[frames] | no frames]

# Module power

source code

Functions

 calc_var_of_window() source code

 smooth(ph, dt_in, dt_out, extra=0.0, wt=None) Smooths a data set, simultaneously resampling to a lower sampling rate. source code

 smooth_guts(ph, dt_in, dt_out, w, wt=None) source code

 old_smooth(ph, dt_in, dt_out, extra=0.0, wt=None) Smooths a data set, simultaneously resampling to a lower sampling rate. source code

 local_power(d, dt_in, dt_out, extra=0.0) THIS IS WRONG! IT PROBABLY SHOULD BE something like sqrt(d**2 + hilbert_transform(d)**2). source code

 test1() source code

 test2() source code

 test3() source code
 Variables __package__ = `'lib'`

Imports: math, numpy, hilbert_xform, P1

 Function Details

### smooth(ph, dt_in, dt_out, extra=0.0, wt=None)

source code

Smooths a data set, simultaneously resampling to a lower sampling rate. It uses successive boxcar averages followed by decimations for the initial smooth, then a convolution with a Gaussian. Even if `dt_out>>dt_in`, it only uses `O[log(dt_out/dt_in)` operations.

Parameters:
• `dt_in` (float (in the same units as extra and dt_out).) - input sampling rate.
• `dt_out` (float (in the same units as dt_in and extra).) - output sampling rate.
• `extra` (float (in the same units as dt_in and dt_out).) - extra smoothing time constant to apply. Extra is the standard deviation of a Gaussian kernel smooth that is applied as the last step. This last step is not implemented efficiently, so if if `extra>>dt_out` it can slow down the algorithm substantially.
• `ph` (`numpy.ndarray`.) - Normally a 1-dimensional array containing data to be smoothed. If the data is higher-dimensional, the time axis is assumed to run along axis=0, and the return value will be an array of the same dimension.
• `ph` (`numpy.ndarray`.) - None (which indicates a uniform weighting) or a `numpy.ndarray` that is the same length (axis 0) as `ph`.
• `wt` (`numpy.ndarray`)
Returns:
`(rv, t0)` where `rv` is a numpy array and `t0` it a `float` offset of the first element, relative to the start of the input data.

### old_smooth(ph, dt_in, dt_out, extra=0.0, wt=None)

source code

Smooths a data set, simultaneously resampling to a lower sampling rate. It uses successive boxcar averages followed by decimations for the initial smooth, then a convolution with a Gaussian. Even if `dt_out>>dt_in`, it only uses `O[log(dt_out/dt_in)` operations.

Parameters:
• `dt_in` (float (in the same units as extra and dt_out).) - input sampling rate.
• `dt_out` (float (in the same units as dt_in and extra).) - output sampling rate.
• `extra` (float (in the same units as dt_in and dt_out).) - extra smoothing time constant to apply. Extra is the standard deviation of a Gaussian kernel smooth that is applied as the last step. This last step is not implemented efficiently, so if if `extra>>dt_out` it can slow down the algorithm substantially.
• `ph` (`numpy.array`.) - a 1???-dimensionan array containing data to be smoothed. (Query: will this work for higher-dimensional data?)
Returns:
`(rv, t0)` where `rv` is a numpy array and `t0` it a `float` offset of the first element, relative to the start of the input data.

### local_power(d, dt_in, dt_out, extra=0.0)

source code

THIS IS WRONG! IT PROBABLY SHOULD BE something like sqrt(d**2 + hilbert_transform(d)**2). The hilbert transform just supplies the imaginary part of an analytic function. Current code leaves out the read part!

 Generated by Epydoc 3.0.1 on Thu Jun 16 20:03:14 2011 http://epydoc.sourceforge.net