rejection sampling used in accuracy vs sample size test

test_synthetic.py

@@ -561,9 +561,9 @@ def test_hypercompression():
 
 
 def test_sample_size():
-    nsim = 500
+    nsim = 1000
     level = 0.05
-    sizes = [100, 500, 1000, 2500, 5000]
+    sizes = [100, 400, 700, 1000, 1300, 1600, 2000]
     suppfX = range(-7, 8)
     srcX = "geometric"
 
@@ -575,14 +575,28 @@ def test_sample_size():
     sys.stdout.flush()
     # progress(0, len(sizes))
     for k, size in enumerate(sizes):
+        nsamples = 0
         nc_dc, nc_dr, nc_cisc, nc_crispe, nc_crisp = 0, 0, 0, 0, 0
-        for j in xrange(nsim):
+        while nsamples < nsim:
             X = generate_X(srcX, size)
             suppX = list(set(X))
             f = map_randomly(suppX, suppfX)
             N = generate_additive_N(size)
             Y = [f[X[i]] + N[i] for i in xrange(size)]
 
+            # check if f(x) + supp N are disjoint for x in domx
+            suppN = set(N)
+            decomps = []
+            for x in suppX:
+                fx = f[x]
+                sum_fx_suppN = set([fx + n for n in suppN])
+                decomps.append(sum_fx_suppN)
+
+            non_overlapping_noise = are_disjoint(decomps)
+            if non_overlapping_noise:
+                continue
+
+            nsamples += 1
             dc_score = dc(dc_compat(X), dc_compat(Y))
             dr_score = dr(X, Y, level)
             cisc_score = cisc(X, Y)
@@ -609,4 +623,5 @@ def test_sample_size():
 
 
 if __name__ == "__main__":
-    test_accuracy()
+    test_sample_size()
+    # test_accuracy()