make quantum and interfaces in init

Author: refraction-ray

tensorcircuit/__init__.py

@@ -18,4 +18,6 @@
 from .densitymatrix import DMCircuit
 from .densitymatrix2 import DMCircuit2
 from .gates import num_to_tensor, array_to_tensor
+from . import interfaces
 from . import templates
+from . import quantum

tensorcircuit/interfaces.py

@@ -5,7 +5,6 @@
 from typing import Any, Callable, Tuple, Optional
 
 import numpy as np
-import torch
 
 from .cons import backend, dtypestr
 from .backends import get_backend  # type: ignore
@@ -78,6 +77,8 @@ def is_sequence(x: Any) -> bool:
 
 
 def torch_interface(fun: Callable[..., Any], jit: bool = False) -> Callable[..., Any]:
+    import torch
+
     def vjp_fun(x: Tensor, v: Tensor) -> Tuple[Tensor, Tensor]:
         return backend.vjp(fun, x, v)  # type: ignore
 

tests/test_interfaces.py

@@ -11,14 +11,16 @@
 
 try:
     import torch
+
+    is_torch = True
 except ImportError:
-    pytest.skip("torch not available", allow_module_level=True)
+    is_torch = False
 
 import numpy as np
 import tensorcircuit as tc
-from tensorcircuit import interfaces
 
 
+@pytest.mark.skipif(is_torch is False, reason="torch not installed")
 @pytest.mark.parametrize("backend", [lf("tfb"), lf("jaxb")])
 def test_torch_interface(backend):
     n = 4
@@ -38,7 +40,7 @@ def f(param):
 
     f_jit = tc.backend.jit(f)
 
-    f_jit_torch = interfaces.torch_interface(f_jit)
+    f_jit_torch = tc.interfaces.torch_interface(f_jit)
 
     param = torch.ones([4, n], requires_grad=True)
     l = f_jit_torch(param)
@@ -76,7 +78,7 @@ def f2(paramzz, paramx):
         )
         return tc.backend.real(loss1), tc.backend.real(loss2)
 
-    f2_torch = interfaces.torch_interface(f2, jit=True)
+    f2_torch = tc.interfaces.torch_interface(f2, jit=True)
 
     paramzz = torch.ones([2, n], requires_grad=True)
     paramx = torch.ones([2, n], requires_grad=True)
@@ -92,7 +94,7 @@ def f2(paramzz, paramx):
     def f3(x):
         return tc.backend.real(x ** 2)
 
-    f3_torch = interfaces.torch_interface(f3)
+    f3_torch = tc.interfaces.torch_interface(f3)
     param3 = torch.ones([2], dtype=torch.complex64, requires_grad=True)
     l3 = f3_torch(param3)
     l3 = torch.sum(l3)
@@ -120,7 +122,7 @@ def f(param):
         )
         return tc.backend.real(loss)
 
-    f_scipy = interfaces.scipy_optimize_interface(f, shape=[2, n])
+    f_scipy = tc.interfaces.scipy_optimize_interface(f, shape=[2, n])
     r = optimize.minimize(f_scipy, np.zeros([2 * n]), method="L-BFGS-B", jac=True)
     # L-BFGS-B may has issue with float32
     # see: https://github.com/scipy/scipy/issues/5832