diff --git a/_unittests/ut_onnxrt/test_onnxrt_python_runtime_.py b/_unittests/ut_onnxrt/test_onnxrt_python_runtime_.py
index bbdd667c7..84551bebe 100644
--- a/_unittests/ut_onnxrt/test_onnxrt_python_runtime_.py
+++ b/_unittests/ut_onnxrt/test_onnxrt_python_runtime_.py
@@ -21,7 +21,7 @@
 except ImportError:
     from sklearn.utils.testing import ignore_warnings
 from skl2onnx.algebra.onnx_ops import (  # pylint: disable=E0611
-    OnnxAbs, OnnxAdd, OnnxArgMax, OnnxArgMin,
+    OnnxAbs, OnnxAdd, OnnxArgMax, OnnxArgMin, OnnxAtan,
     OnnxBatchNormalization,
     OnnxConcat,
     OnnxCeil, OnnxClip, OnnxConstant, OnnxConstantOfShape,
@@ -396,6 +396,28 @@ def test_onnxt_runtime_argmin_12(self):
         self.assertEqualArray(numpy.array([2, 1], dtype=numpy.int64),
                               got['Y'], decimal=6)
 
+    def test_onnxt_runtime_atan(self):
+        self.common_test_onnxt_runtime_unary(OnnxAtan, numpy.arctan)
+
+    def test_onnxt_runtime_atan2(self):
+        test_pairs = [[y, x] for x in [3., -4., 0.] for y in [5., -6., 0.]]
+        y_val = numpy.array([y for y, x in test_pairs], dtype=numpy.float32)
+        x_val = numpy.array([x for y, x in test_pairs], dtype=numpy.float32)
+
+        def atan2(y, x):
+            # size: 100000
+            # timeit arctan: 0.00205
+            # timeit arctan2: 0.00361
+            # timeit atan2: 0.00599
+            sx = numpy.sign(x)
+            sy = numpy.sign(y)
+            pi_part = (sy + sx * (sy ** 2 - 1)) * (sx - 1) * (-numpy.pi / 2)
+            atan_part = numpy.arctan(y / (x + (1 - sx ** 2))) * sx ** 2
+            return atan_part + pi_part
+
+        self.assertEqualArray(
+            numpy.arctan2(y_val, x_val), atan2(y_val, x_val))
+
     def test_onnxt_runtime_batch_normalization(self):
         # input size: (1, 2, 1, 3)
         x = numpy.array([[[[-1, 0, 1]], [[2, 3, 4]]]]).astype(numpy.float32)
diff --git a/mlprodict/onnxrt/onnx_inference.py b/mlprodict/onnxrt/onnx_inference.py
index e4a6d9460..54892bdef 100644
--- a/mlprodict/onnxrt/onnx_inference.py
+++ b/mlprodict/onnxrt/onnx_inference.py
@@ -1007,6 +1007,9 @@ def _build_compile_run(self):
                 # to the onnx graph
                 print(oinf2)
         """
+        def clean_name(name):
+            return name.replace(":", "_").replace('.', '_')
+
         # inits
         code = ['def compiled_run(dict_inputs):']
         context = {}
@@ -1018,25 +1021,33 @@ def _build_compile_run(self):
         inputs = self.input_names
         code.append("    # inputs")
         for inp in inputs:
-            code.append("    {0} = dict_inputs['{0}']".format(inp))
+            code.append("    {0} = dict_inputs['{1}']".format(
+                clean_name(inp), inp))
 
         # code
         for i, node in enumerate(self.sequence_):
             name = "n{}_{}".format(i, node.ops_.__class__.__name__.lower())
             context[name] = node.ops_._run
             code.append('    ({1}, ) = {2}({0})'.format(
-                ', '.join(node.inputs), ', '.join(node.outputs), name))
+                ', '.join(map(clean_name, node.inputs)),
+                ', '.join(map(clean_name, node.outputs)),
+                name))
 
         # return
         code.append('    return {')
         for out in self.output_names:
-            code.append("        '{0}': {0},".format(out))
+            code.append("        '{1}': {0},".format(
+                clean_name(out), out))
         code.append('    }')
         final_code = '\n'.join(code)
 
         # compile the outcome
         context['self'] = self
-        obj = compile(final_code, "<string>", 'exec')
+        try:
+            obj = compile(final_code, "<string>", 'exec')
+        except SyntaxError as e:
+            raise SyntaxError(
+                "Unable to compile\n#####\n{}".format(final_code)) from e
         fcts_obj = [_ for _ in obj.co_consts
                     if _ is not None and not isinstance(_, (bool, str, int))]
         fct = make_callable("compiled_run", fcts_obj[0], final_code, context)
diff --git a/mlprodict/onnxrt/ops_cpu/_op_list.py b/mlprodict/onnxrt/ops_cpu/_op_list.py
index 628ebb734..40b9412b5 100644
--- a/mlprodict/onnxrt/ops_cpu/_op_list.py
+++ b/mlprodict/onnxrt/ops_cpu/_op_list.py
@@ -11,6 +11,7 @@
 from .op_argmax import ArgMax
 from .op_argmin import ArgMin
 from .op_array_feature_extractor import ArrayFeatureExtractor
+from .op_atan import Atan
 from .op_batch_normalization import BatchNormalization
 from .op_binarizer import Binarizer
 from .op_cast import Cast
diff --git a/mlprodict/onnxrt/ops_cpu/op_atan.py b/mlprodict/onnxrt/ops_cpu/op_atan.py
new file mode 100644
index 000000000..430e4a49f
--- /dev/null
+++ b/mlprodict/onnxrt/ops_cpu/op_atan.py
@@ -0,0 +1,26 @@
+# -*- encoding: utf-8 -*-
+# pylint: disable=E0203,E1101,C0111
+"""
+@file
+@brief Runtime operator.
+"""
+import numpy
+from ._op import OpRunUnaryNum
+
+
+class Atan(OpRunUnaryNum):
+
+    def __init__(self, onnx_node, desc=None, **options):
+        OpRunUnaryNum.__init__(self, onnx_node, desc=desc,
+                               **options)
+
+    def _run(self, x):  # pylint: disable=W0221
+        if self.inplaces.get(0, False):
+            return self._run_inplace(x)
+        return (numpy.arctan(x), )
+
+    def _run_inplace(self, x):
+        return (numpy.arctan(x, out=x), )
+
+    def to_python(self, inputs):
+        return self._to_python_numpy(inputs, 'arctan')