def test_linear_cg():
rng = numpy.random.RandomState([1,2,3])
n = 5
M = rng.randn(2*n,n)
M = numpy.dot(M.T,M).astype(config.floatX)
b = rng.randn(n).astype(config.floatX)
c = rng.randn(1).astype(config.floatX)[0]
x = theano.tensor.vector('x')
f = 0.5 * tensor.dot(x,tensor.dot(M,x)) - tensor.dot(b,x) + c
sol = linear_cg.linear_cg(f,[x])
fn_sol = theano.function([x], sol)
start = time.time()
sol = fn_sol( rng.randn(n).astype(config.floatX))[0]
my_lcg = time.time() -start
eval_f = theano.function([x],f)
cgf = eval_f(sol)
print "conjugate gradient's value of f:", str(cgf), 'time (s)', my_lcg
skip_if_no_scipy()
spf = eval_f( scipy.linalg.solve(M,b) )
print "scipy.linalg.solve's value of f: "+str(spf)
abs_diff = abs(cgf - spf)
if not (abs_diff < 1e-5):
raise AssertionError("Expected abs_diff < 1e-5, got abs_diff of " +
str(abs_diff))
def test_linear_cg():
rng = numpy.random.RandomState([1, 2, 3])
n = 5
M = rng.randn(2 * n, n)
M = numpy.dot(M.T, M)
b = rng.randn(n)
c = rng.randn()
x = theano.shared(rng.randn(n))
f = 0.5 * tensor.dot(x, tensor.dot(M, x)) - tensor.dot(b, x) + c
sol = linear_cg.linear_cg(f, [x])
fn_sol = theano.function([], sol)
start = time.time()
sol = fn_sol()[0]
my_lcg = time.time() - start
eval_f = theano.function([], f)
x.set_value(sol)
cgf = eval_f()
print "conjugate gradient's value of f:", str(cgf), 'time (s)', my_lcg
x.set_value(scipy.linalg.solve(M, b), borrow=True)
spf = eval_f()
print "scipy.linalg.solve's value of f: " + str(spf)
assert abs(cgf - spf) < 1e-5
def test_linear_cg():
rng = numpy.random.RandomState([1,2,3])
n = 5
M = rng.randn(2*n,n)
M = numpy.dot(M.T,M)
b = rng.randn(n)
c = rng.randn()
x = theano.shared(rng.randn(n))
f = 0.5 * tensor.dot(x,tensor.dot(M,x)) - tensor.dot(b,x) + c
sol = linear_cg.linear_cg(f,[x])
fn_sol = theano.function([], sol)
start = time.time()
sol = fn_sol()[0]
my_lcg = time.time() -start
eval_f = theano.function([],f)
x.set_value(sol)
cgf = eval_f()
print "conjugate gradient's value of f:", str(cgf), 'time (s)', my_lcg
x.set_value( scipy.linalg.solve(M,b) , borrow = True )
spf = eval_f()
print "scipy.linalg.solve's value of f: "+str(spf)
assert abs(cgf - spf) < 1e-5
def test_linear_cg():
rng = numpy.random.RandomState([1, 2, 3])
n = 5
M = rng.randn(2 * n, n)
M = numpy.dot(M.T, M).astype(config.floatX)
b = rng.randn(n).astype(config.floatX)
c = rng.randn(1).astype(config.floatX)[0]
x = theano.tensor.vector("x")
f = 0.5 * tensor.dot(x, tensor.dot(M, x)) - tensor.dot(b, x) + c
sol = linear_cg.linear_cg(f, [x])
fn_sol = theano.function([x], sol)
start = time.time()
sol = fn_sol(rng.randn(n).astype(config.floatX))[0]
my_lcg = time.time() - start
eval_f = theano.function([x], f)
cgf = eval_f(sol)
print "conjugate gradient's value of f:", str(cgf), "time (s)", my_lcg
spf = eval_f(scipy.linalg.solve(M, b))
print "scipy.linalg.solve's value of f: " + str(spf)
assert abs(cgf - spf) < 1e-5
