample.ResamplingFilterParams — The parameters controlling the resampling filter.
n_threads int — The number of threads to run the resampling on. For nontrivial data sizes, one should expect a linear speedup with the number of threads.

Returns numpy.ndarray — The receiver data resampled along the time axis.

Classes

`ResamplingFilterParams`

class ResamplingFilterParams(builtins.object):
    def __init__(
        self,
        num_zeros: int = 32,
        precision: int = 9,
        window: Callable[[int], numpy.ndarray] = numpy.blackman,
        rolloff: float = 0.945,
    ) -> None:
        ...

Filter parameters for resampling.

Default values were gathered from the inspection of some common use cases. Note that the filter width is quite a bit smaller than that used by default in resampy (num_zeros = 512). This looks to allow for a significant speedup without a significant reduction in quality for our use cases. Nevertheless, it is always good to check these parameters when working with a new dataset to ensure that they remain optimal.

Parameters

num_zeros int — The number of zero-crossings that are considered in the sinc filter.
precision int — The number of digits of precision for the linear window interpolation. Precision here is related to the relative difference input / output sampling rates.
window Callable[[int], numpy.ndarray] — The type of window used to truncate the sinc function.
rolloff float — Rolloff.