def test_1(self):
G = Group(5, 'V = I; I')
Z = np.ones(5)
DenseConnection(Z, G('I'), 1)
DenseConnection(Z, G('I'), 2)
G.run()
assert np_equal(3 * np.ones(5), G('V'))
def test_2(self):
kernel = np.ones(3) * np.NaN
C = DenseConnection(np.ones(3),
np.ones(3),
kernel,
equation='dW/dt = 1')
C.evaluate(dt=.1)
assert np_equal(C.weights, np.zeros((3, 3)))
def test_1(self):
kernel = np.ones(1)
C = DenseConnection(np.ones(3),
np.ones(3),
kernel,
equation='dW/dt = 1')
C.evaluate(dt=.1)
assert np_equal(C.weights, np.identity(3) * 1.1)
def test_3(self):
kernel = np.ones(3)
kernel[1] = np.NaN
C = DenseConnection(np.ones(3), np.ones(3), kernel,
equation = 'dW/dt = 1')
C.evaluate(dt=.1)
assert np_equal(C.weights, np.array([[0,1,0],
[1,0,1],
[0,1,0]])*1.1)
def test_3(self):
kernel = np.ones(3)
kernel[1] = np.NaN
C = DenseConnection(np.ones(3),
np.ones(3),
kernel,
equation='dW/dt = 1')
C.evaluate(dt=.1)
assert np_equal(C.weights,
np.array([[0, 1, 0], [1, 0, 1], [0, 1, 0]]) * 1.1)
def test_4(self):
src = np.ones((3, ))
dst = zeros((3, ), 'V=I; I')
kernel = np.ones(1)
C = DenseConnection(src, dst('I'), kernel, equation='dW/dt = post.I')
dst.run(dt=0.1)
assert np_equal(C.weights, np.identity(3) * 1.1)
def test_5(self):
n = 10
Z = np.random.random(n)
K = np.random.random(n)
Z1 = DenseConnection(Z,Z,K,toric=True).output()
Z2 = SharedConnection(Z,Z,K,toric=True,fft=False).output()
Z3 = SharedConnection(Z,Z,K,toric=True,fft=True).output()
Z4 = convolve(Z, K[::-1], mode='wrap')
assert np_equal(Z1,Z4)
assert np_equal(Z2,Z4)
assert np_equal(Z3,Z4)
def test_7(self):
n = 9
Z = np.random.random(n)
K = np.random.random(n//2)
Z1 = DenseConnection(Z,Z,K,toric=False).output()
Z2 = SharedConnection(Z,Z,K,toric=False,fft=False).output()
Z3 = SharedConnection(Z,Z,K,toric=False,fft=True).output()
Z4 = convolve(Z, K[::-1], mode='constant')
assert np_equal(Z1,Z4)
assert np_equal(Z2,Z4)
assert np_equal(Z3,Z4)
def test_12(self):
n = 10
Z = np.random.random((n,n))
K = np.random.random((2*n,2*n))
Z1 = DenseConnection(Z,Z,K,toric=False).output()
Z2 = SparseConnection(Z,Z,K,toric=False).output()
Z3 = SharedConnection(Z,Z,K,toric=False,fft=False).output()
Z4 = SharedConnection(Z,Z,K,toric=False,fft=True).output()
Z5 = convolve(Z, K[::-1,::-1], mode='constant')
assert np_equal(Z1,Z5)
assert np_equal(Z2,Z5)
assert np_equal(Z3,Z5)
assert np_equal(Z4,Z5)
def test_1(self):
n = 9
Z = np.random.random((n,n))
K = np.random.random((n//2,n//2))
Z1 = DenseConnection(Z,Z,K,toric=True).output()
Z2 = SparseConnection(Z,Z,K,toric=True).output()
Z3 = SharedConnection(Z,Z,K,toric=True,fft=False).output()
Z4 = SharedConnection(Z,Z,K,toric=True,fft=True).output()
Z5 = convolve(Z, K[::-1,::-1], mode='wrap')
assert np_equal(Z1,Z5)
assert np_equal(Z2,Z5)
assert np_equal(Z3,Z5)
assert np_equal(Z4,Z5)
def test_6(self):
n = 10
Z = np.random.random((n,n))
K = np.random.random((2*n,2*n))
Z1 = DenseConnection(Z,Z,K,toric=True).output()
Z2 = SparseConnection(Z,Z,K,toric=True).output()
Z3 = SharedConnection(Z,Z,K,toric=True,fft=False).output()
Z4 = SharedConnection(Z,Z,K,toric=True,fft=True).output()
K_ = K[5:15,5:15]
Z5 = convolve(Z, K_[::-1,::-1], mode='wrap')
assert np_equal(Z1,Z5)
assert np_equal(Z2,Z5)
assert np_equal(Z3,Z5)
assert np_equal(Z4,Z5)
def test_3(self):
net = Network(Clock(0.0, 1.0, 0.001))
src = Group((1, ), 'dV/dt=1')
src[...] = 1
dst = Group((1, ), 'I')
dst[...] = 0
kernel = np.ones(1)
C = DenseConnection(src('V'), dst('I'), kernel)
net.append(src)
net.append(dst)
V, I = [], []
@net.clock.every(0.1, order=-1)
def do(t):
V.append(src['V'][0])
I.append(dst['I'][0])
net.run(time=1.0, dt=0.1)
assert np_equal(
np.array(V),
[1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2.0])
assert np_equal(
np.array(I),
[0.0, 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9])
def test_2(self):
kernel = np.ones(3)*np.NaN
C = DenseConnection(np.ones(3), np.ones(3), kernel,
equation = 'dW/dt = 1')
C.evaluate(dt=.1)
assert np_equal(C.weights, np.zeros((3,3)))
def test_1(self):
kernel = np.ones(1)
C = DenseConnection(np.ones(3), np.ones(3), kernel,
equation = 'dW/dt = 1')
C.evaluate(dt=.1)
assert np_equal(C.weights, np.identity(3)*1.1)