Reduced precision requirement to 7 digits

Author: haoxingz

pymatbridge/tests/test_precision.py

@@ -23,7 +23,7 @@ def test_float64_roundtrip(self):
         for i in range(0,10):
             val = np.float64(rd.random())
             res = self.mlab.run_func('precision_pass.m', {'val':val})['result']
-            npt.assert_almost_equal(res, val, decimal=8, err_msg="Float64 roundtrip error")
+            npt.assert_almost_equal(res, val, decimal=7, err_msg="Float64 roundtrip error")
 
     # Add two 64-bit floating number in Matlab and return the sum
     def test_float64_sum(self):
@@ -32,7 +32,7 @@ def test_float64_sum(self):
             val2 = np.float64(rd.random())
 
             res = self.mlab.run_func('precision_sum.m', {'val1':val1, 'val2':val2})['result']
-            npt.assert_almost_equal(res, val1 + val2, decimal=8, err_msg="Float64 sum error")
+            npt.assert_almost_equal(res, val1 + val2, decimal=7, err_msg="Float64 sum error")
 
     # Multiply two 64-bit floating number in Matlab and return the product
     def test_float64_multiply(self):
@@ -41,7 +41,7 @@ def test_float64_multiply(self):
             val2 = np.float64(rd.random())
 
             res = self.mlab.run_func('precision_multiply.m', {'val1':val1, 'val2':val2})['result']
-            npt.assert_almost_equal(res, val1 * val2, decimal=8, err_msg="Float64 multiply error")
+            npt.assert_almost_equal(res, val1 * val2, decimal=7, err_msg="Float64 multiply error")
 
     # Make a division in Matlab and return the results
     def test_float64_divide(self):
@@ -50,12 +50,12 @@ def test_float64_divide(self):
             val2 = np.float64(rd.random())
 
             res = self.mlab.run_func('precision_divide.m', {'val1':val1, 'val2':val2})['result']
-            npt.assert_almost_equal(res, val1 / val2, decimal=8, err_msg="Float64 divide error")
+            npt.assert_almost_equal(res, val1 / val2, decimal=7, err_msg="Float64 divide error")
 
     # Calculate the square root in Matlab and return the result
     def test_float64_sqrt(self):
         for i in range(0,10):
             val = np.float64(rd.random())
 
             res = self.mlab.run_func('precision_sqrt.m', {'val':val})['result']
-            npt.assert_almost_equal(res, np.sqrt(val), decimal=8, err_msg="Float64 square root error")
+            npt.assert_almost_equal(res, np.sqrt(val), decimal=7, err_msg="Float64 square root error")