Flake8 fixes.

theano/sandbox/gpuarray/nerv.py

 import os.path
-import numpy
 import theano
-from theano import Op, Apply, Variable, tensor
+from theano import Apply, Variable, tensor
 
 from theano.compile import optdb
 from theano.compile.ops import shape_i
@@ -9,7 +8,7 @@ from theano.gof import local_optimizer, COp
 from theano.scalar import as_scalar, constant
 
 from . import opt
-from .basic_ops import (as_gpuarray_variable, gpu_alloc, gpu_from_host,
+from .basic_ops import (as_gpuarray_variable, gpu_from_host,
                         host_from_gpu, GpuAllocEmpty)
 from .opt_util import alpha_merge, output_merge
 from .pycuda_helper import ensure_pycuda_context
@@ -54,7 +53,7 @@ class Gemm16(COp):
         COp.__init__(self, ["gemm16.c"], "gemm16")
         self.relu = relu
         # relu = True will require more work in optimizations.
-        assert self.relu == False
+        assert self.relu is False
         self.inplace = inplace
         if self.inplace:
             self.destroy_map = {0: [0]}
@@ -75,7 +74,7 @@ class Gemm16(COp):
         return Apply(self, [C, alpha, A, B, beta], [C.type()])
 
     def perform(self, node, inputs, outputs):
-        ctx = ensure_pycuda_context()
+        ensure_pycuda_context()
         C, alpha, A, B, beta = inputs
         # The nervana code does not support the case where both inputs
         # are trans, so we need to copy one if them if that is the
@@ -143,7 +142,7 @@ if (GpuKernel_init(&k_%(name)s, c->ops, c->ctx, 1, &bcode, &sz,
     def c_init_code_struct(self, node, nodename, sub):
         codel = [super(Gemm16, self).c_init_code_struct(node, nodename, sub)]
         for name in self.KERN_NAMES:
-            codel.append("memset(&k_{0}, 0, sizeof(GpuKernel));".format(name));
+            codel.append("memset(&k_{0}, 0, sizeof(GpuKernel));".format(name))
         codel.append("const char *bcode;")
         codel.append("size_t sz;")
         codel.append("PyGpuContextObject *c = pygpu_default_context();")
@@ -175,6 +174,7 @@ def local_dot_to_gemm16(node):
             shape_i(A, 0, fgraph), shape_i(B, 1, fgraph))
         return [host_from_gpu(Gemm16()(C, 1.0, A, B, 0.0))]
 
+
 @opt.register_opt()
 @alpha_merge(Gemm16, alpha_in=1, beta_in=4, nd=2)
 def local_gemm16_alpha_merge(node, *inputs):

theano/sandbox/gpuarray/opt_util.py

@@ -2,7 +2,6 @@ from functools import wraps
 
 import numpy
 
-import theano
 from theano import scalar as scal, Constant
 from theano.gof import local_optimizer
 from theano.tensor import (DimShuffle, get_scalar_constant_value,
@@ -13,11 +12,12 @@ from .elemwise import GpuDimShuffle, GpuElemwise
 
 _one = scal.constant(numpy.asarray(1.0, dtype='float32'))
 
+
 def grab_cpu_scalar(v, nd):
     if v.owner is not None:
         n = v.owner
         if (isinstance(n.op, GpuDimShuffle) and
-            n.op.new_order == ('x',) * nd):
+                n.op.new_order == ('x',) * nd):
             return host_from_gpu(n.inputs[0])
         elif (isinstance(n.op, DimShuffle) and
               n.op.new_order == ('x',) * nd):
@@ -28,7 +28,7 @@ def grab_cpu_scalar(v, nd):
             return None
     else:
         if (isinstance(v, Constant) and
-            v.broadcastable == (True,) * nd):
+                v.broadcastable == (True,) * nd):
             return v.dimshuffle(())
 
 
@@ -64,8 +64,8 @@ def alpha_merge(cls, alpha_in, beta_in, nd):
         @wraps(maker)
         def opt(node):
             if (isinstance(node.op, GpuElemwise) and
-                node.op.scalar_op == scal.mul and
-                node.nin == 2):
+                    node.op.scalar_op == scal.mul and
+                    node.nin == 2):
                 targ = find_node(node.inputs[0], cls)
                 if targ is None:
                     targ = find_node(node.inputs[1], cls)
@@ -88,8 +88,8 @@ def output_merge(cls, alpha_in, beta_in, out_in, nd):
         @wraps(maker)
         def opt(node):
             if (isinstance(node.op, GpuElemwise) and
-                node.op.scalar_op == scal.add and
-                node.nin == 2):
+                    node.op.scalar_op == scal.add and
+                    node.nin == 2):
                 targ = find_node(node.inputs[0], cls)
                 W = node.inputs[1]
                 if targ is None:

theano/sandbox/gpuarray/pycuda_helper.py

@@ -8,6 +8,7 @@ except ImportError:
 pycuda_initialized = False
 pycuda_context = None
 
+
 def ensure_pycuda_context():
     global pycuda_context, pycuda_initialized
     if not pycuda_initialized: