general_purpose_python_scripts/irafLikeFunctions.py at main · stubbslab/general_purpose_python_scripts · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
from cantrips import readInDataFromFitsFile
from cantrips import safeSortOneListByAnother
import numpy as np
import scipy.optimize as optimize
import matplotlib.pyplot as plt
import time

def gaussFunctToMinimizeOld(fit_vars, data, selection_radius, val_to_remove, show = 0):
    #print ('Hi!' )
    x0, y0, sig, A, shift = fit_vars
    #print ('fit_vars = ' + str(fit_vars))
    #selection_radius = int(sig * n_fwhm_to_fit)
    #print 'selection_radius = ' + str(selection_radius)
    #print 'int(x0)-selection_radius = ' + str(int(x0)-selection_radius)
    #print 'int(x0)+selection_radius+1 = ' + str(int(x0)+selection_radius+1)
    #print 'int(y0)-selection_radius = ' + str(int(y0)-selection_radius)
    #print 'int(y0)+selection_radius+1 = ' + str(int(y0)+selection_radius+1)
    #print 'np.shape(data) = ' + str(np.shape(data))
    fit_counts = data[int(y0)-selection_radius:int(y0)+selection_radius+1, int(x0)-selection_radius:int(x0)+selection_radius+1]
    x_dist_mesh, y_dist_mesh = np.meshgrid(np.array(range(-selection_radius, selection_radius+1)) - (y0 - int(y0)), np.array(range(-selection_radius, selection_radius+1)) - (x0 - int(x0)))
    #print ('data[int(y0), int(x0)] = ' + str(data[int(y0), int(x0)]))
    #print ('[y0, x0] = ' + str([y0, x0]))
    #print ('x_dist_mesh = ' + str(x_dist_mesh))
    #print ('y_dist_mesh = ' + str(y_dist_mesh))
    #print ('fit_counts = ' + str(fit_counts) )
    sqr_rad_mesh = x_dist_mesh ** 2.0 + y_dist_mesh ** 2.0
    #print 'np.shape(sqr_rad_mesh) = ' + str(np.shape(sqr_rad_mesh))
    #print 'np.shape(fit_counts) = ' + str(np.shape(fit_counts))
    sorted_sqr_rads, sorted_counts = safeSortOneListByAnother(sqr_rad_mesh.flatten(), [sqr_rad_mesh.flatten(), fit_counts.flatten()])
    sqr_rads_in_circle = [sorted_sqr_rads[i] for i in range(len(sorted_sqr_rads)) if sorted_sqr_rads[i] <= selection_radius ** 2.0]
    counts_in_circle = [sorted_counts[i] for i in range(len(sorted_sqr_rads)) if sorted_sqr_rads[i] <= selection_radius ** 2.0]

    rd_noise = 5.0
    raw_sigs = (rd_noise + np.sqrt(np.abs(np.array(counts_in_circle) - shift)))
    size_scaled_sigs = ((raw_sigs * np.sqrt(sqr_rads_in_circle)) + 1.0) / np.mean(np.sqrt(sqr_rads_in_circle))
    size_scaled_sigs = [1.0 for sig in size_scaled_sigs]

    vals_from_fit = A * np.exp(- np.array(sqr_rads_in_circle) / (2.0 * sig ** 2.0) ) + shift
    sum_of_sqrs = sum([((counts_in_circle[i] - vals_from_fit[i]) ** 2.0 / size_scaled_sigs[i])  for i in range(len(counts_in_circle)) ])

    #print ('For fit_vars in ' + str(fit_vars) + ', sum_of_sqrs = ' + str(sum_of_sqrs)  )

    if show:
        plt.scatter(np.sqrt(sqr_rads_in_circle), counts_in_circle)
        plt.errorbar(np.sqrt(sqr_rads_in_circle), counts_in_circle, yerr = size_scaled_sigs, fmt = 'none')
        plt.plot(np.linspace(-np.sqrt(max(sqr_rads_in_circle)), np.sqrt(max(sqr_rads_in_circle)), 500), A * np.exp(- np.array(np.linspace(-np.sqrt(max(sqr_rads_in_circle)), np.sqrt(max(sqr_rads_in_circle)), 500) ** 2.0) / (2.0 * sig ** 2.0) ) + shift, c = 'r')
        plt.show()

    print ('for fit vars ' + str(fit_vars) + ', returning_val = ' + str(sum_of_sqrs - val_to_remove)  )
    #return mean_sum_of_sqrs - val_to_remove
    return sum_of_sqrs - val_to_remove

def gaussFunctToMinimize(fit_vars, data, selection_radius, expected_fwhm_in_pix, val_to_remove, show = 0):
    x0, y0 = fit_vars

    centroid = [x0, y0]
    print ('centroid = ' + str(centroid) )

    data_to_find_max = data[int(centroid[1])-selection_radius:int(centroid[1])+selection_radius+1,
                            int(centroid[0])-selection_radius:int(centroid[0])+selection_radius+1]
    x_dists_mesh, y_dists_mesh = np.meshgrid(np.array(range(-selection_radius, selection_radius+1)),
                                             np.array(range(-selection_radius, selection_radius+1)))
    radius_sqr_mesh = x_dists_mesh ** 2.0 + y_dists_mesh ** 2.0
    radii_sqr = radius_sqr_mesh.flatten()
    data_to_find_max = data_to_find_max.flatten()
    #point_of_max = np.unravel_index(data_to_find_max.argmax(), data_to_find_max.shape)
    max_val = np.max(data_to_find_max)
    #x0_guess = point_of_max[0] + int(seed_point[1])-selection_radius
    #y0_guess = point_of_max[1] + int(seed_point[0])-selection_radius
    background_guess = np.median(data_to_find_max)
    init_guess = [expected_fwhm_in_pix, max_val - background_guess, background_guess]
    bounds = [(expected_fwhm_in_pix * 0.05, 0.0, 0.0),
              (selection_radius, np.inf, max_val)]

    raw_sigs = (5.0 + np.sqrt(np.abs(np.array(data_to_find_max) - background_guess)))
    size_scaled_sigs = raw_sigs * (np.sqrt(radii_sqr) + 1) / np.mean(np.sqrt(radii_sqr))

    print ('init_guess = ' + str(init_guess) )
    fit_funct = lambda rs_sqrd, sig, A, shift: A * np.exp(-np.array(rs_sqrd / (2.0 * sig ** 2.0))) + shift
    best_fit = optimize.curve_fit (fit_funct, radii_sqr, data_to_find_max, p0 = init_guess, bounds = bounds, sigma = size_scaled_sigs)[0]
    #selection_radius = int(sig * n_fwhm_to_fit)
    #print 'selection_radius = ' + str(selection_radius)
    #print 'int(x0)-selection_radius = ' + str(int(x0)-selection_radius)
    #print 'int(x0)+selection_radius+1 = ' + str(int(x0)+selection_radius+1)
    #print 'int(y0)-selection_radius = ' + str(int(y0)-selection_radius)
    #print 'int(y0)+selection_radius+1 = ' + str(int(y0)+selection_radius+1)
    #print 'np.shape(data) = ' + str(np.shape(data))
    fit_counts = data[int(x0)-selection_radius:int(x0)+selection_radius+1, int(y0)-selection_radius:int(y0)+selection_radius+1]
    x_dist_mesh, y_dist_mesh = np.meshgrid(np.array(range(-selection_radius, selection_radius+1)) - (x0 - int(x0)), np.array(range(-selection_radius, selection_radius+1)) - (y0 - int(y0)))
    sqr_rad_mesh = x_dist_mesh ** 2.0 + y_dist_mesh ** 2.0
    #print 'np.shape(sqr_rad_mesh) = ' + str(np.shape(sqr_rad_mesh))
    #print 'np.shape(fit_counts) = ' + str(np.shape(fit_counts))
    sorted_sqr_rads, sorted_counts = safeSortOneListByAnother(sqr_rad_mesh.flatten(), [sqr_rad_mesh.flatten(), fit_counts.flatten()])
    sqr_rads_in_circle = [sorted_sqr_rads[i] for i in range(len(sorted_sqr_rads)) if sorted_sqr_rads[i] <= selection_radius ** 2.0]
    counts_in_circle = [sorted_counts[i] for i in range(len(sorted_sqr_rads)) if sorted_sqr_rads[i] <= selection_radius ** 2.0]

    vals_from_fit = fit_funct(radii_sqr, *best_fit)
    sum_of_sqrs = sum([(counts_in_circle[i] - vals_from_fit[i]) ** 2.0 for i in range(len(counts_in_circle)) ])
    print ('centroid + best_fit = ' + str(list(centroid) + list(best_fit)))
    print ('sum_of_sqrs = ' + str(sum_of_sqrs) )
    print ('return_val = ' + str(sum_of_sqrs - val_to_remove) )

    #print ('For fit_vars in ' + str(fit_vars) + ', sum_of_sqrs = ' + str(sum_of_sqrs)  )

    if show:
        print ( 'here 1' )
        plt.scatter(np.sqrt(radii_sqr), data_to_find_max)
        plt.plot(np.sqrt(radii_sqr), fit_funct(radii_sqr, *best_fit), c = 'r')
        plt.show()

    #return mean_sum_of_sqrs - val_to_remove

    return ((centroid[0] - 1509.1) ** 2.0 + (centroid[1] - 2171.5) ** 2.0)
    #return (sum_of_sqrs - val_to_remove)

def determineCenterOfObject(guess_x, guess_y, selection_rad, data, n_iterations, max_iterations):

    guess_x = int(guess_x)
    guess_y = int(guess_y)
    selection_rad = int(selection_rad)

    print ('[guess_x, guess_y] = ' + str([guess_x, guess_y]))

    #print 'val_at_guess_point = ' + str(data[guess_x, guess_y])
    x_dist_mesh, y_dist_mesh = np.meshgrid(range(0, np.shape(data)[1]), range(0, np.shape(data)[0]))
    #print 'x_dist_mesh = ' + str(x_dist_mesh)
    #print 'y_dist_mesh = ' + str(y_dist_mesh)
    #print 'guess_x = ' + str(guess_x)
    #print 'guess_y = ' + str(guess_y)
    #print 'np.shape(data) = ' + str(np.shape(data))
    x_dist_mesh = x_dist_mesh[max(0, guess_y - selection_rad):min(np.shape(data)[0], guess_y + selection_rad + 1),
                           max(0, guess_x - selection_rad):min(np.shape(data)[1], guess_x + selection_rad + 1)]
    y_dist_mesh = y_dist_mesh[max(0, guess_y - selection_rad):min(np.shape(data)[0], guess_y + selection_rad + 1),
                              max(0, guess_x - selection_rad):min(np.shape(data)[1], guess_x + selection_rad + 1)]

    line_numbers = range(max(0, guess_y - selection_rad), min(np.shape(data)[0], guess_y + selection_rad + 1))
    col_numbers = range(max(0, guess_x - selection_rad), min(np.shape(data)[1], guess_x + selection_rad + 1))

    data_to_measure = data[max(0, guess_y - selection_rad):min(np.shape(data)[0], guess_y + selection_rad + 1),
                           max(0, guess_x - selection_rad):min(np.shape(data)[1], guess_x + selection_rad + 1)]

    line_profile = np.sum(data_to_measure, axis = 1)
    col_profile = np.sum(data_to_measure, axis = 0)

    line_mean = np.mean(line_profile)
    col_mean = np.mean(col_profile)

    line_centroid = (np.sum(np.array(line_numbers) * np.array([count if count >= line_mean else 0.0 for count in line_profile]))
                     / np.sum(np.array([count for count in line_profile if count >= line_mean]) ) )
    col_centroid = (np.sum(np.array(col_numbers) * np.array([count if count >= col_mean else 0.0 for count in col_profile]))
                    / np.sum(np.array([count for count in col_profile if count >= col_mean]) ) )

    if n_iterations < max_iterations and col_centroid != guess_x and line_centroid != guess_y:
        #print '[col_centroid, line_centroid] = ' + str([col_centroid, line_centroid])
        return determineCenterOfObject(col_centroid, line_centroid, selection_rad, data, n_iterations + 1, max_iterations)
    else:
        return [col_centroid, line_centroid]


def computeRadialData(image_file, data_dir, seed_point, expected_fwhm_in_pix,
                      n_fwhm_to_peak = 2, n_fwhm_to_fit = 5, fit_funct = 'gauss', tol = 0.001,
                      max_centroid_iterations = 5, rdnoise = 5.0):

    if fit_funct in ['gauss', 'gaussian', 'normal']:
        #fit_funct = gaussFunctToMinimize
        fit_funct = lambda rs_sqrd, sig, A, shift: A * np.exp(-np.array(rs_sqrd / (2.0 * sig ** 2.0))) + shift

    selection_radius = int(n_fwhm_to_fit * expected_fwhm_in_pix)
    data, header = readInDataFromFitsFile(image_file, data_dir)
    print ('image_file = ' + str(image_file))
    centroid = determineCenterOfObject(seed_point[0], seed_point[1], selection_radius, data, 0, max_centroid_iterations)
    print ('Found centroid = ' + str(centroid))

    data_to_find_max = data[int(centroid[1])-selection_radius:int(centroid[1])+selection_radius+1,
                            int(centroid[0])-selection_radius:int(centroid[0])+selection_radius+1]
    x_dists_mesh, y_dists_mesh = np.meshgrid(np.array(range(-selection_radius, selection_radius+1)),
                                             np.array(range(-selection_radius, selection_radius+1)))
    radius_sqr_mesh = x_dists_mesh ** 2.0 + y_dists_mesh ** 2.0
    radii_sqr = radius_sqr_mesh.flatten()
    point_of_max = np.unravel_index(data_to_find_max.argmax(), data_to_find_max.shape)
    data_to_find_max = data_to_find_max.flatten()
    max_val = np.max(data_to_find_max)
    x0_guess = point_of_max[0] + int(centroid[1])-selection_radius
    y0_guess = point_of_max[1] + int(centroid[0])-selection_radius
    print ('[x0_guess, y0_guess] = ' + str([x0_guess, y0_guess]) )
    background_guess = np.median(data_to_find_max)
    init_guess = [expected_fwhm_in_pix, max_val - background_guess, background_guess]

    bounds = [(y0_guess - expected_fwhm_in_pix * n_fwhm_to_fit, y0_guess + expected_fwhm_in_pix * n_fwhm_to_fit),
              (x0_guess - expected_fwhm_in_pix * n_fwhm_to_fit, x0_guess + expected_fwhm_in_pix * n_fwhm_to_fit),
              (expected_fwhm_in_pix * 0.05, expected_fwhm_in_pix * n_fwhm_to_fit),
              (0.0, np.inf),
              (0.0, max_val)]
    #bounds = [(expected_fwhm_in_pix * 0.05, 0.0, 0.0),
    #          (expected_fwhm_in_pix * n_fwhm_to_fit, np.inf, max_val)]
    #print ('init_guess = ' + str(init_guess))
    #print ('bounds = ' + str(bounds) )
    start = time.time()
    #init_sum_of_sqrs = gaussFunctToMinimize(centroid, data, 0.0, show = 1)
    init_sum_of_sqrs = gaussFunctToMinimizeOld(list([y0_guess, x0_guess]) + list(init_guess), data, selection_radius, 0.0, show = 1)
    print ('init_sum_of_sqrs = ' + str(init_sum_of_sqrs))

    #older style where I allowed the position of the star center to be varied during the fit.  It was a bit more advance, but could be made to work.
    # Left here, just in case I end up needing it sometime/somewhere.
    #best_fit = optimize.minimize(gaussFunctToMinimize, centroid + init_guess, args = (data, selection_radius, expected_fwhm_in_pix, init_sum_of_sqrs), tol = tol, bounds = bounds, method = 'L-BFGS-B')['x']
    print ('bounds = ' + str(bounds))
    print ('list([y0_guess, x0_guess]) + list(init_guess) = ' + str(list([y0_guess, x0_guess]) + list(init_guess)))
    print (('Here 1'))
    optimize.minimize(gaussFunctToMinimizeOld, list([y0_guess, x0_guess]) + list(init_guess), args = (data, selection_radius, init_sum_of_sqrs), tol = tol, bounds = bounds, method = 'L-BFGS-B')['x']
    best_fit = optimize.minimize(gaussFunctToMinimizeOld, list([y0_guess, x0_guess]) + list(init_guess), args = (data, selection_radius, init_sum_of_sqrs), tol = tol, bounds = bounds, method = 'L-BFGS-B')['x']
    print ('Here 2')


    final_sum_of_sqrs = gaussFunctToMinimizeOld(best_fit, data, selection_radius, 0.0, show = 1)
    print ('for best_fit = ' + str(best_fit) + ', sum_of_sqrs = ' + str(final_sum_of_sqrs) )

    #print ('init_guess = ' + str(init_guess)
    #print ('bounds = ' + str(bounds)
    #We determine the unce
    raw_sigs = (5.0 + np.sqrt(np.abs(np.array(data_to_find_max) - background_guess)))
    size_scaled_sigs = raw_sigs * (np.sqrt(radii_sqr) + 1) / np.mean(np.sqrt(radii_sqr))
    #print 'np.shape(size_scaled_sigs) = ' + str(np.shape(size_scaled_sigs))
    #print 'size_scaled_sigs = ' + str(size_scaled_sigs)
    #print 'np.min(size_scaled_sigs) = ' + str(np.min(size_scaled_sigs))
    #best_fit = optimize.curve_fit (fit_funct, radii_sqr, data_to_find_max, p0 = init_guess, bounds = bounds, sigma = size_scaled_sigs)[0]
    end = time.time()
    print ('Took ' + str(end - start) + 's to optimize.' )
    print ('best_fit = ' + str(best_fit) )
    #plt.scatter(np.sqrt(radii_sqr), data_to_find_max)
    #plt.errorbar(np.sqrt(radii_sqr), data_to_find_max, yerr = size_scaled_sigs, fmt = 'none')
    #plt.plot(np.linspace(0.0, max(np.sqrt(radii_sqr)), 201), fit_funct(np.linspace(0.0, max(np.sqrt(radii_sqr)), 201) ** 2.0, *init_guess), c = 'b')
    #plt.plot(np.linspace(0.0, max(np.sqrt(radii_sqr)), 201), fit_funct(np.linspace(0.0, max(np.sqrt(radii_sqr)), 201) ** 2.0, *best_fit), c = 'r')
    #plt.show()
    end = time.time()
    #print ('Took ' + str(end - start) + 's')
    #print ('best_fit = ' + str(best_fit))

    #gaussFunctToMinimize(best_fit, data, selection_radius, show = 1)
    sig = best_fit[2]
    fwhm = 2.0 * np.sqrt(np.log(2) * 2) * sig
    print ('fwhm = ' + str(fwhm) )

    return fwhm