[910] | 1 | ############################################### |
---|
| 2 | ## PLANETOPLOT ## |
---|
| 3 | ## --> PPCOMPUTE ## |
---|
| 4 | ############################################### |
---|
| 5 | # python built-in librairies |
---|
| 6 | import os |
---|
| 7 | # added librairies |
---|
| 8 | import numpy as np |
---|
| 9 | import scipy.signal as sp_signal |
---|
| 10 | ############################################### |
---|
| 11 | |
---|
| 12 | ## first a useful function to find settings in a folder in PYTHONPATH |
---|
| 13 | def findset(whereset,string="planetoplot_v2"): |
---|
| 14 | # ... set a default whereset if it was set to None |
---|
| 15 | # ... default is in the planetoplot_v2 folder |
---|
| 16 | if whereset is None: |
---|
| 17 | for path in os.environ['PYTHONPATH'].split(os.pathsep): |
---|
| 18 | if string in path: whereset = path |
---|
| 19 | if whereset is None: |
---|
| 20 | print "!! ERROR !! "+ string + "not in $PYTHONPATH" |
---|
| 21 | print "--> either put it in $PYTHONPATH or change whereset" |
---|
| 22 | exit() |
---|
| 23 | # ... if the last / is missing put it |
---|
| 24 | if whereset[-1] != "/": whereset = whereset + "/" |
---|
| 25 | return whereset |
---|
| 26 | |
---|
| 27 | ########################## |
---|
| 28 | #### MAX MEAN MIN SUM #### |
---|
| 29 | ##################################### |
---|
| 30 | #### WITH SUPPORT FOR NaN VALUES #### |
---|
| 31 | ##################################### |
---|
| 32 | |
---|
| 33 | ## compute min |
---|
| 34 | ## author AS + AC |
---|
| 35 | def min (field,axis=None): |
---|
| 36 | if field is None: return None |
---|
| 37 | if type(field).__name__=='MaskedArray': |
---|
| 38 | field.set_fill_value(np.NaN) |
---|
| 39 | return np.ma.array(field).min(axis=axis) |
---|
| 40 | elif (np.isnan(np.sum(field)) and (type(field).__name__ not in 'MaskedArray')): |
---|
| 41 | return np.ma.masked_invalid(field).min(axis=axis) |
---|
| 42 | else: return np.array(field).min(axis=axis) |
---|
| 43 | |
---|
| 44 | ## compute max |
---|
| 45 | ## author AS + AC |
---|
| 46 | def max (field,axis=None): |
---|
| 47 | if field is None: return None |
---|
| 48 | if type(field).__name__=='MaskedArray': |
---|
| 49 | field.set_fill_value(np.NaN) |
---|
| 50 | return np.ma.array(field).max(axis=axis) |
---|
| 51 | elif (np.isnan(np.sum(field)) and (type(field).__name__ not in 'MaskedArray')): |
---|
| 52 | return np.ma.masked_invalid(field).max(axis=axis) |
---|
| 53 | else: return np.array(field).max(axis=axis) |
---|
| 54 | |
---|
| 55 | ## compute mean |
---|
| 56 | ## author AS + AC |
---|
| 57 | def mean (field,axis=None): |
---|
| 58 | if field is None: return None |
---|
| 59 | else: |
---|
| 60 | if type(field).__name__=='MaskedArray': |
---|
| 61 | field.set_fill_value(np.NaN) |
---|
| 62 | zout=np.ma.array(field).mean(axis=axis) |
---|
| 63 | if axis is not None: |
---|
| 64 | zout.set_fill_value(np.NaN) |
---|
| 65 | return zout.filled() |
---|
| 66 | else:return zout |
---|
| 67 | elif (np.isnan(np.sum(field)) and (type(field).__name__ not in 'MaskedArray')): |
---|
| 68 | zout=np.ma.masked_invalid(field).mean(axis=axis) |
---|
| 69 | if axis is not None: |
---|
| 70 | zout.set_fill_value([np.NaN]) |
---|
| 71 | return zout.filled() |
---|
| 72 | else:return zout |
---|
| 73 | else: |
---|
| 74 | return np.array(field).mean(axis=axis) |
---|
| 75 | |
---|
| 76 | ## compute sum |
---|
| 77 | ## author AS + AC |
---|
| 78 | def sum (field,axis=None): |
---|
| 79 | if field is None: return None |
---|
| 80 | else: |
---|
| 81 | if type(field).__name__=='MaskedArray': |
---|
| 82 | field.set_fill_value(np.NaN) |
---|
| 83 | zout=np.ma.array(field).sum(axis=axis) |
---|
| 84 | if axis is not None: |
---|
| 85 | zout.set_fill_value(np.NaN) |
---|
| 86 | return zout.filled() |
---|
| 87 | else:return zout |
---|
| 88 | elif (np.isnan(np.sum(field)) and (type(field).__name__ not in 'MaskedArray')): |
---|
| 89 | zout=np.ma.masked_invalid(field).sum(axis=axis) |
---|
| 90 | if axis is not None: |
---|
| 91 | zout.set_fill_value([np.NaN]) |
---|
| 92 | return zout.filled() |
---|
| 93 | else:return zout |
---|
| 94 | else: |
---|
| 95 | return np.array(field).sum(axis=axis) |
---|
| 96 | |
---|
| 97 | ################ |
---|
| 98 | #### SMOOTH #### |
---|
| 99 | ################ |
---|
| 100 | ### TBD: works with missing values |
---|
| 101 | |
---|
| 102 | ## Author: AS. uses gauss_kern and blur_image. |
---|
| 103 | def smooth2d(field, window=10): |
---|
| 104 | ## actually blur_image could work with different coeff on x and y |
---|
| 105 | if True in np.isnan(field): |
---|
| 106 | print "!! ERROR !! Smooth is a disaster with missing values. This will be fixed." |
---|
| 107 | exit() |
---|
| 108 | if window > 1: result = blur_image(field,int(window)) |
---|
| 109 | else: result = field |
---|
| 110 | return result |
---|
| 111 | |
---|
| 112 | ## FROM COOKBOOK http://www.scipy.org/Cookbook/SignalSmooth |
---|
| 113 | def gauss_kern(size, sizey=None): |
---|
| 114 | # Returns a normalized 2D gauss kernel array for convolutions |
---|
| 115 | size = int(size) |
---|
| 116 | if not sizey: |
---|
| 117 | sizey = size |
---|
| 118 | else: |
---|
| 119 | sizey = int(sizey) |
---|
| 120 | x, y = np.mgrid[-size:size+1, -sizey:sizey+1] |
---|
| 121 | g = np.exp(-(x**2/float(size)+y**2/float(sizey))) |
---|
| 122 | return g / g.sum() |
---|
| 123 | |
---|
| 124 | ## FROM COOKBOOK http://www.scipy.org/Cookbook/SignalSmooth |
---|
| 125 | def blur_image(im, n, ny=None) : |
---|
| 126 | # blurs the image by convolving with a gaussian kernel of typical size n. |
---|
| 127 | # The optional keyword argument ny allows for a different size in the y direction. |
---|
| 128 | g = gauss_kern(n, sizey=ny) |
---|
| 129 | improc = sp_signal.convolve(im, g, mode='same') |
---|
| 130 | return improc |
---|
| 131 | |
---|
| 132 | ## FROM COOKBOOK http://www.scipy.org/Cookbook/SignalSmooth |
---|
| 133 | def smooth1d(x,window=11,window_type='hanning'): |
---|
| 134 | """smooth the data using a window with requested size. |
---|
| 135 | This method is based on the convolution of a scaled window with the signal. |
---|
| 136 | The signal is prepared by introducing reflected copies of the signal |
---|
| 137 | (with the window size) in both ends so that transient parts are minimized |
---|
| 138 | in the begining and end part of the output signal. |
---|
| 139 | input: |
---|
| 140 | x: the input signal |
---|
| 141 | window: the dimension of the smoothing window; should be an odd integer |
---|
| 142 | window_type: the type of window from 'flat', 'hanning', 'hamming', 'bartlett', 'blackman' |
---|
| 143 | flat window will produce a moving average smoothing. |
---|
| 144 | output: |
---|
| 145 | the smoothed signal |
---|
| 146 | example: |
---|
| 147 | t=linspace(-2,2,0.1) |
---|
| 148 | x=sin(t)+randn(len(t))*0.1 |
---|
| 149 | y=smooth(x) |
---|
| 150 | see also: |
---|
| 151 | numpy.hanning, numpy.hamming, numpy.bartlett, numpy.blackman, numpy.convolve |
---|
| 152 | scipy.signal.lfilter |
---|
| 153 | TODO: the window parameter could be the window itself if an array instead of a string |
---|
| 154 | """ |
---|
| 155 | if True in np.isnan(field): |
---|
| 156 | print "!! ERROR !! Smooth is a disaster with missing values. This will be fixed." |
---|
| 157 | exit() |
---|
| 158 | x = np.array(x) |
---|
| 159 | if x.ndim != 1: |
---|
| 160 | raise ValueError, "smooth only accepts 1 dimension arrays." |
---|
| 161 | if x.size < window: |
---|
| 162 | raise ValueError, "Input vector needs to be bigger than window size." |
---|
| 163 | if window<3: |
---|
| 164 | return x |
---|
| 165 | if not window_type in ['flat', 'hanning', 'hamming', 'bartlett', 'blackman']: |
---|
| 166 | raise ValueError, "Window is on of 'flat', 'hanning', 'hamming', 'bartlett', 'blackman'" |
---|
| 167 | s=np.r_[x[window-1:0:-1],x,x[-1:-window:-1]] |
---|
| 168 | #print(len(s)) |
---|
| 169 | if window == 'flat': #moving average |
---|
| 170 | w=np.ones(window,'d') |
---|
| 171 | else: |
---|
| 172 | w=eval('np.'+window_type+'(window)') |
---|
| 173 | y=np.convolve(w/w.sum(),s,mode='valid') |
---|
| 174 | return y |
---|