1 | import sys, time |
---|
2 | from calcthread import CalcThread |
---|
3 | from sans.pr.invertor import Invertor |
---|
4 | import numpy |
---|
5 | from config import printEVT |
---|
6 | |
---|
7 | class CalcPr(CalcThread): |
---|
8 | """ |
---|
9 | Compute 2D model |
---|
10 | This calculation assumes a 2-fold symmetry of the model |
---|
11 | where points are computed for one half of the detector |
---|
12 | and I(qx, qy) = I(-qx, -qy) is assumed. |
---|
13 | """ |
---|
14 | |
---|
15 | def __init__(self, pr, nfunc=5, error_func=None, |
---|
16 | completefn = None, |
---|
17 | updatefn = None, |
---|
18 | yieldtime = 0.01, |
---|
19 | worktime = 0.01 |
---|
20 | ): |
---|
21 | CalcThread.__init__(self,completefn, |
---|
22 | updatefn, |
---|
23 | yieldtime, |
---|
24 | worktime) |
---|
25 | self.pr = pr |
---|
26 | self.nfunc = nfunc |
---|
27 | self.error_func = error_func |
---|
28 | self.starttime = 0 |
---|
29 | |
---|
30 | def isquit(self): |
---|
31 | try: |
---|
32 | CalcThread.isquit(self) |
---|
33 | except KeyboardInterrupt: |
---|
34 | printEVT("P(r) calc interrupted") |
---|
35 | raise KeyboardInterrupt |
---|
36 | |
---|
37 | def compute(self): |
---|
38 | import time |
---|
39 | try: |
---|
40 | self.starttime = time.time() |
---|
41 | #out, cov = self.pr.invert(self.nfunc) |
---|
42 | out, cov = self.pr.lstsq(self.nfunc) |
---|
43 | elapsed = time.time()-self.starttime |
---|
44 | self.complete(out=out, cov=cov, pr=self.pr, elapsed=elapsed) |
---|
45 | except: |
---|
46 | if not self.error_func==None: |
---|
47 | self.error_func("CalcPr.compute: %s" % sys.exc_value) |
---|
48 | |
---|
49 | class EstimatePr(CalcThread): |
---|
50 | """ |
---|
51 | Compute 2D model |
---|
52 | This calculation assumes a 2-fold symmetry of the model |
---|
53 | where points are computed for one half of the detector |
---|
54 | and I(qx, qy) = I(-qx, -qy) is assumed. |
---|
55 | """ |
---|
56 | |
---|
57 | def __init__(self, pr, nfunc=5, error_func=None, |
---|
58 | completefn = None, |
---|
59 | updatefn = None, |
---|
60 | yieldtime = 0.01, |
---|
61 | worktime = 0.01 |
---|
62 | ): |
---|
63 | CalcThread.__init__(self,completefn, |
---|
64 | updatefn, |
---|
65 | yieldtime, |
---|
66 | worktime) |
---|
67 | self.pr = pr |
---|
68 | self.nfunc = nfunc |
---|
69 | self.error_func = error_func |
---|
70 | self.starttime = 0 |
---|
71 | |
---|
72 | def isquit(self): |
---|
73 | try: |
---|
74 | CalcThread.isquit(self) |
---|
75 | except KeyboardInterrupt: |
---|
76 | printEVT("P(r) calc interrupted") |
---|
77 | raise KeyboardInterrupt |
---|
78 | |
---|
79 | def compute(self): |
---|
80 | import time |
---|
81 | try: |
---|
82 | self.starttime = time.time() |
---|
83 | # If the current alpha is zero, try |
---|
84 | # another value |
---|
85 | if self.pr.alpha<=0: |
---|
86 | self.pr.alpha = 0.0001 |
---|
87 | |
---|
88 | # Perform inversion to find the largest alpha |
---|
89 | out, cov = self.pr.lstsq(self.nfunc) |
---|
90 | elapsed = time.time()-self.starttime |
---|
91 | initial_alpha = self.pr.alpha |
---|
92 | initial_peaks = self.pr.get_peaks(out) |
---|
93 | |
---|
94 | # Try the inversion with the estimated alpha |
---|
95 | self.pr.alpha = self.pr.suggested_alpha |
---|
96 | out, cov = self.pr.lstsq(self.nfunc) |
---|
97 | |
---|
98 | npeaks = self.pr.get_peaks(out) |
---|
99 | # if more than one peak to start with |
---|
100 | # just return the estimate |
---|
101 | if npeaks>1: |
---|
102 | message = "Your P(r) is not smooth, please check your inversion parameters" |
---|
103 | self.complete(alpha=self.pr.suggested_alpha, message=message, elapsed=elapsed) |
---|
104 | else: |
---|
105 | |
---|
106 | # Look at smaller values |
---|
107 | # We assume that for the suggested alpha, we have 1 peak |
---|
108 | # if not, send a message to change parameters |
---|
109 | alpha = self.pr.suggested_alpha |
---|
110 | best_alpha = self.pr.suggested_alpha |
---|
111 | found = False |
---|
112 | for i in range(10): |
---|
113 | self.pr.alpha = (0.33)**(i+1)*alpha |
---|
114 | out, cov = self.pr.lstsq(self.nfunc) |
---|
115 | #osc = self.pr.oscillations(out) |
---|
116 | #print self.pr.alpha, osc |
---|
117 | |
---|
118 | peaks = self.pr.get_peaks(out) |
---|
119 | print self.pr.alpha, peaks |
---|
120 | if peaks>1: |
---|
121 | found = True |
---|
122 | break |
---|
123 | best_alpha = self.pr.alpha |
---|
124 | |
---|
125 | # If we didn't find a turning point for alpha and |
---|
126 | # the initial alpha already had only one peak, |
---|
127 | # just return that |
---|
128 | if not found and initial_peaks==1 and initial_alpha<best_alpha: |
---|
129 | best_alpha = initial_alpha |
---|
130 | |
---|
131 | # Check whether the size makes sense |
---|
132 | message=None |
---|
133 | |
---|
134 | if not found: |
---|
135 | message = "None" |
---|
136 | elif best_alpha>=0.5*self.pr.suggested_alpha: |
---|
137 | # best alpha is too big, return a |
---|
138 | # reasonable value |
---|
139 | message = "The estimated alpha for your system is too large. " |
---|
140 | message += "Try increasing your maximum distance." |
---|
141 | |
---|
142 | self.complete(alpha=best_alpha, message=None, elapsed=elapsed) |
---|
143 | |
---|
144 | except: |
---|
145 | if not self.error_func==None: |
---|
146 | printEVT("EstimatePr.compute: %s" % sys.exc_value) |
---|
147 | |
---|
148 | |
---|