def test_ifelse(self):
config1 = theano.config.profile
config2 = theano.config.profile_memory
try:
theano.config.profile = True
theano.config.profile_memory = True
a, b = T.scalars('a', 'b')
x, y = T.scalars('x', 'y')
z = ifelse(T.lt(a, b), x * 2, y * 2)
p = theano.ProfileStats(False)
if theano.config.mode in ["DebugMode", "DEBUG_MODE", "FAST_COMPILE"]:
m = "FAST_RUN"
else:
m = None
f_ifelse = theano.function([a, b, x, y], z, profile=p, name="test_ifelse",
mode=m)
val1 = 0.
val2 = 1.
big_mat1 = 10
big_mat2 = 11
out = f_ifelse(val1, val2, big_mat1, big_mat2)
finally:
theano.config.profile = config1
theano.config.profile_memory = config2
def test_pickle(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
f = function([x, In(a, value=1.0, name='a'), In(s, value=0.0, update=s+a*x, mutable=True)], s+a*x)
try:
# Note that here we also test protocol 0 on purpose, since it
# should work (even though one should not use it).
g = pickle.loads(pickle.dumps(f, protocol=0))
g = pickle.loads(pickle.dumps(f, protocol=-1))
except NotImplementedError as e:
if e[0].startswith('DebugMode is not picklable'):
return
else:
raise
# if they both return, assume that they return equivalent things.
# print [(k,id(k)) for k in f.finder.keys()]
# print [(k,id(k)) for k in g.finder.keys()]
self.assertFalse(g.container[0].storage is f.container[0].storage)
self.assertFalse(g.container[1].storage is f.container[1].storage)
self.assertFalse(g.container[2].storage is f.container[2].storage)
self.assertFalse(x in g.container)
self.assertFalse(x in g.value)
self.assertFalse(g.value[1] is f.value[1]) # should not have been copied
self.assertFalse(g.value[2] is f.value[2]) # should have been copied because it is mutable.
self.assertFalse((g.value[2] != f.value[2]).any()) # its contents should be identical
self.assertTrue(f(2, 1) == g(2)) # they should be in sync, default value should be copied.
self.assertTrue(f(2, 1) == g(2)) # they should be in sync, default value should be copied.
f(1, 2) # put them out of sync
self.assertFalse(f(1, 2) == g(1, 2)) # they should not be equal anymore.
def test_multiple_functions(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
v = T.vector('v')
# put in some inputs
list_of_things = [s, x, v]
# some derived thing, whose inputs aren't all in the list
list_of_things.append(a * x + s )
f1 = function([x, In(a, value=1.0, name='a'), In(s, value=0.0, update=s+a*x, mutable=True)], s+a*x)
list_of_things.append(f1)
# now put in a function sharing container with the previous one
f2 = function([x, In(a, value=1.0, name='a'), In(s, value=f1.container[s], update=s+a*x, mutable=True)], s+a*x)
list_of_things.append(f2)
assert isinstance(f2.container[s].storage, list)
assert f2.container[s].storage is f1.container[s].storage
# now put in a function with non-scalar
v_value = numpy.asarray([2, 3, 4.], dtype=config.floatX)
f3 = function([x, In(v, value=v_value)], x+v)
list_of_things.append(f3)
# try to pickle the entire things
try:
saved_format = cPickle.dumps(list_of_things, protocol=-1)
new_list_of_things = cPickle.loads(saved_format)
except NotImplementedError, e:
if e[0].startswith('DebugMode is not picklable'):
return
else:
raise
def __init__(self):
x, y, z = T.scalars('xyz')
e = T.nnet.sigmoid((x + y + z)**2)
op = th.OpFromGraph([x, y, z], [e])
e2 = op(x, y, z) + op(z, y, x)
self.inputs = [x, y, z]
self.outputs = [e2]
def test_callback_with_ifelse(self):
a, b, c = tensor.scalars("abc")
f = function(
[a, b, c], ifelse(a, 2 * b, 2 * c), mode=Mode(optimizer=None, linker=vm.VM_Linker(callback=self.callback))
)
f(1, 2, 3)
assert self.n_callbacks["IfElse"] == 2
def test_callback(self):
a, b, c = tensor.scalars("abc")
f = function([a, b, c], (a + b) + c, mode=Mode(optimizer=None, linker=vm.VM_Linker(callback=self.callback)))
f(1, 2, 3)
assert sum(self.n_callbacks.values()) == len(f.maker.fgraph.toposort())
f(1, 2, 3)
assert sum(self.n_callbacks.values()) == len(f.maker.fgraph.toposort()) * 2
def test_naming_rule1(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
f = function([a, s], a/s)
self.assertTrue(f(1, 2) == 0.5)
self.assertTrue(f(2, 1) == 2.0)
self.assertTrue(f(2, s=1) == 2.0)
checkfor(self, lambda: f(q=2, s=1), TypeError) # got unexpected keyword argument 'q'
checkfor(self, lambda: f(a=2, s=1), TypeError) # got unexpected keyword argument 'a'
def test_weird_names(self):
a, x, s = T.scalars('xxx')
checkfor(self, lambda: function([In(a, name=[])], []), TypeError)
def t():
f = function([In(a, name=set(['adsf', ()]), value=1.0),
In(x, name=(), value=2.0),
In(s, name=T.scalar(), value=3.0)], a+x+s)
checkfor(self, t, TypeError)
def __init__(self):
x, y, z = T.scalars('xyz')
e = x * y
op = th.OpFromGraph([x, y], [e])
e2 = op(x, y) + z
op2 = th.OpFromGraph([x, y, z], [e2])
e3 = op2(x, y, z) + z
self.inputs = [x, y, z]
self.outputs = [e3]
def test_naming_rule2(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x,s = T.scalars('xs')
#x's name is ignored because it is followed by anonymous parameter a.
f = function([x, a, s], a/s)
self.assertTrue(f(9,1,2) == 0.5)
self.assertTrue(f(9,2,1) == 2.0)
self.assertTrue(f(9,2, s=1) == 2.0)
checkfor(self, lambda:f(x=9,a=2,s=1), TypeError) #got unexpected keyword argument 'x'
checkfor(self, lambda:f(5.0,x=9), TypeError) #got unexpected keyword argument 'x'
def test_naming_rule0(self):
x, s = T.scalars('xs')
f = function([x, s], x/s)
self.assertTrue(f(1, 2) == 0.5)
self.assertTrue(f(2, 1) == 2.0)
self.assertTrue(f(s=2, x=1) == 0.5)
self.assertTrue(f(x=2, s=1) == 2.0)
self.assertTrue(f(2, s=1) == 2.0)
checkfor(self, lambda : f(2, x=2.0), TypeError) # got multiple values for keyword argument 'x'
checkfor(self, lambda : f(x=1), TypeError) # takes exactly 2 non-keyword arguments (1 given)
checkfor(self, lambda : f(s=1), TypeError) # takes exactly 2 non-keyword arguments (0 given)
def test_deepcopy(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x,s = T.scalars('xs')
f = function([x, In(a, value=1.0,name='a'), In(s, value=0.0, update=s+a*x, mutable=True)], s+a*x)
try:
g = copy.deepcopy(f)
except NotImplementedError, e:
if e[0].startswith('DebugMode is not picklable'):
return
else:
raise
def test_naming_rule3(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
# x's name is not ignored (as in test_naming_rule2) because a has a default value.
f = function([x, In(a, value=1.0), s], a/s+x)
self.assertTrue(f(9, 2, 4) == 9.5) # can specify all args in order
self.assertTrue(f(9, 2, s=4) == 9.5) # can give s as kwarg
self.assertTrue(f(9, s=4) == 9.25) # can give s as kwarg, get default a
self.assertTrue(f(x=9, s=4) == 9.25) # can give s as kwarg, omit a, x as kw
checkfor(self, lambda: f(x=9, a=2, s=4), TypeError) # got unexpected keyword argument 'a'
checkfor(self, lambda: f(), TypeError) # takes exactly 3 non-keyword arguments (0 given)
checkfor(self, lambda: f(x=9), TypeError) # takes exactly 3 non-keyword arguments (1 given)
def test_shared_state0(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
f = function([x, In(a, value=1.0, name='a'), In(s, value=0.0, update=s+a*x, mutable=True)], s+a*x)
g = function([x, In(a, value=1.0, name='a'), In(s, value=f.container[s], update=s-a*x, mutable=True)], s+a*x)
f(1, 2)
self.assertTrue(f[s] == 2)
self.assertTrue(g[s] == 2)
g(1, 2)
self.assertTrue(f[s] == 0)
self.assertTrue(g[s] == 0)
def __init__(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
v = T.vector('v')
self.s = s
self.x = x
self.v = v
self.e = a * x + s
self.f1 = function([x, In(a, value=1.0, name='a'), In(s, value=0.0, update=s+a*x, mutable=True)], s+a*x)
self.f2 = function([x, In(a, value=1.0, name='a'), In(s, value=self.f1.container[s], update=s+a*x, mutable=True)], s+a*x)
def test_naming_rule4(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
f = function([x, In(a, value=1.0, name='a'), s], a/s+x)
self.assertTrue(f(9, 2, 4) == 9.5) # can specify all args in order
self.assertTrue(f(9, 2, s=4) == 9.5) # can give s as kwarg
self.assertTrue(f(9, s=4) == 9.25) # can give s as kwarg, get default a
self.assertTrue(f(9, a=2, s=4) == 9.5) # can give s as kwarg, a as kwarg
self.assertTrue(f(x=9, a=2, s=4) == 9.5) # can give all kwargs
self.assertTrue(f(x=9, s=4) == 9.25) # can give all kwargs
checkfor(self, lambda: f(), TypeError) # takes exactly 3 non-keyword arguments (0 given)
checkfor(self, lambda: f(5.0, x=9), TypeError) # got multiple values for keyword argument 'x'
def test_pickle(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x,s = T.scalars('xs')
f = function([x, In(a, value=1.0,name='a'), In(s, value=0.0, update=s+a*x, mutable=True)], s+a*x)
try:
# Note that here we also test protocol 0 on purpose, since it
# should work (even though one should not use it).
g = cPickle.loads(cPickle.dumps(f, protocol=0))
g = cPickle.loads(cPickle.dumps(f, protocol=-1))
except NotImplementedError, e:
if e[0].startswith('DebugMode is not picklable'):
return
else:
raise
def test_c_thunks():
a = tensor.scalars('a')
b, c = tensor.vectors('bc')
cases = [False]
if theano.config.cxx:
cases.append(True)
for c_thunks in cases:
f = function([a, b, c], ifelse(a, a * b, b * c),
mode=Mode(
optimizer=None,
linker=vm.VM_Linker(c_thunks=c_thunks,
use_cloop=False)))
f(1, [2], [3, 2])
from nose.tools import assert_raises
assert_raises(ValueError, f, 0, [2], [3, 4])
assert any([hasattr(t, 'cthunk') for t in f.fn.thunks]) == c_thunks
def _policy_function(self):
epoch, gm, powr, step = T.scalars('epoch', 'gm', 'powr', 'step')
if self.lr_policy == 'inv':
decay = T.power(1.0+gm*epoch, -powr)
elif self.lr_policy == 'exp':
decay = gm ** epoch
elif self.lr_policy == 'step':
decay = T.switch(T.eq(T.mod_check(epoch, step), 0.0),
T.power(gm, T.floor_div(epoch, step)),
1.0)
elif self.lr_policy == 'fixed':
decay = T.constant(1.0, name='fixed', dtype=theano.config.floatX)
return theano.function([gm, epoch, powr, step],
decay,
allow_input_downcast=True,
on_unused_input='ignore')
def test_shared_state_not_implicit(self):
# This test is taken from the documentation in
# doc/topics/function.txt. If it does not pass anymore and yet the
# behavior is still intended the doc and the test should both be
# updated accordingly.
x, s = T.scalars('xs')
inc = function([x, In(s, update=(s+x), value=10.0)], [])
dec = function([x, In(s, update=(s-x), value=inc.container[s],
implicit=False)], [])
self.assertTrue(dec[s] is inc[s])
inc[s] = 2
self.assertTrue(dec[s] == 2)
dec(1)
self.assertTrue(inc[s] == 1)
dec(1, 0)
self.assertTrue(inc[s] == -1)
self.assertTrue(dec[s] == -1)
def policy_update(self, lr_policy):
epoch, gm, powr, step = T.scalars('epoch', 'gm', 'powr', 'step')
if lr_policy == 'inv':
decay = T.power(1+gm*epoch, -powr)
elif lr_policy == 'exp':
decay = gm ** epoch
elif lr_policy == 'step':
decay = T.switch(T.eq(T.mod_check(epoch, step), 0),
T.power(gm, T.floor_div(epoch, step)),
1.0)
elif lr_policy == 'fixed':
decay = T.constant(1.0, name='fixed', dtype='float32')
return theano.function([gm, epoch, powr, step],
decay,
updates=[(self.shared_lr,
self.shared_lr * decay)],
on_unused_input='ignore')
def test_deepcopy(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
f = function([x, In(a, value=1.0, name='a'),
In(s, value=0.0, update=s + a * x, mutable=True)],
s + a * x)
try:
g = copy.deepcopy(f)
except NotImplementedError as e:
if e[0].startswith('DebugMode is not picklable'):
return
else:
raise
# if they both return, assume that they return equivalent things.
# print [(k,id(k)) for k in f.finder.keys()]
# print [(k,id(k)) for k in g.finder.keys()]
self.assertFalse(g.container[0].storage is f.container[0].storage)
self.assertFalse(g.container[1].storage is f.container[1].storage)
self.assertFalse(g.container[2].storage is f.container[2].storage)
self.assertFalse(x in g.container)
self.assertFalse(x in g.value)
self.assertTrue(len(f.defaults) == len(g.defaults))
self.assertTrue(f._check_for_aliased_inputs is g._check_for_aliased_inputs)
self.assertTrue(f.name == g.name)
self.assertTrue(f.maker.fgraph.name == g.maker.fgraph.name)
# print 'f.defaults = %s' % (f.defaults, )
# print 'g.defaults = %s' % (g.defaults, )
self.assertTrue(all([f_req == g_req and f_feed == g_feed and
f_val == g_val
for ((f_req, f_feed, f_val), (g_req, g_feed, g_val)) in zip(
f.defaults, g.defaults)]))
self.assertFalse(g.value[1] is f.value[1]) # should not have been copied
self.assertFalse(g.value[2] is f.value[2]) # should have been copied because it is mutable.
self.assertFalse((g.value[2] != f.value[2]).any()) # its contents should be identical
self.assertTrue(f(2, 1) == g(2)) # they should be in sync, default value should be copied.
self.assertTrue(f(2, 1) == g(2)) # they should be in sync, default value should be copied.
f(1, 2) # put them out of sync
self.assertFalse(f(1, 2) == g(1, 2)) # they should not be equal anymore.
g(1, 2) # put them back in sync
self.assertTrue(f(3) == g(3)) # They should be in sync again.
def test_deepcopy_shared_container(self):
# Ensure that shared containers remain shared after a deep copy.
a, x = T.scalars('ax')
h = function([In(a, value = 0.0)], a)
f = function([x, In(a, value=h.container[a], implicit = True)], x + a)
try:
memo = {}
ac = copy.deepcopy(a)
memo.update({id(a): ac})
hc = copy.deepcopy(h, memo = memo)
memo.update({id(h): hc})
fc = copy.deepcopy(f, memo = memo)
except NotImplementedError, e:
if e[0].startswith('DebugMode is not picklable'):
return
else:
raise
def test_state_access(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
f = function([x, In(a, value=1.0, name='a'), In(s, value=0.0, update=s+a*x)], s+a*x)
self.assertTrue(f[a] == 1.0)
self.assertTrue(f[s] == 0.0)
self.assertTrue(f(3.0) == 3.0)
self.assertTrue(f(3.0, a=2.0) == 9.0) # 3.0 + 2*3.0
self.assertTrue(f[a] == 1.0) # state hasn't changed permanently, we just overrode it last line
self.assertTrue(f[s] == 9.0)
f[a] = 5.0
self.assertTrue(f[a] == 5.0)
self.assertTrue(f(3.0) == 24.0) # 9 + 3*5
self.assertTrue(f[s] == 24.0)
def test_4_conditionals():
a, b = T.scalars('a', 'b')
x, y = T.matrices('x', 'y')
f_switch = theano.function([a, b, x, y], T.switch(T.lt(a, b), T.mean(x), T.mean(y)))
f_lazy_ifelse = theano.function([a, b, x, y], ifelse(T.lt(a, b), T.mean(x), T.mean(y)))
x_val = np.ones((100, 100), dtype=theano.config.floatX)*1
y_val = np.ones((100, 100), dtype=theano.config.floatX)*2
# vectorized switch is going to evaluate both options
np.testing.assert_almost_equal(
f_switch(1, 2, x_val, y_val),
1
)
# lazy evaluation is going to evaluate only single option
np.testing.assert_almost_equal(
f_lazy_ifelse(2, 1, x_val, y_val),
2
)
def test_copy(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
f = function([x, In(a, value=1.0, name='a'), In(s, value=0.0, update=s+a*x, mutable=True)], s+a*x)
g = copy.copy(f)
# if they both return, assume that they return equivalent things.
self.assertFalse(g.container[x].storage is f.container[x].storage)
self.assertFalse(g.container[a].storage is f.container[a].storage)
self.assertFalse(g.container[s].storage is f.container[s].storage)
self.assertFalse(g.value[a] is not f.value[a]) # should not have been copied
self.assertFalse(g.value[s] is f.value[s]) # should have been copied because it is mutable.
self.assertFalse((g.value[s] != f.value[s]).any()) # its contents should be identical
self.assertTrue(f(2, 1) == g(2)) # they should be in sync, default value should be copied.
self.assertTrue(f(2, 1) == g(2)) # they should be in sync, default value should be copied.
f(1, 2) # put them out of sync
self.assertFalse(f(1, 2) == g(1, 2)) # they should not be equal anymore.
def learning_updates(model, details, inputs):
# create a list of all model parameters to be fit by gradient descent
params = []
for i in xrange(len(model.params_to_train)):
for j in model.params_to_train[i]:
params.append(model.params[i][j])
updates = []
# list of gradients for all model parameters
grads = T.grad(model.cost, params)
# theano variables for input (learning hyperparameters)
l_r, mom = T.scalars('l_r', 'mom')
#initialize parameter updates for momentum
param_updates = []
for i in xrange(len(params)):
param = params[i]
init = np.zeros(param.get_value(borrow=True).shape, dtype=theano.config.floatX)
param_updates.append(theano.shared(init))
# build updates pairs
for param_i, grad_i, prev in zip(params, grads, param_updates):
upd = mom * prev - l_r * grad_i
upd_param = param_i + upd
upd_param = regularize_weights(details['regularizer'], param_i, upd, upd_param)
updates.append((param_i, upd_param))
inputs += [l_r, mom]
return inputs, params, updates
def test_input_anon_singleton(self):
x, s = T.scalars('xs')
fn = function([s, x], [x + s])
self.assertTrue(fn(2, 3) == [5])
# no state
self.assertTrue(fn(2, 3) == [5])
def test_input_anon_unpack(self):
x, s = T.scalars('xs')
fn = function([s, x], x + s)
self.assertTrue(fn(2, 3) == 5)
def test_same_names(self):
a, x, s = T.scalars('xxx')
# implicit names would cause error. What do we do?
f = function([a, x, s], a + x + s)
self.assertTrue(f(1, 2, 3) == 6)
checkfor(self, lambda: f(1, 2, x=3), TypeError)
def test_extra_inputs(self):
x, s = T.scalars('xs')
fn = function([x], [x])
self.assertRaises(TypeError, fn, 1, 2)
def fn():
x, s = T.scalars('xs')
# Ignore unused input s, as it hides the other error
function([s], [x], on_unused_input='ignore')
def test_multiple_functions(self):
a = T.scalar() # the a is for 'anonymous' (un-named).
x, s = T.scalars('xs')
v = T.vector('v')
# put in some inputs
list_of_things = [s, x, v]
# some derived thing, whose inputs aren't all in the list
list_of_things.append(a * x + s)
f1 = function([x, In(a, value=1.0, name='a'),
In(s, value=0.0, update=s + a * x, mutable=True)],
s + a * x)
list_of_things.append(f1)
# now put in a function sharing container with the previous one
f2 = function([x, In(a, value=1.0, name='a'),
In(s, value=f1.container[s], update=s + a * x, mutable=True)],
s + a * x)
list_of_things.append(f2)
assert isinstance(f2.container[s].storage, list)
assert f2.container[s].storage is f1.container[s].storage
# now put in a function with non-scalar
v_value = numpy.asarray([2, 3, 4.], dtype=config.floatX)
f3 = function([x, In(v, value=v_value)], x + v)
list_of_things.append(f3)
# try to pickle the entire things
try:
saved_format = pickle.dumps(list_of_things, protocol=-1)
new_list_of_things = pickle.loads(saved_format)
except NotImplementedError as e:
if e[0].startswith('DebugMode is not picklable'):
return
else:
raise
# now test our recovered new_list_of_things
# it should be totally unrelated to the original
# it should be interdependent in the same way as the original
ol = list_of_things
nl = new_list_of_things
for i in range(4):
assert nl[i] != ol[i]
assert nl[i].type == ol[i].type
assert nl[i].type is not ol[i].type
# see if the implicit input got stored
assert ol[3].owner.inputs[1] is s
assert nl[3].owner.inputs[1] is not s
assert nl[3].owner.inputs[1].type == s.type
# moving on to the functions...
for i in range(4, 7):
assert nl[i] != ol[i]
# looking at function number 1, input 's'
assert nl[4][nl[0]] is not ol[4][ol[0]]
assert nl[4][nl[0]] == ol[4][ol[0]]
assert nl[4](3) == ol[4](3)
# looking at function number 2, input 's'
# make sure it's shared with the first function
assert ol[4].container[ol[0]].storage is ol[5].container[ol[0]].storage
assert nl[4].container[nl[0]].storage is nl[5].container[nl[0]].storage
assert nl[5](3) == ol[5](3)
assert nl[4].value[nl[0]] == 6
assert numpy.all(nl[6][nl[2]] == numpy.asarray([2, 3., 4]))
def fn():
x, s = T.scalars('xs')
function([In(x, update=s + x)], x)
def fn():
x, s = T.scalars('xs')
function([In(x, update=((s * s) + x))], x)
def fn():
x, s = T.scalars('xs')
function([s], [x])
def setUp(self):
self.x, self.y = tensor.scalars("x", "y")
self.z = self.x + self.y
self.w = 2 * self.z
def fn():
x,s = T.scalars('xs')
fn = function([], [x])
def setUp(self):
self.x, self.y = tensor.scalars('x', 'y')
self.z = self.x + self.y
self.w = 2 * self.z
def setup_method(self):
self.x, self.y = tensor.scalars("x", "y")
self.z = self.x + self.y
self.w = 2 * self.z
def fn():
x, s = T.scalars('xs')
fn = function([In(x, update=s+x)], x)
def fn():
x,s = T.scalars('xs')
fn = function([s], Out(x))
from __future__ import absolute_import, print_function, division
import time
import numpy
import theano
from theano import tensor as tt
from six.moves import xrange
from theano.ifelse import ifelse
a, b = tt.scalars('a', 'b')
x, y = tt.matrices('x', 'y')
z_switch = tt.switch(tt.lt(a, b), tt.mean(x), tt.mean(y))
z_lazy = ifelse(tt.lt(a, b), tt.mean(x), tt.mean(y))
f_switch = theano.function([a, b, x, y], z_switch)
f_lazyifelse = theano.function([a, b, x, y], z_lazy)
val1 = 0.
val2 = 1.
big_mat1 = numpy.ones((10000, 1000))
big_mat2 = numpy.ones((10000, 1000))
n_times = 10
tic = time.clock()
for i in xrange(n_times):
f_switch(val1, val2, big_mat1, big_mat2)
print('time spent evaluating both values %f sec' % (time.clock() - tic))
from theano import tensor as T
from theano.ifelse import ifelse
import theano, time, numpy
a, b = T.scalars('a', 'b')
x, y = T.matrices('x', 'y')
z_switch = T.switch(T.lt(a, b), T.mean(x), T.mean(y))
z_lazy = ifelse(T.lt(a, b), T.mean(x), T.mean(y))
f_switch = theano.function([a, b, x, y],
z_switch,
mode=theano.Mode(linker='vm'))
f_lazyifelse = theano.function([a, b, x, y],
z_lazy,
mode=theano.Mode(linker='vm'))
val1 = 0.
val2 = 1.
big_mat1 = numpy.ones((10000, 1000))
big_mat2 = numpy.ones((10000, 1000))
n_times = 10
tic = time.clock()
for i in xrange(n_times):
f_switch(val1, val2, big_mat1, big_mat2)
print 'time spent evaluating both values %f sec' % (time.clock() - tic)
tic = time.clock()
for i in xrange(n_times):
