fix: prevent bin-edges machine rounding bug when requesting number of bins

- Added dedicated unit test to cover this.

Author: mbaak

histogrammar/primitives/bin.py

@@ -1097,7 +1097,7 @@ def bin_edges(self, low=None, high=None):
                 maxBin -= 1
             high = self.low + self.bin_width() * (maxBin + 1)
         # new low and high values reset, so redo num_bins
-        num_bins = self.num_bins(low, high)
+        num_bins = self.num_bins(low + np.finfo(float).eps, high - np.finfo(float).eps)
         edges = np.linspace(low, high, num_bins + 1)
         return edges
 

histogrammar/primitives/sparselybin.py

@@ -751,7 +751,7 @@ def bin_edges(self, low=None, high=None):
                 maxBin -= 1
             high = self.origin + self.bin_width() * (maxBin + 1)
         # number of bins, after low/high adjustments
-        num_bins = self.num_bins(low, high)
+        num_bins = self.num_bins(low + np.finfo(float).eps, high - np.finfo(float).eps)
         edges = np.linspace(low, high, num_bins + 1)
         return edges
 

tests/testnumpy.py

@@ -723,6 +723,7 @@ def test_num_bins(self):
                 hist3 = hg.SparselyBin(origin=0.0, binWidth=1.0, quantity=unit('A'))
                 hist4 = hg.Bin(num=20, low=0.0, high=20., quantity=unit('A'))
                 hist5 = hg.Bin(num=20, low=0.0, high=20., quantity=unit('A'))
+                hist6 = hg.Bin(num=201, low=0.0, high=1.005)
 
                 # fill them
                 hist2.fill.numpy(df1)
@@ -734,16 +735,19 @@ def test_num_bins(self):
                 assert hist3.num_bins() == 18
                 assert hist4.num_bins() == 20
                 assert hist5.num_bins() == 20
+                assert hist6.num_bins() == 201
 
                 assert hist2.num_bins(low=10, high=25) == 15
                 assert hist3.num_bins(low=10, high=25) == 15
                 assert hist4.num_bins(low=10, high=25) == 10
                 assert hist5.num_bins(low=10, high=25) == 10
+                assert hist6.num_bins(low=0.2089, high=0.9333) == 146
 
                 assert hist2.num_bins(low=-10, high=28) == 38
                 assert hist3.num_bins(low=-10, high=28) == 38
                 assert hist4.num_bins(low=-10, high=28) == 20
                 assert hist5.num_bins(low=-10, high=28) == 20
+                assert hist6.num_bins(low=0.205, high=0.935) == 146
 
     def test_most_probable_value(self):
         """ Test getting most probable value or label from histogram
@@ -917,6 +921,7 @@ def test_bin_edges(self):
                 hist3 = hg.SparselyBin(origin=0.0, binWidth=1.0, quantity=unit('A'))
                 hist4 = hg.Bin(num=10, low=0.0, high=10., quantity=unit('A'))
                 hist5 = hg.Bin(num=10, low=0.0, high=10., quantity=unit('A'))
+                hist6 = hg.Bin(num=201, low=0.0, high=1.005)
 
                 # fill them
                 hist2.fill.numpy(df1)
@@ -937,6 +942,10 @@ def test_bin_edges(self):
                                               2., 3., 4., 5., 6., 7., 8., 9., 10.])
                 np.testing.assert_array_equal(hist5.bin_edges(low=1.1, high=5.4), [1., 2., 3., 4., 5., 6.])
 
+                assert len(hist6.bin_edges()) == 202
+                assert len(hist6.bin_edges(low=0.2089, high=0.9333)) == 147
+                assert len(hist6.bin_edges(low=0.205, high=0.935)) == 147
+
     def test_bin_width(self):
         """ Test getting the bin width of bin and sparselybin histograms
         """