first version of the tutorial adding something on the 2D Ising model

Author: acevedo-s

dadapy/hamming.py

@@ -26,7 +26,6 @@ def __init__(
     ):
         self.q = q
         self.coordinates = coordinates
-        self.metric = "hamming"
         self.distances = distances
         self.crossed_distances = crossed_distances
         self.verbose = verbose
@@ -45,7 +44,8 @@ def compute_distances(
         check_format=True,
     ):
         """
-        Computes all to all distances in dataset and stores them in the matrix self.distances
+        Computes all to all distances in dataset and stores them
+        in the matrix "self.distances" of shape (Ns,Ns), where Ns is the number of samples
         """
         if self.q == 2:
             self.distances = jcompute_distances(
@@ -63,15 +63,20 @@ def D_histogram(
         compute_flag=0,  # 1 to compute histogram (else it is loaded)
         save=False,  # 1 to save computed histogram
         resultsfolder="results/hist/",
+        filename="counts.txt",
     ):
         """
-        Given the computed distances this routine computes the histogram (Pemp) and saves is
+        Given the computed distances, this routine computes the histogram (Pemp).
+        It defines
+        - self.D_values, a vector containing the sampled distances
+        - self.D_counts, a vector containing how many times each distance was sampled
+        - self.D_probs, self.counts normalized by the total number of counts observed.
         """
         assert self.crossed_distances == 0
 
         if save:
             os.makedirs(resultsfolder, exist_ok=True)
-        c_fname = resultsfolder + "counts"
+        _filename = resultsfolder + filename
 
         if compute_flag:
             self.D_values, self.D_counts = np.unique(self.distances, return_counts=True)
@@ -90,17 +95,25 @@ def D_histogram(
 
             if save:
                 np.savetxt(
-                    fname=c_fname + ".txt",
+                    fname=_filename,
                     X=np.transpose([self.D_values, self.D_counts]),
                     fmt="%d,%d",
                 )
         else:
-            f = np.loadtxt(c_fname + ".txt", delimiter=",", dtype=int)
+            f = np.loadtxt(_filename, delimiter=",", dtype=int)
             self.D_values = f[:, 0]
             self.D_counts = f[:, 1]
             self.D_probs = self.D_counts / np.sum(self.D_counts)
 
     def set_r_quantile(self, alpha, round=True, precision=10):
+        """
+        Defines
+
+        - self.r as the quantile of order alpha of self.D_probs,
+        which can be used for rmax or rmin,
+        to discard distances larger than rmax or smaller than rmin, respectively.
+        - self.r_idx as the index of self.r in self.D_values
+        """
         if round:
             alpha = np.round(alpha, precision)
             self.D_probs = np.round(self.D_probs, precision)
@@ -114,28 +127,10 @@ def set_r_quantile(self, alpha, round=True, precision=10):
         self.r = int(self.D_values[self.r_idx])
         return
 
-    def set_r(self, r=None, n_sigma=3):
-        if r is not None:
-            self.r = r
-        else:
-            self.compute_moments()
-            r = int(np.round(self.D_mu_emp - n_sigma * np.sqrt(self.D_var_emp), 0))
-            if r >= self.D_values[0]:
-                self.r = r
-            else:
-                self.r = self.D_values[0]
-
-    def set_r_idx(self, r_idx=None):
-        if r_idx is not None:
-            self.r_idx = r_idx
-        else:
-            aux = np.asarray(self.r == self.D_values).nonzero()[0]
-            if len(aux) > 0:
-                self.r_idx = aux[0]
-            else:
-                self.r_idx = 0
-
     def compute_moments(self):
+        """
+        computes the empirical mean and variance of H.D_probs
+        """
         self.D_mu_emp = np.dot(self.D_probs, self.D_values)
         self.D_var_emp = np.dot(self.D_probs, self.D_values**2) - self.D_mu_emp**2
 
@@ -420,7 +415,7 @@ def minimize_KL(Op):
 class BID:
     def __init__(
         self,
-        H=None,  # instance of Hamming class defined above
+        H=Hamming(),  # instance of Hamming class defined above
         Op=None,  # instance of Optimizer class defined above
         alphamin=0.0,
         alphamax=0.2,

examples/datasets/hamming/Ising2D/hist/L100_T1.00_Ns5000D_counts.txt

@@ -0,0 +1,55 @@
+0,6666
+1,41186
+2,137369
+3,313850
+4,540111
+5,750004
+6,874420
+7,889678
+8,802703
+9,651278
+10,479772
+11,323353
+12,200816
+13,115544
+14,61751
+15,31203
+16,15125
+17,7053
+18,3306
+19,1465
+20,678
+21,252
+22,113
+23,38
+24,15
+25,4
+26,1
+27,2
+9974,1
+9975,2
+9976,13
+9977,48
+9978,100
+9979,309
+9980,659
+9981,1533
+9982,3227
+9983,7158
+9984,15026
+9985,31304
+9986,61676
+9987,115611
+9988,201143
+9989,323535
+9990,480037
+9991,651106
+9992,802960
+9993,889424
+9994,875168
+9995,750171
+9996,540532
+9997,313787
+9998,137434
+9999,41066
+10000,6714

examples/datasets/hamming/Ising2D/hist/L100_T1.10_Ns5000D_counts.txt

@@ -0,0 +1,81 @@
+0,6
+1,33
+2,270
+3,1236
+4,4595
+5,13063
+6,31216
+7,63822
+8,114823
+9,184118
+10,269691
+11,363323
+12,454201
+13,531295
+14,579959
+15,596097
+16,578923
+17,535015
+18,468375
+19,389550
+20,309200
+21,237401
+22,174358
+23,122598
+24,82780
+25,54954
+26,35340
+27,21848
+28,13026
+29,7502
+30,4355
+31,2361
+32,1256
+33,613
+34,298
+35,140
+36,68
+37,34
+38,9
+39,3
+40,1
+9961,2
+9962,8
+9963,28
+9964,54
+9965,121
+9966,288
+9967,599
+9968,1237
+9969,2296
+9970,4323
+9971,7625
+9972,13049
+9973,21842
+9974,35311
+9975,55161
+9976,83068
+9977,122785
+9978,174037
+9979,237456
+9980,310060
+9981,388835
+9982,467920
+9983,534727
+9984,580120
+9985,596245
+9986,580723
+9987,530779
+9988,454515
+9989,363580
+9990,269742
+9991,184248
+9992,114511
+9993,64065
+9994,31117
+9995,13145
+9996,4543
+9997,1266
+9998,264
+9999,45
+10000,4

examples/datasets/hamming/Ising2D/hist/L100_T1.20_Ns5000D_counts.txt

@@ -0,0 +1,111 @@
+1,1
+7,5
+8,18
+9,81
+10,274
+11,798
+12,1775
+13,3685
+14,7443
+15,13464
+16,22482
+17,36597
+18,56221
+19,80779
+20,112896
+21,151290
+22,193759
+23,238575
+24,285791
+25,328497
+26,365695
+27,395586
+28,414940
+29,421795
+30,416480
+31,401741
+32,376444
+33,344360
+34,305943
+35,265864
+36,225529
+37,186434
+38,150782
+39,118859
+40,91679
+41,68849
+42,50612
+43,36474
+44,25792
+45,17612
+46,11700
+47,7736
+48,4938
+49,3062
+50,1917
+51,1110
+52,636
+53,363
+54,194
+55,108
+56,43
+57,29
+58,13
+59,3
+60,2
+61,1
+9939,1
+9940,1
+9941,5
+9942,10
+9943,22
+9944,47
+9945,91
+9946,192
+9947,332
+9948,610
+9949,1111
+9950,1835
+9951,3031
+9952,4968
+9953,7691
+9954,11862
+9955,17512
+9956,25627
+9957,36367
+9958,50879
+9959,68924
+9960,91656
+9961,119019
+9962,150657
+9963,186323
+9964,225640
+9965,266036
+9966,306072
+9967,344005
+9968,377027
+9969,401237
+9970,417891
+9971,421702
+9972,413982
+9973,395937
+9974,366806
+9975,328491
+9976,284850
+9977,239795
+9978,193465
+9979,151035
+9980,112872
+9981,81498
+9982,55695
+9983,36525
+9984,22859
+9985,13419
+9986,7324
+9987,3844
+9988,1807
+9989,745
+9990,294
+9991,99
+9992,18
+9993,1