Merge remote-tracking branch 'upstream/master' into issue-1080

NEWS.txt

@@ -43,13 +43,13 @@ Interface change:
  * Now we do not support unaligned ndarray in python code. (Frederic B.)
    We did not support it in c code and supporting it in python code made
    the detection harder.
- * Now we only support officialy scipy 0.7.2 and numpy 1.5.0 (Frederic B.)
-   We weren't and aren't testing with older version.
+ * Now we only officially support scipy 0.7.2 and numpy 1.5.0 (Frederic B.)
+   We weren't and aren't testing with older versions.
  * The theano.sparse.SparseType is available even when scipy is not (Frederic B.)
  * Fixes issue where members of consider_constant grad parameter
    were treated differently from Constant variables. (Ian G.)
- * Remove the parameter g_cost to theano.grad(). (Ian G.)
-   Use the new more powerfull parameter known_grads instead.
+ * Remove the parameter g_cost from theano.grad(). (Ian G.)
+   Use the new more powerful parameter known_grads instead.
 
 NumPy interface support:
  * theano.tensor.where is an alias for theano.tensor.switch to support NumPy semantic. (Ian G.)
@@ -75,23 +75,23 @@ New Feature:
  * Allow integer axes when keepdims==True (Jeremiah Lowin)
  * Add erfinv and erfcinv op. (Jey Kottalam)
  * Added tensor.batched_dot(). (Caglar Gulcehre)
-   It use scan behind the scene, but making doing this easier.
+   It uses scan behind the scenes, but makes doing this easier.
  * theano.get_constant_value(x) (Frederic B.)
-   This try to do have x as a constant int.
-   This do some constant folding to try to convert x into an int.
-   Used by some optimization.
+   This tries to have x as a constant int.
+   This does some constant folding to try to convert x into an int.
+   Used by some optimizations.
  * Add theano.tensor.io.{MPIRecv,MPIRecvWait,MPISend,MPISendWait} (Matthew Rocklin)
-   Theano do not automatically use them. It is up to you to use them and split your computation.
+   Theano does not automatically use them. It is up to you to use them and split your computation.
  * Added theano.sandbox.linalg.eig (abalkin)
  * Started some support for Python3 (abalkin)
-   setup.py support python3 now.
-   It call 2to3 during the setup.
-   Python3 not fully supported as we didn't update the c code.
+   setup.py supports python3 now.
+   It calls 2to3 during the setup.
+   Python3 is not fully supported as we didn't update the c code.
 
 
 Crash Fix:
  * Fix a crash related to scan.grad due to the new mechanism. (Ian G.)
- * Fix an optimization warning. Now it get optimized. (Frederic B.)
+ * Fix an optimization warning. Now it gets optimized. (Frederic B.)
  * Fix crash introduced in 0.6rc1 in theano.grad (Ian G.)
  * Fix crash introduced in 0.6rc1 in the grad of scan (Razvan P.)
  * Fix crash introduced in 0.6rc1 in the grad of clip (Ian G.)
@@ -125,7 +125,7 @@ Theano 0.6rc1 (October 1st, 2012)
 Highlights:
  * Bug fixes, crash fixes, CPU and GPU speed up.
  * theano_var.eval({other_var: val[,...]} to simplify the usage of Theano (Ian G.)
- * New default linker `cvm`. This is the execution engine that tells what op to run in which order.
+ * New default linker `cvm`. This is the execution engine that tells ops to run in certain orders.
    It is now implemented in C and enables lazy evaluation of ifelse op.
  * Faster theano.function compilation. (Pascal L., Ian G.)
  * Big sparse submodule update and documentation of it. (Nicolas Bouchard)
@@ -153,7 +153,7 @@ Bug fixes:
    with those defaults then stuck at old values if the config variables were
    changed during program execution. (David W-F)
  * Fixed many subtle bugs involving mutable default arguments which may have
-   led to unexpected behaviour, such as objects sharing instance variables
+   led to unexpected behavior, such as objects sharing instance variables
    they were not supposed to share. (David W-F)
  * Correctly record the GPU device number used when we let the driver select it.
    (Frederic B.)
@@ -408,7 +408,7 @@ Documentation:
    (Frederic B.)
  * New installation instructions for Windows using EPD (Pascal L.)
  * New installation on Windows by using a Linux VM from ContinuumIO (Frederic B.)
- * Revisions of Theano tutorial and addition of exercices to it. (Eric L.)
+ * Revisions of Theano tutorial and addition of exercises to it. (Eric L.)
  * New tutorial on Sparse variable. (Nicolas B., Sebastien Lemieux, Frederic Bastien
    http://www.deeplearning.net/software/theano/tutorial/sparse.html
  * Installation documentation for CentOS6 (Frederic B.)

doc/NEWS.txt

@@ -43,13 +43,13 @@ Interface change:
  * Now we do not support unaligned ndarray in python code. (Frederic B.)
    We did not support it in c code and supporting it in python code made
    the detection harder.
- * Now we only support officialy scipy 0.7.2 and numpy 1.5.0 (Frederic B.)
-   We weren't and aren't testing with older version.
+ * Now we only officially support scipy 0.7.2 and numpy 1.5.0 (Frederic B.)
+   We weren't and aren't testing with older versions.
  * The theano.sparse.SparseType is available even when scipy is not (Frederic B.)
  * Fixes issue where members of consider_constant grad parameter
    were treated differently from Constant variables. (Ian G.)
- * Remove the parameter g_cost to theano.grad(). (Ian G.)
-   Use the new more powerfull parameter known_grads instead.
+ * Remove the parameter g_cost from theano.grad(). (Ian G.)
+   Use the new more powerful parameter known_grads instead.
 
 NumPy interface support:
  * theano.tensor.where is an alias for theano.tensor.switch to support NumPy semantic. (Ian G.)
@@ -75,23 +75,23 @@ New Feature:
  * Allow integer axes when keepdims==True (Jeremiah Lowin)
  * Add erfinv and erfcinv op. (Jey Kottalam)
  * Added tensor.batched_dot(). (Caglar Gulcehre)
-   It use scan behind the scene, but making doing this easier.
+   It uses scan behind the scenes, but makes doing this easier.
  * theano.get_constant_value(x) (Frederic B.)
-   This try to do have x as a constant int.
-   This do some constant folding to try to convert x into an int.
-   Used by some optimization.
+   This tries to have x as a constant int.
+   This does some constant folding to try to convert x into an int.
+   Used by some optimizations.
  * Add theano.tensor.io.{MPIRecv,MPIRecvWait,MPISend,MPISendWait} (Matthew Rocklin)
-   Theano do not automatically use them. It is up to you to use them and split your computation.
+   Theano does not automatically use them. It is up to you to use them and split your computation.
  * Added theano.sandbox.linalg.eig (abalkin)
  * Started some support for Python3 (abalkin)
-   setup.py support python3 now.
-   It call 2to3 during the setup.
-   Python3 not fully supported as we didn't update the c code.
+   setup.py supports python3 now.
+   It calls 2to3 during the setup.
+   Python3 is not fully supported as we didn't update the c code.
 
 
 Crash Fix:
  * Fix a crash related to scan.grad due to the new mechanism. (Ian G.)
- * Fix an optimization warning. Now it get optimized. (Frederic B.)
+ * Fix an optimization warning. Now it gets optimized. (Frederic B.)
  * Fix crash introduced in 0.6rc1 in theano.grad (Ian G.)
  * Fix crash introduced in 0.6rc1 in the grad of scan (Razvan P.)
  * Fix crash introduced in 0.6rc1 in the grad of clip (Ian G.)
@@ -125,7 +125,7 @@ Theano 0.6rc1 (October 1st, 2012)
 Highlights:
  * Bug fixes, crash fixes, CPU and GPU speed up.
  * theano_var.eval({other_var: val[,...]} to simplify the usage of Theano (Ian G.)
- * New default linker `cvm`. This is the execution engine that tells what op to run in which order.
+ * New default linker `cvm`. This is the execution engine that tells ops to run in certain orders.
    It is now implemented in C and enables lazy evaluation of ifelse op.
  * Faster theano.function compilation. (Pascal L., Ian G.)
  * Big sparse submodule update and documentation of it. (Nicolas Bouchard)
@@ -153,7 +153,7 @@ Bug fixes:
    with those defaults then stuck at old values if the config variables were
    changed during program execution. (David W-F)
  * Fixed many subtle bugs involving mutable default arguments which may have
-   led to unexpected behaviour, such as objects sharing instance variables
+   led to unexpected behavior, such as objects sharing instance variables
    they were not supposed to share. (David W-F)
  * Correctly record the GPU device number used when we let the driver select it.
    (Frederic B.)
@@ -408,7 +408,7 @@ Documentation:
    (Frederic B.)
  * New installation instructions for Windows using EPD (Pascal L.)
  * New installation on Windows by using a Linux VM from ContinuumIO (Frederic B.)
- * Revisions of Theano tutorial and addition of exercices to it. (Eric L.)
+ * Revisions of Theano tutorial and addition of exercises to it. (Eric L.)
  * New tutorial on Sparse variable. (Nicolas B., Sebastien Lemieux, Frederic Bastien
    http://www.deeplearning.net/software/theano/tutorial/sparse.html
  * Installation documentation for CentOS6 (Frederic B.)

doc/install.txt

@@ -769,34 +769,10 @@ You can then proceed to the :ref:`windows_basic` or the :ref:`windows_bleeding_e
 Alternative: Anaconda 0.8.3 (Linux VM on Windows)
 #################################################
 
-ContinuumIO_ provides a free Windows VM with Theano install. The VM is the CentOS6.2 64 bit OS.
-
-- If you do not have VMWare installed, install VMWare player(free): http://www.vmware.com/products/player/
-- Download the VM: http://continuum.io/downloads.html
-- Follow the instruction on the ContinuumIO website to start the VM
-- Configure Theano by executing this:
-
-.. code-block:: bash
-
-    echo "[blas]" >> ~/.theanorc
-    echo "ldflags=" >> ~/.theanorc
-
-- [Optional] To enable the network, go into the VMWare setting for the vm and set the networking to NAT. Start the VM. In the VM, comment all lines in the file /etc/udev/rules.d/70-persistent-net.rules and restart the VM.
-- [Optional] to install easy_install:
-
-.. code-block:: bash
-
-    wget -c http://pypi.python.org/packages/source/s/setuptools/setuptools-0.6c11.tar.gz#md5=7df2a529a074f613b509fb44feefe74e
-    tar -zxf setuptools-0.6c11.tar.gz
-    cd setuptools-0.6c11
-    sudo /opt/anaconda/bin/python setup.py install
-
-- [Optional] To install pip
-
-.. code-block:: bash
-
-    #install setuptools
-    sudo /opt/anaconda/bin/easy_install pip
+ContinuumIO_ was providing a free VM with Theano installed. Now they
+provide a new installation system that install itself on Windows. We
+don't have the time now to update the docs, so we remove the old
+documentations that don't work.
 
 .. _ContinuumIO: http://continuum.io
 

theano/compile/debugmode.py

@@ -147,7 +147,7 @@ class BadThunkOutput(DebugModeError):
     val2 = None
     """The value computed by `thunk2`"""
 
-    def __init__(self, r, thunk1, val1, thunk2, val2):
+    def __init__(self, r, thunk1, val1, thunk2, val2, inputs_val=()):
         """Initialize members"""
         DebugModeError.__init__(self)  # to be compatible with python2.4
         self.r = r
@@ -155,6 +155,7 @@ class BadThunkOutput(DebugModeError):
         self.val1 = val1
         self.thunk2 = thunk2
         self.val2 = val2
+        self.inputs_val = inputs_val
 
     def offending_op(self):
         """Return the Op class whose c_code and perform
@@ -171,7 +172,11 @@ class BadThunkOutput(DebugModeError):
         print >> sio, "BadThunkOutput"
         print >> sio, "  variable    :", self.r
         print >> sio, "  Outputs Type:", self.r.type
-        print >> sio, "  Inputs Type :", [i.type for i in self.r.owner.inputs]
+        print >> sio, "  Inputs Type :", [i.type for i in self.r.owner.inputs],
+        print >> sio, "  Inputs Shape:", [getattr(val, 'shape', None)
+                                          for val in self.inputs_val]
+        print >> sio, "  Inputs Strides:", [getattr(val, 'strides', None)
+                                            for val in self.inputs_val]
         print >> sio, "  Apply   :", self.r.owner
         print >> sio, "  thunk1  :", self.thunk1
         print >> sio, "  thunk2  :", self.thunk2
@@ -1331,9 +1336,11 @@ def _check_preallocated_output(node, thunk, prealloc_modes, def_val,
             for r in node.outputs:
                 if not r.type.values_eq_approx(r_vals[r], storage_map[r][0]):
                     # TODO: indicate it is not a C/Py problem
+                    inputs_val = [storage_map[inp] for inp in r.owner.inputs]
                     raise BadThunkOutput(r,
                             thunk1='Reference value', val1=r_vals[r],
-                            thunk2=thunk_name, val2=storage_map[r][0])
+                            thunk2=thunk_name, val2=storage_map[r][0],
+                            inputs_val=inputs_val)
 
             # Clear storage_map
             for r in node.outputs:
@@ -1911,9 +1918,11 @@ class _Linker(gof.link.LocalLinker):
                                 if not r.type.values_eq_approx(r_vals[r], storage_map[r][0]):
                                     #import pdb; pdb.set_trace()
                                     #r.type.values_eq_approx(r_vals[r], storage_map[r][0])
+                                    inputs_val = [storage_map[inp] for inp in r.owner.inputs]
                                     raise BadThunkOutput(r,
                                             thunk1='perform', val1=r_vals[r],
-                                            thunk2='c_code', val2=storage_map[r][0])
+                                            thunk2='c_code', val2=storage_map[r][0],
+                                            inputs_val=inputs_val)
                             else:
                                 #print >> sys.stderr, i, "DEBUGMODE storing reference output %x" % id(storage_map[r][0])
                                 #retrieve each output from the storage_map

theano/tensor/basic.py

@@ -2308,6 +2308,8 @@ class SpecifyShape(Op):
 
     @note:     Maybe in the future we will never do the assert!
     @note:     We currently don't support specifying partial shape information.
+
+    @todo:     test this op with sparse and cuda ndarray. Do c code for them too.
     """
     view_map = {0: [0]}
 
@@ -2324,11 +2326,16 @@ class SpecifyShape(Op):
         if not isinstance(x, Variable):
             x = as_tensor_variable(x)
         shape = as_tensor_variable(shape)
+        assert shape.ndim == 1
+        assert "int" in shape.dtype
+        if isinstance(shape, TensorConstant):
+            assert shape.data.size == x.ndim
         return Apply(self, [x, shape], [x.type()])
 
     def perform(self, node, inp, out_):
         x, shape = inp
         out, = out_
+        assert x.ndim == shape.size
         assert numpy.all(x.shape == shape), ("got shape", x.shape,
                                            "expected", shape)
         out[0] = x
@@ -2368,6 +2375,47 @@ class SpecifyShape(Op):
             return [None]
         return self.make_node(eval_points[0], *inputs[1:]).outputs
 
+    def c_code(self, node, nodename, inp, out, sub):
+        if not isinstance(node.inputs[0], TensorVariable):
+            # The c code bellow support only Tensor.  super.c_code
+            # will raise an exception to tell that there isn't c code
+            # for the other cases.
+            return super(SpecifyShape, self).c_code(node, nodename,
+                                                    inp, out, sub)
+        iname, shape = inp
+        oname, = out
+        fail = sub['fail']
+
+        return """
+        if (PyArray_NDIM(%(iname)s) != PyArray_DIMS(%(shape)s)[0]) {
+            PyErr_Format(PyExc_AssertionError,
+                         "SpecifyShape: vector of shape have %%d element,"
+                         " but the input have %%d dimensions.",
+                         PyArray_NDIM(%(iname)s),
+                         PyArray_DIMS(%(shape)s)[0]);
+            %(fail)s;
+        }
+        for(int i = 0; i < PyArray_NDIM(%(iname)s); i++){
+            dtype_%(shape)s shp = ((dtype_%(shape)s*)PyArray_GETPTR1(%(shape)s,
+                                                                     i))[0];
+            if (PyArray_DIMS(%(iname)s)[i] != shp) {
+                PyErr_Format(PyExc_AssertionError,
+                             "SpecifyShape: dim %%d of input have shape %%d,"
+                             " expected %%d.",
+                             i, PyArray_DIMS(%(iname)s)[i],
+                             shp);
+                %(fail)s;
+            }
+        }
+        Py_XDECREF(%(oname)s);
+        %(oname)s = %(iname)s;
+        Py_XINCREF(%(oname)s);
+        """ % locals()
+
+    def c_code_cache_version(self):
+        return (1,)
+
+
 specify_shape = SpecifyShape()
 
 

theano/tensor/elemwise.py

@@ -340,6 +340,7 @@ class DimShuffle(Op):
                             #borrow only the writable flag from the base
                             # the NPY_OWNDATA flag will default to 0.
                             '(NPY_ARRAY_WRITEABLE*PyArray_ISWRITEABLE(%(basename)s)), NULL)'),
+                'if (%(res)s == NULL) %(fail)s;',
                 #recalculate flags: CONTIGUOUS, FORTRAN, ALIGNED
                 'PyArray_UpdateFlags(%(res)s, NPY_ARRAY_UPDATE_ALL)',
                 #we are making a view in both inplace and non-inplace cases

theano/tensor/tests/test_basic.py

@@ -6326,6 +6326,54 @@ def test_transpose():
     assert tensor.transpose(tensor.dmatrix()).name is None
 
 
+class TestSpecifyShape(unittest.TestCase):
+    def shortDescription(self):
+        return None
+
+    def test_bad_shape(self):
+        """ Test that at run time we raise an exception when the shape
+        is not the one specified"""
+        specify_shape = SpecifyShape()
+
+        x = vector()
+        xval = numpy.random.rand(2).astype(floatX)
+        f = theano.function([x], specify_shape(x, [2]))
+        f(xval)
+        xval = numpy.random.rand(3).astype(floatX)
+        self.assertRaises(AssertionError, f, xval)
+
+        x = matrix()
+        xval = numpy.random.rand(2, 3).astype(floatX)
+        f = theano.function([x], specify_shape(x, [2, 3]))
+        f(xval)
+        for shape in [(1, 3), (2, 2), (5, 5)]:
+            xval = numpy.random.rand(*shape).astype(floatX)
+            self.assertRaises(AssertionError, f, xval)
+
+    def test_bad_number_of_shape(self):
+        """ Test that the number of dimensions provided is good"""
+        specify_shape = SpecifyShape()
+
+        x = vector()
+        shape_vec = ivector()
+        xval = numpy.random.rand(2).astype(floatX)
+        self.assertRaises(AssertionError, specify_shape, x, [])
+        self.assertRaises(AssertionError, specify_shape, x, [2, 2])
+
+        f = theano.function([x, shape_vec], specify_shape(x, shape_vec))
+        self.assertRaises(AssertionError, f, xval, [])
+        self.assertRaises(AssertionError, f, xval, [2, 2])
+
+        x = matrix()
+        xval = numpy.random.rand(2, 3).astype(floatX)
+        for shape in [(),
+                      (1,),
+                      (2, 3, 4)]:
+            self.assertRaises(AssertionError, specify_shape, x, shape)
+            f = theano.function([x, shape_vec], specify_shape(x, shape_vec))
+            self.assertRaises(AssertionError, f, xval, shape)
+
+
 class TestInferShape(utt.InferShapeTester):
 
     def test_infer_shape(self):

theano/tensor/tests/test_elemwise.py

@@ -95,6 +95,10 @@ class test_DimShuffle(unittest_tools.InferShapeTester):
                             [DimShuffle(ib, shuffle)(adtens)],
                             [adtens_val], DimShuffle)
 
+    def test_too_big_rank(self):
+        x = tensor.dscalar()
+        y = x.dimshuffle(('x',) * (numpy.MAXDIMS + 1))
+        self.assertRaises(ValueError, y.eval, {x: 0})
                           
 class test_Broadcast(unittest.TestCase):
     def setUp(self):