merge

theano/compile/function_module.py

@@ -16,6 +16,9 @@ from mode import *
 from io import *
 
 
+# used by function and module as the default compilation mode
+mode_default = 'FAST_COMPILE'
+
 def infer_reuse_pattern(env, outputs_to_disown):
     """
     Given an env and a list of results, returns the list of all
@@ -451,7 +454,7 @@ class FunctionMaker(object):
             raise TypeError("Unknown output type: %s (%s)", type(output), output)
 
     def __init__(self, inputs, outputs, 
-            mode = 'FAST_COMPILE', accept_inplace = False, function_builder = Function):
+            mode = mode_default, accept_inplace = False, function_builder = Function):
         """
         :type inputs: a list of SymbolicInput instances
 
@@ -641,7 +644,7 @@ def register_checker(checker):
 
 
 
-def function(inputs, outputs, mode='FAST_RUN', accept_inplace = False):
+def function(inputs, outputs, mode=mode_default, accept_inplace = False):
     """
     Return a function calculating the outputs from the inputs.
 

theano/compile/module.py

@@ -125,7 +125,7 @@ class Component(object):
         """
         raise NotImplementedError
 
-    def make_no_init(self, mode='FAST_COMPILE'):
+    def make_no_init(self, mode=F.mode_default):
         """
         Allocates the necessary containers using allocate() and uses
         build() with the provided mode to make an instance which will
@@ -145,7 +145,7 @@ class Component(object):
         arguments and the keyword arguments. If 'mode' is in the
         keyword arguments it will be passed to build().
         """
-        mode = kwargs.pop('mode', 'FAST_COMPILE')
+        mode = kwargs.pop('mode', F.mode_default)
         rval = self.make_no_init(mode)
         if hasattr(rval, 'initialize'):
             rval.initialize(*args, **kwargs)
@@ -956,7 +956,7 @@ class Module(ComponentDict):
         """
         self.make_mi(args,kwargs)
 
-        mode = kwargs.pop('mode', 'FAST_COMPILE')
+        mode = kwargs.pop('mode', F.mode_default)
         rval = self.make_no_init(mode)
         if hasattr(rval, 'initialize'):
             rval.initialize(*args, **kwargs)

theano/printing.py

@@ -236,6 +236,7 @@ class PPrinter:
         return "\n".join(s[1] for s in strings)
 
     def __call__(self, *args):
+        print sys.stderr, "WARNING: PPrinter bug. Is theano ticket #249 fixed yet?"
         if len(args) == 1:
             return self.process(*args)
         elif len(args) == 2 and isinstance(args[1], (PrinterState, dict)):

theano/sandbox/debugmode.py

@@ -11,6 +11,7 @@ broken.
 
 """
 import time, copy, sys
+from StringIO import StringIO
 
 from .. import gof
 
@@ -21,6 +22,7 @@ from ..gof.cc import OpWiseCLinker, CLinker
 from ..compile.mode import Mode
 import numpy
 
+
 from ..compile.function_module import (convert_function_input, 
         FunctionMaker,
         predefined_modes,
@@ -31,6 +33,24 @@ from ..compile.function_module import (convert_function_input,
         SymbolicOutput,
         Supervisor)
 
+def debugprint(a, prefix='', depth=-1, done=None, file=sys.stdout):
+    if depth==0:
+        return
+    done = set() if done is None else done
+    if hasattr(a, 'op'):
+        print >> file, prefix, a.op, id(a)
+        if id(a) not in done:
+            done.add(id(a))
+            for i in a.inputs:
+                if i.owner:
+                    debugprint(i.owner, prefix+'  ', depth=depth-1, done=done, file=file)
+                else:
+                    print >> file, prefix+'  ', i, id(i)
+    else:
+        print >> file, prefix+'  ', a, id(a)
+
+    return file
+
 class ResultEquivalenceTracker(object):
     def __init__(self):
         self.env = None
@@ -43,6 +63,7 @@ class ResultEquivalenceTracker(object):
         self.env = env
         self.all_results_ever = []
         self.reasons = {}
+        self.snapshots = {}
 
     def on_detach(self, env):
         assert env is self.env
@@ -70,15 +91,22 @@ class ResultEquivalenceTracker(object):
                 self.equiv[r] = set([r])
                 self.all_results_ever.append(r)
                 self.reasons.setdefault(r, [])
+                self.snapshots.setdefault(r, [])
             for r in node.inputs:
                 self.reasons.setdefault(r, [])
+                self.snapshots.setdefault(r, [])
 
     def on_change_input(self, env, node, i, r, new_r, reason=None):
         #print 'CHANGE by', reason, 'to use', new_r, type(new_r)
 
         self.reasons.setdefault(new_r, [])
+        self.snapshots.setdefault(new_r, [])
         if (reason, r) not in self.reasons[new_r]:
             self.reasons[new_r].append((reason, r))
+            self.snapshots[new_r].append((
+                reason, 
+                debugprint(r.owner, prefix='  ', depth=6, file=StringIO()).getvalue(),
+                debugprint(new_r.owner,prefix='  ',  depth=6, file=StringIO()).getvalue()))
             self.reasons[r].append(('replaced by', new_r))
 
         if r in self.equiv:
@@ -252,6 +280,12 @@ class OptCheckLinker(OpWiseCLinker):
                         print "  Value Type:", type(r_vals[r])
                         print "  Value: ", r_vals[r]
                         print "  Reason: ", [(str(reason), id(old_r)) for reason, old_r in env.equivalence_tracker.reasons[r]]
+                        print "  Snapshots:"
+                        for s in env.equivalence_tracker.snapshots[r]:
+                            print "  BEFORE"
+                            print s[1]
+                            print "  AFTER"
+                            print s[2]
                         print ""
 
                     raise Exception("OptCheckFailure")

theano/tensor/opt.py

@@ -801,6 +801,7 @@ def local_greedy_distributor(node):
     increase numerical stability, e.g. when x and/or y tend to 0 in
     example 1.
     """
+
     out = node.outputs[0]
     num, denum = local_mul_canonizer.get_num_denum(out)
     if len(num) == 1 and not denum:
@@ -816,7 +817,6 @@ def local_greedy_distributor(node):
         num.remove(candidate)
         _change, candidate, num, denum = attempt_distribution(candidate, num, denum)
         change |= _change
-        if change:
         new_num.append(candidate)
 
     for candidate in list(denum):
@@ -825,7 +825,6 @@ def local_greedy_distributor(node):
         denum.remove(candidate)
         _change, candidate, denum, num = attempt_distribution(candidate, denum, num)
         change |= _change
-        if change:
         new_denum.append(candidate)
 
     if not change:

theano/tensor/tests/test_opt.py

@@ -13,6 +13,9 @@ from theano import pprint
 import numpy
 #import scalar_opt
 
+from theano.sandbox.debugmode import OptCheck
+from theano import function
+
 
 def inputs(xbc = (0, 0), ybc = (0, 0), zbc = (0, 0)):
     x = Tensor(broadcastable = xbc, dtype = 'float64')('x')
@@ -120,6 +123,25 @@ class test_greedy_distribute(unittest.TestCase):
         gof.TopoOptimizer(gof.LocalOptGroup(local_greedy_distributor), order = 'out_to_in').optimize(g)
         ##print pprint(g.outputs[0])
     
+    def test_kording_bug(self):
+        x, y = vectors('xy')
+        eps = scalar('eps')
+        s = scalar('s')
+
+        #r = theano.tensor.mul(theano.tensor.fill(x, 2.*a), x/a , (y+z) , a)
+        #r = theano.tensor.mul((x/a+y) , a, z)
+        r = mul(
+                s - 1
+                , eps + x/s
+                , eps + y/s
+                , s)
+
+        f = function([s, eps, x,y], r**2, mode=OptCheck())
+
+        r0 = f(4,1.e-6, [1.5,2], [2.3,3.1])
+        r1 = f(4,1.e-6, [1.5,2], [2.3,3.1])
+        r2 = f(4,1.e-6, [1.5,2], [2.3,3.1])
+
 
 
 class test_canonize(unittest.TestCase):