[Tests] Adding tests for math primitives (#412)

Closes #195

tests/frontends/torch/test_torch_interoperability.py

@@ -362,7 +362,7 @@ def test_torch_einsum(equation, operand_shapes):
 
     atol = 5e-2
     if equation == 'abcd,cd->ab':
-        atol = 8e-2
+        atol = 1e-1
 
     check_module(
         FunctionalModule(op=lambda *args: torch.einsum(equation, *args)), args=operands_torch, atol=atol, rtol=1e-4

tests/operators/test_arithmetic.py

@@ -72,21 +72,116 @@ def test_sqrt(shape):
     check_unary(shape, np.float32, np.sqrt, ops.sqrt, positive=True)
 
 
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_sin(shape):
+    check_unary(shape, np.float32, np.sin, ops.sin)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_cos(shape):
+    check_unary(shape, np.float32, np.cos, ops.cos)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_tan(shape):
+    check_unary(shape, np.float32, np.tan, ops.tan)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_sinh(shape):
+    check_unary(shape, np.float32, np.sinh, ops.sinh)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_cosh(shape):
+    check_unary(shape, np.float32, np.cosh, ops.cosh)
+
+
 @pytest.mark.parametrize("shape", unary_op_shapes)
 def test_tanh(shape):
     check_unary(shape, np.float32, np.tanh, ops.tanh)
 
 
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_asin(shape):
+    check_unary(shape, np.float32, np.arcsin, ops.asin)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_acos(shape):
+    check_unary(shape, np.float32, np.arccos, ops.acos)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_atan(shape):
+    check_unary(shape, np.float32, np.arctan, ops.atan)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_asinh(shape):
+    check_unary(shape, np.float32, np.arcsinh, ops.asinh)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_acosh(shape):
+    check_unary(shape, np.float32, np.arccosh, ops.acosh)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_atanh(shape):
+    check_unary(shape, np.float32, np.arctanh, ops.atanh)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_exp(shape):
+    check_unary(shape, np.float32, np.exp, ops.exp)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_expm1(shape):
+    check_unary(shape, np.float32, np.expm1, ops.expm1)
+
+
 @pytest.mark.parametrize("shape", unary_op_shapes)
 def test_erf(shape):
     check_unary(shape, np.float32, np.vectorize(math.erf), ops.erf)
 
 
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_sqrt(shape):
+    check_unary(shape, np.float32, np.sqrt, ops.sqrt, positive=True)
+
+
 @pytest.mark.parametrize("shape", unary_op_shapes)
 def test_rsqrt(shape):
     check_unary(shape, np.float32, lambda v: np.reciprocal(np.sqrt(v)), ops.rsqrt, positive=True)
 
 
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_log(shape):
+    check_unary(shape, np.float32, np.log, ops.log, positive=True)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_log2(shape):
+    check_unary(shape, np.float32, np.log2, ops.log2, positive=True)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_log10(shape):
+    check_unary(shape, np.float32, np.log10, ops.log10, positive=True)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_log1p(shape):
+    check_unary(shape, np.float32, np.log1p, ops.log1p, positive=True)
+
+
+@pytest.mark.parametrize("shape", unary_op_shapes)
+def test_round(shape):
+    check_unary(shape, np.float32, np.round, ops.round)
+
+
 @pytest.mark.parametrize("shape", unary_op_shapes)
 def test_neg(shape):
     check_unary(shape, np.float32, np.negative, ops.negative)
@@ -142,6 +237,31 @@ def test_ceil(a_shape):
     check_unary(a_shape, np.float32, np.ceil, ops.ceil)
 
 
+@pytest.mark.parametrize("a_shape", unary_op_shapes)
+def test_floor(a_shape):
+    check_unary(a_shape, np.float32, np.floor, ops.floor)
+
+
+@pytest.mark.parametrize("a_shape", unary_op_shapes)
+def test_trunc(a_shape):
+    check_unary(a_shape, np.float32, np.trunc, ops.trunc)
+
+
+@pytest.mark.parametrize("a_shape", unary_op_shapes)
+def test_isfinite(a_shape):
+    check_unary(a_shape, np.float32, np.isfinite, ops.isfinite)
+
+
+@pytest.mark.parametrize("a_shape", unary_op_shapes)
+def test_isinf(a_shape):
+    check_unary(a_shape, np.float32, np.isinf, ops.isinf)
+
+
+@pytest.mark.parametrize("a_shape", unary_op_shapes)
+def test_isnan(a_shape):
+    check_unary(a_shape, np.float32, np.isnan, ops.isnan)
+
+
 @pytest.mark.parametrize("a_shape", [[20]])
 def test_cast_from_fp16(a_shape):
     check_unary(a_shape, np.float16, np.int8, lambda x: ops.cast(x, "int8"))