Skip to content

P
pytensor
提交 ec1caa4d authored 作者: Pascal Lamblin's avatar Pascal Lamblin
浏览文件
操作

More pep8

上级 043c3eef
隐藏空白字符变更
内嵌 并排
正在显示 包含 319 行增加 和 330 行删除
theano/tensor/tests/test_opt.py
浏览文件 @ ec1caa4d
...@@ -964,144 +964,133 @@ class test_fusion(unittest.TestCase): ...@@ -964,144 +964,133 @@ class test_fusion(unittest.TestCase):
fwv + fxv + fyv + fzv, 'float32'), fwv + fxv + fyv + fzv, 'float32'),
(fw + (fx + (fy + fz)), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), 1, (fw + (fx + (fy + fz)), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), 1,
fwv + fxv + fyv + fzv, 'float32'), fwv + fxv + fyv + fzv, 'float32'),
((fw+fx)+(fy+fz), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), 1, ((fw + fx) + (fy + fz), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), 1,
fwv+fxv+fyv+fzv, 'float32'), # 10 fwv + fxv + fyv + fzv, 'float32'), # 10
(fw*fx*fy*fz, (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), 1, (fw * fx * fy * fz, (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), 1,
fwv * fxv * fyv * fzv, 'float32'), fwv * fxv * fyv * fzv, 'float32'),
(fw+fx*fy*fz, (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), 1, (fw + fx * fy * fz, (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), 1,
fwv + fxv * fyv * fzv, 'float32'), fwv + fxv * fyv * fzv, 'float32'),
(fx+fy*fz*fx, (fx, fy, fz), (fxv, fyv, fzv), 1, (fx + fy * fz * fx, (fx, fy, fz), (fxv, fyv, fzv), 1,
fxv + fyv * fzv * fxv, 'float32'), fxv + fyv * fzv * fxv, 'float32'),
(fx*fy+fz+fy, (fx, fy, fz), (fxv, fyv, fzv), 1, (fx * fy + fz + fy, (fx, fy, fz), (fxv, fyv, fzv), 1,
fxv * fyv + fzv + fyv, 'float32'), fxv * fyv + fzv + fyv, 'float32'),
(fx*fy*fz*fw+fx+fy+fz+fw, (fw, fx, fy, fz), (fwv, fxv, (fx * fy * fz * fw + fx + fy + fz + fw, (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), 1,
fyv, fzv), 1, fxv*fyv*fzv*fwv+fxv+fyv+fzv+fwv, 'float32'), # 15 fxv * fyv * fzv * fwv + fxv + fyv + fzv + fwv, 'float32'), # 15
# test with constant # test with constant
((fw+fx)+(fy+fz) + 2., (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), ((fw + fx) + (fy + fz) + 2., (fw, fx, fy, fz), (fwv, fxv, fyv, fzv),
1, fwv+fxv+fyv+fzv+2, 'float32'), 1, fwv + fxv + fyv + fzv + 2, 'float32'),
(((fw+fx)+2.+fy)+fz, (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), (((fw + fx) + 2. + fy) + fz, (fw, fx, fy, fz), (fwv, fxv, fyv, fzv),
1, fwv+fxv+fyv+fzv+2, 'float32'), 1, fwv + fxv + fyv + fzv + 2, 'float32'),
((fw+(fx+2.+fy))+fz, (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), ((fw + (fx + 2. + fy)) + fz, (fw, fx, fy, fz), (fwv, fxv, fyv, fzv),
1, fwv+fxv+fyv+fzv+2, 'float32'), 1, fwv + fxv + fyv + fzv + 2, 'float32'),
((fw+(fx+fy)+2+fz), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), ((fw + (fx + fy) + 2 + fz), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv),
1, fwv+fxv+fyv+fzv+2, 'float32'), 1, fwv + fxv + fyv + fzv + 2, 'float32'),
(fw+(fx+(fy+fz)+2.), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), (fw + (fx + (fy + fz) + 2.), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv),
1, fwv+fxv+fyv+fzv+2, 'float32'), # 20 1, fwv + fxv + fyv + fzv + 2, 'float32'), # 20
(2+(fw+fx)+(fy+fz), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv), (2 + (fw + fx) + (fy + fz), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv),
1, fwv+fxv+fyv+fzv+2, 'float32'), 1, fwv + fxv + fyv + fzv + 2, 'float32'),
# mix float32 and float64 # mix float32 and float64
(2+(dw+fx)+(fy+fz), (dw, fx, fy, fz), (dwv, fxv, fyv, fzv), (2 + (dw + fx) + (fy + fz), (dw, fx, fy, fz), (dwv, fxv, fyv, fzv),
1, dwv+fxv+fyv+fzv+2, 'float64'), 1, dwv + fxv + fyv + fzv + 2, 'float64'),
(2+(fw+dw)+(fy+fz), (fw, dw, fy, fz), (fwv, dwv, fyv, fzv), (2 + (fw + dw) + (fy + fz), (fw, dw, fy, fz), (fwv, dwv, fyv, fzv),
1, fwv+dwv+fyv+fzv+2, 'float64'), 1, fwv + dwv + fyv + fzv + 2, 'float64'),
(2+(fw+fx)+(dw+fz), (fw, fx, dw, fz), (fwv, fxv, dwv, fzv), (2 + (fw + fx) + (dw + fz), (fw, fx, dw, fz), (fwv, fxv, dwv, fzv),
1, fwv+fxv+dwv+fzv+2, 'float64'), 1, fwv + fxv + dwv + fzv + 2, 'float64'),
(2+(fw+fx)+(fy+dw), (fw, fx, fy, dw), (fwv, fxv, fyv, dwv), (2 + (fw + fx) + (fy + dw), (fw, fx, fy, dw), (fwv, fxv, fyv, dwv),
1, fwv+fxv+fyv+dwv+2, 'float64'), # 25 1, fwv + fxv + fyv + dwv + 2, 'float64'), # 25
# test when their is other op then elemwise. # test when their is other op then elemwise.
# the good output for the next test. ((fwx.sum()) + (fwx) + (fy + fz), (fw, fx, fy, fz), (fwv, fxv, fyv, fzv),
# (Pdb) p f.maker.fgraph.toposort() 4, (fwv + fxv).sum() + fwv + fxv + fyv + fzv, 'float32'),
#[Elemwise{add,no_inplace}(w, x), Sum(Elemwise{add,no_inplace}.0), InplaceDimShuffle{x,x}(Sum.0), Elemwise{Composite{_impls=[<function <lambda> at 0x2c5c8c0>], nin=4, _c_code={
# npy_float32 V%(id)s_tmp1;
# V%(id)s_tmp1 = %(i2)s + %(i3)s;
# npy_float32 V%(id)s_tmp2;
# V%(id)s_tmp2 = %(i0)s + %(i1)s;
#%(o0)s = V%(id)s_tmp2 + V%(id)s_tmp1;
#}
#, nout=1, fgraph=[add(add(<float32>, <float32>), add(<float32>, <float32>))]}}(InplaceDimShuffle{x,x}.0, Elemwise{add,no_inplace}.0, y, z)]
((fwx.sum())+(fwx)+(fy+fz), (fw, fx, fy, fz), (fwv, fxv,
fyv, fzv), 4, (fwv+fxv).sum()+fwv+fxv+fyv+fzv, 'float32'),
# test other elemwise op # test other elemwise op
(fx+fy+tensor.cos(fz), (fx, fy, fz), (fxv, fyv, fzv), 1, (fx + fy + tensor.cos(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fxv+fyv+numpy.cos(fzv), 'float32'), fxv + fyv + numpy.cos(fzv), 'float32'),
(fx+fy+tensor.cosh(fz), (fx, fy, fz), (fxv, fyv, fzv), 1, (fx + fy + tensor.cosh(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fxv+fyv+numpy.cosh(fzv), 'float32'), fxv + fyv + numpy.cosh(fzv), 'float32'),
(fx+fy+abs(fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv+fyv + (fx + fy + abs(fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv + fyv +
numpy.absolute(fzv), 'float32'), numpy.absolute(fzv), 'float32'),
(ix+iy+abs(iz), (ix, iy, iz), (ixv, iyv, izv), 1, ixv+iyv + (ix + iy + abs(iz), (ix, iy, iz), (ixv, iyv, izv), 1, ixv + iyv +
numpy.absolute(izv), 'int32'), # 30 numpy.absolute(izv), 'int32'), # 30
(fx+fy+theano.tensor.log(fz), (fx, fy, fz), (fxv, fyv, (fx + fy + theano.tensor.log(fz), (fx, fy, fz), (fxv, fyv, fzv),
fzv), 1, fxv+fyv+numpy.log(fzv), 'float32'), 1, fxv + fyv + numpy.log(fzv), 'float32'),
(fx+fy+theano.tensor.log2(fz), (fx, fy, fz), (fxv, fyv, (fx + fy + theano.tensor.log2(fz), (fx, fy, fz), (fxv, fyv, fzv),
fzv), 1, fxv+fyv+numpy.log2(fzv), 'float32'), 1, fxv + fyv + numpy.log2(fzv), 'float32'),
(fx+fy+theano.tensor.log10(fz), (fx, fy, fz), (fxv, fyv, (fx + fy + theano.tensor.log10(fz), (fx, fy, fz), (fxv, fyv, fzv),
fzv), 1, fxv+fyv+numpy.log10(fzv), 'float32'), 1, fxv + fyv + numpy.log10(fzv), 'float32'),
(fx+fy**fz, (fx, fy, fz), (fxv, fyv, fzv), 1, fxv+fyv**fzv, (fx + fy ** fz, (fx, fy, fz), (fxv, fyv, fzv), 1, fxv + fyv ** fzv,
'float32'), # pow 'float32'), # pow
(fx+fy+theano.tensor.exp(fz), (fx, fy, fz), (fxv, fyv, (fx + fy + theano.tensor.exp(fz), (fx, fy, fz), (fxv, fyv, fzv),
fzv), 1, fxv+fyv+numpy.exp(fzv), 'float32'), # 35 1, fxv + fyv + numpy.exp(fzv), 'float32'), # 35
(fx-fy-fz, (fx, fy, fz), (fxv, fyv, fzv), 1, fxv-fyv-fzv, 'float32'), (fx - fy - fz, (fx, fy, fz), (fxv, fyv, fzv), 1, fxv - fyv - fzv, 'float32'),
(fx-(fy/fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv-(fyv/fzv), 'float32'), (fx - (fy / fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv - (fyv / fzv), 'float32'),
(fx-theano.tensor.true_div(fy, 2), (fx, fy), (fxv, fyv), (fx - theano.tensor.true_div(fy, 2), (fx, fy), (fxv, fyv),
1, fxv-(fyv/2), 'float32'), 1, fxv - (fyv / 2), 'float32'),
(fx-theano.tensor.true_div(fy, fz), (fx, fy, fz), (fxv, (fx - theano.tensor.true_div(fy, fz), (fx, fy, fz), (fxv, fyv, fzv),
fyv, fzv), 1, fxv-(fyv/fzv), 'float32'), 1, fxv - (fyv / fzv), 'float32'),
(fx-theano.tensor.int_div(ix*100, iy*1000), (fx, ix, (fx - theano.tensor.int_div(ix * 100, iy * 1000), (fx, ix, iy), (fxv, ixv, iyv),
iy), (fxv, ixv, iyv), 1, fxv-((ixv*100)//(iyv*1000)), {'custom': 'float64', 'numpy+floatX': config.floatX, 'numpy': 'float64'}), # 40 1, fxv - ((ixv * 100) // (iyv * 1000)), {'custom': 'float64', 'numpy + floatX': config.floatX, 'numpy': 'float64'}), # 40
(fx-(fy/2), (fx, fy), (fxv, fyv), 1, fxv-(fyv/2), 'float32'), (fx - (fy / 2), (fx, fy), (fxv, fyv), 1, fxv - (fyv / 2), 'float32'),
(fx-(fy%fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv-(fyv%fzv), 'float32'), (fx - (fy % fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv - (fyv % fzv), 'float32'),
(fx-(fy > fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv-(fyv > fzv), 'float32'), (fx - (fy > fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv - (fyv > fzv), 'float32'),
(fx-(fy >= fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv-(fyv >= fzv), 'float32'), (fx - (fy >= fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv - (fyv >= fzv), 'float32'),
(fx-(fy < fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv-(fyv < fzv), 'float32'), # 45 (fx - (fy < fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv - (fyv < fzv), 'float32'), # 45
(fx-(fy <= fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv-(fyv <= fzv), 'float32'), (fx - (fy <= fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv - (fyv <= fzv), 'float32'),
(fx-T.eq(fy, fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv-( (fx - T.eq(fy, fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fyv == fzv), 'float32'), fxv - (fyv == fzv), 'float32'),
(fx-T.neq(fy, fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv-( (fx - T.neq(fy, fz), (fx, fy, fz), (fxv, fyv, fzv), 1, fxv - (
fyv != fzv), 'float32'), fyv != fzv), 'float32'),
(fx-fy+tensor.tan(fz), (fx, fy, fz), (fxv, fyv, fzv), 1, (fx - fy + tensor.tan(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fxv-fyv+numpy.tan(fzv), 'float32'), fxv - fyv + numpy.tan(fzv), 'float32'),
(fx-fy+tensor.tanh(fz), (fx, fy, fz), (fxv, fyv, fzv), 1, (fx - fy + tensor.tanh(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fxv-fyv+numpy.tanh(fzv), 'float32'), # 50 fxv - fyv + numpy.tanh(fzv), 'float32'), # 50
(fx-fy+tensor.sin(fz), (fx, fy, fz), (fxv, fyv, fzv), 1, (fx - fy + tensor.sin(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fxv-fyv+numpy.sin(fzv), 'float32'), fxv - fyv + numpy.sin(fzv), 'float32'),
(fx-fy+tensor.sinh(fz), (fx, fy, fz), (fxv, fyv, fzv), 1, (fx - fy + tensor.sinh(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fxv-fyv+numpy.sinh(fzv), 'float32'), fxv - fyv + numpy.sinh(fzv), 'float32'),
(fx-fy+theano.tensor.sqr(fz), (fx, fy, fz), (fxv, fyv, (fx - fy + theano.tensor.sqr(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fzv), 1, fxv-fyv+(fzv*fzv), 'float32'), fxv - fyv + (fzv * fzv), 'float32'),
(fx-fy+theano.tensor.sqrt(fz), (fx, fy, fz), (fxv, fyv, (fx - fy + theano.tensor.sqrt(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fzv), 1, fxv-fyv+numpy.sqrt(fzv), 'float32'), fxv - fyv + numpy.sqrt(fzv), 'float32'),
(fx-fy+theano.tensor.inv(fz), (fx, fy, fz), (fxv, fyv, (fx - fy + theano.tensor.inv(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fzv), 1, fxv-fyv+(1/fzv), 'float32'), # 55 fxv - fyv + (1 / fzv), 'float32'), # 55
(fx-fy+theano.tensor.neg(fz), (fx, fy, fz), (fxv, fyv, (fx - fy + theano.tensor.neg(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fzv), 1, fxv-fyv+(-fzv), 'float32'), fxv - fyv + (-fzv), 'float32'),
(fx-fy+theano.tensor.round(fz), (fx, fy, fz), (fxv, fyv, (fx - fy + theano.tensor.round(fz), (fx, fy, fz), (fxv, fyv, fzv), 1,
fzv), 1, fxv-fyv+numpy.round(fzv), 'float32'), fxv - fyv + numpy.round(fzv), 'float32'),
(ix-iy+theano.tensor.iround(fz), (ix, iy, fz), (ixv, (ix - iy + theano.tensor.iround(fz), (ix, iy, fz), (ixv, iyv, fzv), 1,
iyv, fzv), 1, ixv-iyv+numpy.round(fzv), 'int64'), ixv - iyv + numpy.round(fzv), 'int64'),
# Bit op # Bit op
(fx-theano.tensor.or_(iy, iz), (fx, iy, iz), (fxv, iyv, (fx - theano.tensor.or_(iy, iz), (fx, iy, iz), (fxv, iyv, izv), 1,
izv), 1, fxv-(iyv|izv), {'custom': 'float64', 'numpy+floatX': config.floatX, 'numpy': 'float64'}), fxv - (iyv | izv), {'custom': 'float64', 'numpy + floatX': config.floatX, 'numpy': 'float64'}),
(fx-theano.tensor.xor(iy, iz), (fx, iy, iz), (fxv, iyv, (fx - theano.tensor.xor(iy, iz), (fx, iy, iz), (fxv, iyv, izv), 1,
izv), 1, fxv-(iyv^izv), {'custom': 'float64', 'numpy+floatX': config.floatX, 'numpy': 'float64'}), # 60 fxv - (iyv ^ izv), {'custom': 'float64', 'numpy + floatX': config.floatX, 'numpy': 'float64'}), # 60
(fx-theano.tensor.and_(iy, iz), (fx, iy, iz), (fxv, iyv, (fx - theano.tensor.and_(iy, iz), (fx, iy, iz), (fxv, iyv, izv), 1,
izv), 1, fxv-(iyv&izv), {'custom': 'float64', 'numpy+floatX': config.floatX, 'numpy': 'float64'}), fxv - (iyv & izv), {'custom': 'float64', 'numpy + floatX': config.floatX, 'numpy': 'float64'}),
(fx-theano.tensor.invert(iy), (fx, iy), (fxv, iyv), 1, (fx - theano.tensor.invert(iy), (fx, iy), (fxv, iyv), 1,
fxv-(~iyv), {'custom': 'float64', 'numpy+floatX': config.floatX, 'numpy': 'float64'}), fxv - (~iyv), {'custom': 'float64', 'numpy + floatX': config.floatX, 'numpy': 'float64'}),
(fx-theano.tensor.cast(fy, dtype='float64'), (fx, fy), (fxv, fyv), 1, (fx - theano.tensor.cast(fy, dtype='float64'), (fx, fy), (fxv, fyv), 1,
fxv-numpy.asarray(fyv, 'float64'), 'float64'), fxv - numpy.asarray(fyv, 'float64'), 'float64'),
(theano.tensor.pow(fx*fy+fz, fx*fy), (fx, fy, fz), (fxv, (theano.tensor.pow(fx * fy + fz, fx * fy), (fx, fy, fz), (fxv, fyv, fzv), 1,
fyv, fzv), 1, numpy.power(fxv*fyv+fzv, fxv*fyv), 'float32'), numpy.power(fxv * fyv + fzv, fxv * fyv), 'float32'),
(fv+fy**fz, (fv, fy, fz), (fvv, fyv, fzv), 2, fvv+fyv**fzv, (fv + fy ** fz, (fv, fy, fz), (fvv, fyv, fzv), 2, fvv + fyv ** fzv, 'float32'), # fused with a dimshuffle #65
'float32'), # fused with a dimshuffle #65 (fv - fy + tensor.tanh(fz), (fv, fy, fz), (fvv, fyv, fzv), 2,
(fv-fy+tensor.tanh(fz), (fv, fy, fz), (fvv, fyv, fzv), 2, fvv - fyv + numpy.tanh(fzv), 'float32'), # fused with a dimshuffle
fvv-fyv+numpy.tanh(fzv), 'float32'), # fused with a dimshuffle
# Cases where the same input is reused many times. # Cases where the same input is reused many times.
(theano.tensor.mul(fx, fx, fx, fx), (fx,), (fxv,), 1, fxv * (theano.tensor.mul(fx, fx, fx, fx), (fx,), (fxv,), 1, fxv *
fxv*fxv*fxv, 'float32'), fxv * fxv * fxv, 'float32'),
(theano.tensor.mul(fx, ftanx, ftanx), (fx,), (fxv,), 1, (theano.tensor.mul(fx, ftanx, ftanx), (fx,), (fxv,), 1,
fxv*numpy.tan(fxv)*numpy.tan(fxv), 'float32'), fxv * numpy.tan(fxv) * numpy.tan(fxv), 'float32'),
(theano.tensor.mul(fx, ftanx, ftanx, fx), (fx,), (fxv,), (theano.tensor.mul(fx, ftanx, ftanx, fx), (fx,), (fxv,),
1, fxv*numpy.tan(fxv)*numpy.tan(fxv)*fxv, 'float32'), 1, fxv * numpy.tan(fxv) * numpy.tan(fxv) * fxv, 'float32'),
(theano.tensor.mul(ftanx, ftanx, fx+fy), (fx, fy), (fxv, (theano.tensor.mul(ftanx, ftanx, fx + fy), (fx, fy), (fxv, fyv),
fyv), 1, numpy.tan(fxv)*numpy.tan(fxv)*(fxv+fyv), 'float32'), # 70 1, numpy.tan(fxv) * numpy.tan(fxv) * (fxv + fyv), 'float32'), # 70
# Cases with different broadcast pattern. They should not # Cases with different broadcast pattern. They should not
# be merged as this would duplicate computation # be merged as this would duplicate computation
# The graph should have 2 elemwise and 1 dimshuffle # The graph should have 2 elemwise and 1 dimshuffle
(fx*theano.tensor.sin(fs), (fx, fs), (fxv, (fx * theano.tensor.sin(fs), (fx, fs), (fxv, fsv), 3,
fsv), 3, fxv*numpy.sin(fsv), 'float32'), fxv * numpy.sin(fsv), 'float32'),
] ]
if slice: if slice:
cases = cases[slice] cases = cases[slice]
...@@ -1149,15 +1138,14 @@ class test_fusion(unittest.TestCase): ...@@ -1149,15 +1138,14 @@ class test_fusion(unittest.TestCase):
print(val_inputs) print(val_inputs)
print(out) print(out)
print(answer * nb_repeat) print(answer * nb_repeat)
#assert 0
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
if gpu: if gpu:
import theano.sandbox.cuda as cuda import theano.sandbox.cuda as cuda
topo_ = [x for x in topo if not isinstance( topo_ = [x for x in topo if not isinstance(
x.op, (cuda.basic_ops.GpuFromHost, cuda.basic_ops.HostFromGpu))] x.op, (cuda.basic_ops.GpuFromHost, cuda.basic_ops.HostFromGpu))]
gpu_ = [x for x in topo if isinstance(x.op, gpu_ = [x for x in topo
cuda.basic_ops.GpuFromHost)] if isinstance(x.op, cuda.basic_ops.GpuFromHost)]
if not len(gpu_) == len(sym_inputs): if not len(gpu_) == len(sym_inputs):
fail2.append((id, gpu_, sym_inputs)) fail2.append((id, gpu_, sym_inputs))
else: else:
...@@ -1173,7 +1161,7 @@ class test_fusion(unittest.TestCase): ...@@ -1173,7 +1161,7 @@ class test_fusion(unittest.TestCase):
if len(set(g.owner.inputs)) == len(g.owner.inputs): if len(set(g.owner.inputs)) == len(g.owner.inputs):
expected_len_sym_inputs = numpy.sum( expected_len_sym_inputs = numpy.sum(
[not isinstance(x, theano.gof.Constant) [not isinstance(x, theano.gof.Constant)
for x in topo_[0].inputs]) for x in topo_[0].inputs])
assert expected_len_sym_inputs == len(sym_inputs) assert expected_len_sym_inputs == len(sym_inputs)
if not out_dtype == out.dtype: if not out_dtype == out.dtype:
...@@ -1183,7 +1171,7 @@ class test_fusion(unittest.TestCase): ...@@ -1183,7 +1171,7 @@ class test_fusion(unittest.TestCase):
print("Executed", len(cases), "cases", "failed", failed) print("Executed", len(cases), "cases", "failed", failed)
if failed > 0: if failed > 0:
raise Exception("Failed %d cases" % failed, fail1, raise Exception("Failed %d cases" % failed, fail1,
fail2, fail3, fail4) fail2, fail3, fail4)
return times return times
...@@ -1213,11 +1201,11 @@ class test_fusion(unittest.TestCase): ...@@ -1213,11 +1201,11 @@ class test_fusion(unittest.TestCase):
# we need the optimisation enabled, debug do this. # we need the optimisation enabled, debug do this.
if theano.config.mode == "FAST_COMPILE": if theano.config.mode == "FAST_COMPILE":
mode = theano.compile.mode.get_mode("FAST_RUN").including( mode = theano.compile.mode.get_mode("FAST_RUN").including(
'local_elemwise_fusion', 'composite_elemwise_fusion', 'local_elemwise_fusion', 'composite_elemwise_fusion',
'canonicalize', 'gpu') 'canonicalize', 'gpu')
else: else:
mode = theano.compile.mode.get_default_mode().including( mode = theano.compile.mode.get_default_mode().including(
'local_elemwise_fusion', 'composite_elemwise_fusion', 'local_elemwise_fusion', 'composite_elemwise_fusion',
'canonicalize', 'gpu') 'canonicalize', 'gpu')
import theano.sandbox.cuda as cuda import theano.sandbox.cuda as cuda
if not cuda.cuda_available: if not cuda.cuda_available:
...@@ -1230,11 +1218,11 @@ class test_fusion(unittest.TestCase): ...@@ -1230,11 +1218,11 @@ class test_fusion(unittest.TestCase):
# we need the optimisation enabled, debug do this. # we need the optimisation enabled, debug do this.
if theano.config.mode == "FAST_COMPILE": if theano.config.mode == "FAST_COMPILE":
mode = theano.compile.mode.get_mode("FAST_RUN").including( mode = theano.compile.mode.get_mode("FAST_RUN").including(
'local_elemwise_fusion', 'composite_elemwise_fusion', 'local_elemwise_fusion', 'composite_elemwise_fusion',
'canonicalize', 'gpu') 'canonicalize', 'gpu')
else: else:
mode = theano.compile.mode.get_default_mode().including( mode = theano.compile.mode.get_default_mode().including(
'local_elemwise_fusion', 'composite_elemwise_fusion', 'local_elemwise_fusion', 'composite_elemwise_fusion',
'canonicalize', 'gpu') 'canonicalize', 'gpu')
import theano.sandbox.cuda as cuda import theano.sandbox.cuda as cuda
if not cuda.cuda_available: if not cuda.cuda_available:
...@@ -1301,8 +1289,8 @@ class test_fusion(unittest.TestCase): ...@@ -1301,8 +1289,8 @@ class test_fusion(unittest.TestCase):
shp = (3000, 3000) shp = (3000, 3000)
shp = (1000, 1000) shp = (1000, 1000)
nb_repeat = 50 nb_repeat = 50
# linker=gof.CLinker # linker=gof.CLinker
# linker=gof.OpWiseCLinker # linker=gof.OpWiseCLinker
mode1 = copy.copy(compile.get_default_mode()) mode1 = copy.copy(compile.get_default_mode())
mode1._optimizer = mode1._optimizer.including('local_elemwise_fusion') mode1._optimizer = mode1._optimizer.including('local_elemwise_fusion')
...@@ -1325,15 +1313,15 @@ class test_fusion(unittest.TestCase): ...@@ -1325,15 +1313,15 @@ class test_fusion(unittest.TestCase):
print("times2/times1") print("times2/times1")
print(d) print(d)
print("min", d.min(), "argmin", d.argmin(), "max", d.max(), \ print("min", d.min(), "argmin", d.argmin(), "max", d.max(),
"mean", d.mean(), "std", d.std()) "mean", d.mean(), "std", d.std())
def test_fusion_inplace(self): def test_fusion_inplace(self):
mode = copy.copy(compile.mode.get_default_mode()) mode = copy.copy(compile.mode.get_default_mode())
# we need the optimisation enabled and the canonicalize. # we need the optimisation enabled and the canonicalize.
# the canonicalize is needed to merge multiplication/addition by constant. # the canonicalize is needed to merge multiplication/addition by constant.
mode._optimizer = mode._optimizer.including( mode._optimizer = mode._optimizer.including(
'local_elemwise_fusion', 'composite_elemwise_fusion', 'local_elemwise_fusion', 'composite_elemwise_fusion',
'canonicalize', 'inplace') 'canonicalize', 'inplace')
x, y, z = dmatrices('xyz') x, y, z = dmatrices('xyz')
...@@ -1346,8 +1334,8 @@ class test_fusion(unittest.TestCase): ...@@ -1346,8 +1334,8 @@ class test_fusion(unittest.TestCase):
def speed_fusion_gpu(self): def speed_fusion_gpu(self):
import theano.sandbox.cuda as cuda import theano.sandbox.cuda as cuda
self.speed_fusion(shared_fn=cuda. self.speed_fusion(shared_fn=cuda.float32_shared_constructor,
float32_shared_constructor, gpu=True, s=slice(0, 15)) gpu=True, s=slice(0, 15))
def speed_log_exp(self): def speed_log_exp(self):
s = slice(31, 36) s = slice(31, 36)
...@@ -1385,13 +1373,15 @@ class test_fusion(unittest.TestCase): ...@@ -1385,13 +1373,15 @@ class test_fusion(unittest.TestCase):
gc.collect() gc.collect()
gc.collect() gc.collect()
gc.collect() gc.collect()
# print 'v1',v1 v1 = weakref.ref(v) # noqa
v1 = weakref.ref(v)
pdb.set_trace() pdb.set_trace()
# no memory leak
# f = orig_function([compile.In(fx),compile.In(variable=fy, value=None)], # f = orig_function([compile.In(fx),compile.In(variable=fy, value=None)],
# [fy+fx],mode=mode)#no memory leak # [fy+fx],mode=mode)
f = orig_function([compile.In(fx), compile.In(variable=fy, value=v)], # memory leak
[fy + fx], mode=mode) # memory leak f = orig_function( # noqa
[compile.In(fx), compile.In(variable=fy, value=v)],
[fy + fx], mode=mode)
del v del v
gc.collect() gc.collect()
gc.collect() gc.collect()
...@@ -1428,8 +1418,7 @@ class test_fusion(unittest.TestCase): ...@@ -1428,8 +1418,7 @@ class test_fusion(unittest.TestCase):
for x in ndl: for x in ndl:
cmp = not isinstance(x, list) cmp = not isinstance(x, list)
if not cmp and x: if not cmp and x:
cmp = x[0].__class__.__name__ != \ cmp = (x[0].__class__.__name__ != 'array_converter')
'array_converter'
if cmp: if cmp:
cmp = x[0] != 'Option' cmp = x[0] != 'Option'
if cmp: if cmp:
...@@ -1497,13 +1486,13 @@ class TestCompositeCodegen(unittest.TestCase): ...@@ -1497,13 +1486,13 @@ class TestCompositeCodegen(unittest.TestCase):
self.scal_times_2 = TimesN(2, upgrade_to_float, name='times_2') self.scal_times_2 = TimesN(2, upgrade_to_float, name='times_2')
self.times_2 = theano.tensor.elemwise.Elemwise( self.times_2 = theano.tensor.elemwise.Elemwise(
self.scal_times_2, self.scal_times_2,
name='times_2') name='times_2')
self.scal_times_3 = TimesN(3, upgrade_to_float, name='times_3') self.scal_times_3 = TimesN(3, upgrade_to_float, name='times_3')
self.times_3 = theano.tensor.elemwise.Elemwise( self.times_3 = theano.tensor.elemwise.Elemwise(
self.scal_times_3, self.scal_times_3,
name='times_3') name='times_3')
self.x = fvector() self.x = fvector()
...@@ -1525,8 +1514,9 @@ class TestCompositeCodegen(unittest.TestCase): ...@@ -1525,8 +1514,9 @@ class TestCompositeCodegen(unittest.TestCase):
y = self.times_2(self.x) y = self.times_2(self.x)
z = self.times_3(y) z = self.times_3(y)
f = theano.function([self.x], cuda.gpu_from_host(z), f = theano.function(
mode=theano.compile.mode.get_default_mode().including('gpu')) [self.x], cuda.gpu_from_host(z),
mode=theano.compile.mode.get_default_mode().including('gpu'))
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
if config.mode != "FAST_COMPILE": if config.mode != "FAST_COMPILE":
assert len(topo) == 2 assert len(topo) == 2
...@@ -1564,8 +1554,8 @@ def test_log1p(): ...@@ -1564,8 +1554,8 @@ def test_log1p():
assert [node.op for node in f.maker.fgraph.toposort()][3:] == [ assert [node.op for node in f.maker.fgraph.toposort()][3:] == [
T.log1p, tensor.alloc] T.log1p, tensor.alloc]
f = function([x, y], T.log(2 + (x) - tensor.fill(y, 1.0)), mode=m) f = function([x, y], T.log(2 + (x) - tensor.fill(y, 1.0)), mode=m)
assert [node.op for node in f.maker.fgraph.toposort()][3:] \ assert ([node.op for node in f.maker.fgraph.toposort()][3:] ==
== [T.log1p, tensor.alloc] [T.log1p, tensor.alloc])
f([1e-7, 10], [[0, 0], [0, 0]]) # debugmode will verify values f([1e-7, 10], [[0, 0], [0, 0]]) # debugmode will verify values
...@@ -1636,8 +1626,8 @@ def test_local_useless_slice(): ...@@ -1636,8 +1626,8 @@ def test_local_useless_slice():
f_unopt = theano.function([x], o, mode=mode_unopt) f_unopt = theano.function([x], o, mode=mode_unopt)
f_opt = theano.function([x], o, mode=mode_opt) f_opt = theano.function([x], o, mode=mode_opt)
test_inp = numpy.random.randint(-10, 10, (4, 4)).astype('float32') test_inp = numpy.random.randint(-10, 10, (4, 4)).astype('float32')
assert all(f_opt(test_inp) == f_unopt(test_inp)),\ assert all(f_opt(test_inp) == f_unopt(test_inp)),\
"The optimization caused a mismatch in the result" "The optimization caused a mismatch in the result"
# test to see if the slice is truely gone # test to see if the slice is truely gone
apply_node = f_opt.maker.fgraph.toposort()[0] apply_node = f_opt.maker.fgraph.toposort()[0]
subtens = apply_node.op subtens = apply_node.op
...@@ -1669,6 +1659,7 @@ def test_local_useless_slice(): ...@@ -1669,6 +1659,7 @@ def test_local_useless_slice():
assert check_stack_trace(f_opt_check, ops_to_check=Subtensor) assert check_stack_trace(f_opt_check, ops_to_check=Subtensor)
assert check_stack_trace(f_opt_check_apply, ops_to_check=Subtensor) assert check_stack_trace(f_opt_check_apply, ops_to_check=Subtensor)
def test_local_useless_inc_subtensor(): def test_local_useless_inc_subtensor():
x = tensor.matrix('x') x = tensor.matrix('x')
y = tensor.matrix('y') y = tensor.matrix('y')
...@@ -1943,7 +1934,7 @@ class test_local_subtensor_make_vector(unittest.TestCase): ...@@ -1943,7 +1934,7 @@ class test_local_subtensor_make_vector(unittest.TestCase):
v = make_vector(x, y, z) v = make_vector(x, y, z)
mode = theano.compile.mode.get_default_mode().including( mode = theano.compile.mode.get_default_mode().including(
"local_subtensor_make_vector") "local_subtensor_make_vector")
# list of subtensor cases, where local_subtensor_make_vector # list of subtensor cases, where local_subtensor_make_vector
# inserts a new MakeVector node # inserts a new MakeVector node
...@@ -1977,7 +1968,7 @@ class test_local_subtensor_lift(unittest.TestCase): ...@@ -1977,7 +1968,7 @@ class test_local_subtensor_lift(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=[ self.assertTrue(check_stack_trace(f, ops_to_check=[
Subtensor, tensor.Elemwise])) Subtensor, tensor.Elemwise]))
prog = f.maker.fgraph.toposort() prog = f.maker.fgraph.toposort()
assert prog[0].op == tensor.exp assert prog[0].op == tensor.exp
...@@ -1995,7 +1986,7 @@ class test_local_subtensor_lift(unittest.TestCase): ...@@ -1995,7 +1986,7 @@ class test_local_subtensor_lift(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=[ self.assertTrue(check_stack_trace(f, ops_to_check=[
Subtensor, tensor.DimShuffle])) Subtensor, tensor.DimShuffle]))
prog = f.maker.fgraph.toposort() prog = f.maker.fgraph.toposort()
assert isinstance(prog[0].op, tensor.Subtensor) assert isinstance(prog[0].op, tensor.Subtensor)
...@@ -2004,8 +1995,8 @@ class test_local_subtensor_lift(unittest.TestCase): ...@@ -2004,8 +1995,8 @@ class test_local_subtensor_lift(unittest.TestCase):
assert isinstance(prog[3].op.scalar_op, theano.scalar. assert isinstance(prog[3].op.scalar_op, theano.scalar.
Composite) # Composite{add,add} Composite) # Composite{add,add}
assert len(prog) == 4 assert len(prog) == 4
# let debugmode test something
f([[0, 1], [2, 3]], 4, [[4, 5], [6, 7]]) f([[0, 1], [2, 3]], 4, [[4, 5], [6, 7]])
# let debugmode test something
def test2(self): def test2(self):
# as 1, but take a slice # as 1, but take a slice
...@@ -2016,7 +2007,7 @@ class test_local_subtensor_lift(unittest.TestCase): ...@@ -2016,7 +2007,7 @@ class test_local_subtensor_lift(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=[ self.assertTrue(check_stack_trace(f, ops_to_check=[
Subtensor, tensor.DimShuffle])) Subtensor, tensor.DimShuffle]))
prog = f.maker.fgraph.toposort() prog = f.maker.fgraph.toposort()
assert isinstance(prog[0].op, tensor.Subtensor) assert isinstance(prog[0].op, tensor.Subtensor)
...@@ -2025,8 +2016,8 @@ class test_local_subtensor_lift(unittest.TestCase): ...@@ -2025,8 +2016,8 @@ class test_local_subtensor_lift(unittest.TestCase):
assert isinstance(prog[3].op.scalar_op, theano.scalar. assert isinstance(prog[3].op.scalar_op, theano.scalar.
Composite) # Composite{add,add} Composite) # Composite{add,add}
assert len(prog) == 4 assert len(prog) == 4
# let debugmode test something
f([[0, 1], [2, 3]], 4, [[4, 5], [6, 7]]) f([[0, 1], [2, 3]], 4, [[4, 5], [6, 7]])
# let debugmode test something
def test3(self): def test3(self):
# basic test that the optimization does work with broadcasting # basic test that the optimization does work with broadcasting
...@@ -2142,7 +2133,7 @@ class test_local_subtensor_lift(unittest.TestCase): ...@@ -2142,7 +2133,7 @@ class test_local_subtensor_lift(unittest.TestCase):
f3 = function([y], newy[:, 3, 0], mode=mode_opt) f3 = function([y], newy[:, 3, 0], mode=mode_opt)
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f3, ops_to_check=[ self.assertTrue(check_stack_trace(f3, ops_to_check=[
Subtensor, tensor.Rebroadcast])) Subtensor, tensor.Rebroadcast]))
prog = f3.maker.fgraph.toposort() prog = f3.maker.fgraph.toposort()
assert isinstance(prog[0].op, tensor.Subtensor) assert isinstance(prog[0].op, tensor.Subtensor)
assert isinstance(prog[1].op, tensor.Rebroadcast) assert isinstance(prog[1].op, tensor.Rebroadcast)
...@@ -2155,11 +2146,10 @@ class test_local_subtensor_lift(unittest.TestCase): ...@@ -2155,11 +2146,10 @@ class test_local_subtensor_lift(unittest.TestCase):
newz = tensor.Rebroadcast((3, True))(z) newz = tensor.Rebroadcast((3, True))(z)
assert newz.broadcastable == (False, False, False, True) assert newz.broadcastable == (False, False, False, True)
out = newz[:, 3, 0]
f4 = function([z], newz[:, 3, 0], mode=mode_opt) f4 = function([z], newz[:, 3, 0], mode=mode_opt)
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f4, ops_to_check=[ self.assertTrue(check_stack_trace(f4, ops_to_check=[
Subtensor, tensor.Rebroadcast])) Subtensor, tensor.Rebroadcast]))
prog = f4.maker.fgraph.toposort() prog = f4.maker.fgraph.toposort()
assert isinstance(prog[0].op, tensor.Subtensor) assert isinstance(prog[0].op, tensor.Subtensor)
assert isinstance(prog[1].op, tensor.Rebroadcast) assert isinstance(prog[1].op, tensor.Rebroadcast)
...@@ -2208,7 +2198,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2208,7 +2198,7 @@ class test_local_subtensor_merge(unittest.TestCase):
f = function([x, y], x[y::][-1], mode=mode_opt) f = function([x, y], x[y::][-1], mode=mode_opt)
g = function([x, y], x[y::][-1], g = function([x, y], x[y::][-1],
mode=mode_opt.excluding('local_subtensor_merge')) mode=mode_opt.excluding('local_subtensor_merge'))
#theano.printing.debugprint(f, print_type=True) # theano.printing.debugprint(f, print_type=True)
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor)) self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor))
...@@ -2243,7 +2233,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2243,7 +2233,7 @@ class test_local_subtensor_merge(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor)) self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor))
#theano.printing.debugprint(f, print_type=True) # theano.printing.debugprint(f, print_type=True)
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
# print [t for t in topo if isinstance(t.op, tensor.Subtensor)] # print [t for t in topo if isinstance(t.op, tensor.Subtensor)]
assert len([t for t in topo assert len([t for t in topo
...@@ -2269,7 +2259,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2269,7 +2259,7 @@ class test_local_subtensor_merge(unittest.TestCase):
f = function([x, y], x[::-1][y], mode=mode_opt) f = function([x, y], x[::-1][y], mode=mode_opt)
g = function([x, y], x[::-1][y], g = function([x, y], x[::-1][y],
mode=mode_opt.excluding('local_subtensor_merge')) mode=mode_opt.excluding('local_subtensor_merge'))
#theano.printing.debugprint(f, print_type=True) # theano.printing.debugprint(f, print_type=True)
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor)) self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor))
...@@ -2299,7 +2289,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2299,7 +2289,7 @@ class test_local_subtensor_merge(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor)) self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor))
#theano.printing.debugprint(f, print_type=True) # theano.printing.debugprint(f, print_type=True)
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
# print [t for t in topo if isinstance(t.op, tensor.Subtensor)] # print [t for t in topo if isinstance(t.op, tensor.Subtensor)]
assert len([t for t in topo assert len([t for t in topo
...@@ -2320,7 +2310,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2320,7 +2310,7 @@ class test_local_subtensor_merge(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor)) self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor))
#theano.printing.debugprint(f, print_type=True) # theano.printing.debugprint(f, print_type=True)
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
# print [t for t in topo if isinstance(t.op, tensor.Subtensor)] # print [t for t in topo if isinstance(t.op, tensor.Subtensor)]
...@@ -2344,7 +2334,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2344,7 +2334,7 @@ class test_local_subtensor_merge(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check='all')) self.assertTrue(check_stack_trace(f, ops_to_check='all'))
#theano.printing.debugprint(f, print_type=True) # theano.printing.debugprint(f, print_type=True)
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
# print [t for t in topo if isinstance(t.op, tensor.Subtensor)] # print [t for t in topo if isinstance(t.op, tensor.Subtensor)]
assert len([t for t in topo assert len([t for t in topo
...@@ -2366,7 +2356,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2366,7 +2356,7 @@ class test_local_subtensor_merge(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor)) self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor))
#theano.printing.debugprint(f, print_type=True) # theano.printing.debugprint(f, print_type=True)
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
# print [t for t in topo if isinstance(t.op, tensor.Subtensor)] # print [t for t in topo if isinstance(t.op, tensor.Subtensor)]
...@@ -2398,7 +2388,6 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2398,7 +2388,6 @@ class test_local_subtensor_merge(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor)) self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor))
x_val = self.rng.uniform(size=shape).astype(config.floatX) x_val = self.rng.uniform(size=shape).astype(config.floatX)
f(x_val) f(x_val)
...@@ -2418,7 +2407,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2418,7 +2407,7 @@ class test_local_subtensor_merge(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor)) self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor))
#theano.printing.debugprint(f, print_type=True) # theano.printing.debugprint(f, print_type=True)
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
# print [t for t in topo if isinstance(t.op, tensor.Subtensor)] # print [t for t in topo if isinstance(t.op, tensor.Subtensor)]
...@@ -2447,7 +2436,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2447,7 +2436,7 @@ class test_local_subtensor_merge(unittest.TestCase):
for s2 in s2r: for s2 in s2r:
f(x_val, b1, e1, s1, b2, e2, s2) f(x_val, b1, e1, s1, b2, e2, s2)
def test_const4(self): def test_const5(self):
# Bug reported by Razvan # Bug reported by Razvan
data = numpy.asarray(numpy.arange(8), data = numpy.asarray(numpy.arange(8),
dtype=theano.config.floatX) dtype=theano.config.floatX)
...@@ -2460,7 +2449,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2460,7 +2449,7 @@ class test_local_subtensor_merge(unittest.TestCase):
val = fun(data) val = fun(data)
assert val == data[7:1:-1][0] assert val == data[7:1:-1][0]
def test_const5(self): def test_const6(self):
# Bug reported by Graham # Bug reported by Graham
data = self.rng.uniform(size=(8, 8, 8)).astype(theano.config.floatX) data = self.rng.uniform(size=(8, 8, 8)).astype(theano.config.floatX)
x = theano.tensor.tensor3('x') x = theano.tensor.tensor3('x')
...@@ -2506,7 +2495,7 @@ class test_local_subtensor_merge(unittest.TestCase): ...@@ -2506,7 +2495,7 @@ class test_local_subtensor_merge(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor)) self.assertTrue(check_stack_trace(f, ops_to_check=Subtensor))
#theano.printing.debugprint(f, print_type=True) # theano.printing.debugprint(f, print_type=True)
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
# print [t for t in topo if isinstance(t.op, tensor.Subtensor)] # print [t for t in topo if isinstance(t.op, tensor.Subtensor)]
...@@ -2762,7 +2751,6 @@ class test_local_adv_sub1_adv_inc_sub1(unittest.TestCase): ...@@ -2762,7 +2751,6 @@ class test_local_adv_sub1_adv_inc_sub1(unittest.TestCase):
dx = numpy.random.rand(4, 5).astype(config.floatX) dx = numpy.random.rand(4, 5).astype(config.floatX)
dy = numpy.random.rand(2, 5).astype(config.floatX) dy = numpy.random.rand(2, 5).astype(config.floatX)
didx = numpy.asarray([1, 3], "int32")
# set_subtensor # set_subtensor
inc = tensor.set_subtensor(x[idx], y) inc = tensor.set_subtensor(x[idx], y)
...@@ -2809,8 +2797,8 @@ class Test_alloc_zero(unittest.TestCase): ...@@ -2809,8 +2797,8 @@ class Test_alloc_zero(unittest.TestCase):
y0 = tensor.zeros_like(y) y0 = tensor.zeros_like(y)
z = tensor.set_subtensor(x0[:4], y0) z = tensor.set_subtensor(x0[:4], y0)
f = theano.function([x, y], z, mode=self.mode) f = theano.function([x, y], z, mode=self.mode)
assert numpy.all([not isinstance(n.op, tensor.IncSubtensor) for n in assert numpy.all([not isinstance(n.op, tensor.IncSubtensor)
f.maker.fgraph.toposort()]) for n in f.maker.fgraph.toposort()])
def test_setsubtensor_allocs1(self): def test_setsubtensor_allocs1(self):
y = tensor.matrix() y = tensor.matrix()
...@@ -2819,8 +2807,8 @@ class Test_alloc_zero(unittest.TestCase): ...@@ -2819,8 +2807,8 @@ class Test_alloc_zero(unittest.TestCase):
y0 = tensor.zeros_like(y) y0 = tensor.zeros_like(y)
z = tensor.set_subtensor(x0[:4], y0) z = tensor.set_subtensor(x0[:4], y0)
f = theano.function([y], z, mode=self.mode) f = theano.function([y], z, mode=self.mode)
assert numpy.all([not isinstance(n.op, tensor.IncSubtensor) for n in assert numpy.all([not isinstance(n.op, tensor.IncSubtensor)
f.maker.fgraph.toposort()]) for n in f.maker.fgraph.toposort()])
def test_setsubtensor_allocs1t(self): def test_setsubtensor_allocs1t(self):
y = tensor.matrix() y = tensor.matrix()
...@@ -2829,8 +2817,8 @@ class Test_alloc_zero(unittest.TestCase): ...@@ -2829,8 +2817,8 @@ class Test_alloc_zero(unittest.TestCase):
y0 = tensor.zeros_like(y) y0 = tensor.zeros_like(y)
z = tensor.set_subtensor(x0[:4], y0.T) z = tensor.set_subtensor(x0[:4], y0.T)
f = theano.function([y], z, mode=mode_opt) f = theano.function([y], z, mode=mode_opt)
assert numpy.all([not isinstance(n.op, tensor.IncSubtensor) for n in assert numpy.all([not isinstance(n.op, tensor.IncSubtensor)
f.maker.fgraph.toposort()]) for n in f.maker.fgraph.toposort()])
def test_setsubtensor_allocs2(self): def test_setsubtensor_allocs2(self):
x = tensor.matrix() x = tensor.matrix()
...@@ -2839,8 +2827,8 @@ class Test_alloc_zero(unittest.TestCase): ...@@ -2839,8 +2827,8 @@ class Test_alloc_zero(unittest.TestCase):
x0 = tensor.zeros_like(x) x0 = tensor.zeros_like(x)
z = tensor.set_subtensor(x0[:4], y0) z = tensor.set_subtensor(x0[:4], y0)
f = theano.function([x], z, mode=self.mode) f = theano.function([x], z, mode=self.mode)
assert numpy.all([not isinstance(x.op, tensor.IncSubtensor) for x in assert numpy.all([not isinstance(x.op, tensor.IncSubtensor)
f.maker.fgraph.toposort()]) for x in f.maker.fgraph.toposort()])
def test_incsubtensor_allocs0(self): def test_incsubtensor_allocs0(self):
x = tensor.matrix() x = tensor.matrix()
...@@ -2848,8 +2836,8 @@ class Test_alloc_zero(unittest.TestCase): ...@@ -2848,8 +2836,8 @@ class Test_alloc_zero(unittest.TestCase):
y0 = tensor.zeros_like(y) y0 = tensor.zeros_like(y)
z = tensor.inc_subtensor(x[:4], y0) z = tensor.inc_subtensor(x[:4], y0)
f = theano.function([x, y], z, mode=self.mode) f = theano.function([x, y], z, mode=self.mode)
assert numpy.all([not isinstance(n.op, tensor.IncSubtensor) for n in assert numpy.all([not isinstance(n.op, tensor.IncSubtensor)
f.maker.fgraph.toposort()]) for n in f.maker.fgraph.toposort()])
def test_incsubtensor_allocs0t(self): def test_incsubtensor_allocs0t(self):
x = tensor.matrix() x = tensor.matrix()
...@@ -2857,8 +2845,8 @@ class Test_alloc_zero(unittest.TestCase): ...@@ -2857,8 +2845,8 @@ class Test_alloc_zero(unittest.TestCase):
y0 = tensor.zeros_like(y) y0 = tensor.zeros_like(y)
z = tensor.inc_subtensor(x[:4], y0.T) z = tensor.inc_subtensor(x[:4], y0.T)
f = theano.function([x, y], z, mode=mode_opt) f = theano.function([x, y], z, mode=mode_opt)
assert numpy.all([not isinstance(n.op, tensor.IncSubtensor) for n in assert numpy.all([not isinstance(n.op, tensor.IncSubtensor)
f.maker.fgraph.toposort()]) for n in f.maker.fgraph.toposort()])
def test_incsubtensor_allocs1(self): def test_incsubtensor_allocs1(self):
x = tensor.matrix() x = tensor.matrix()
...@@ -2866,8 +2854,8 @@ class Test_alloc_zero(unittest.TestCase): ...@@ -2866,8 +2854,8 @@ class Test_alloc_zero(unittest.TestCase):
dtype=config.floatX)) dtype=config.floatX))
z = tensor.inc_subtensor(x[:4], y0) z = tensor.inc_subtensor(x[:4], y0)
f = theano.function([x], z, mode=self.mode) f = theano.function([x], z, mode=self.mode)
assert numpy.all([not isinstance(x.op, tensor.IncSubtensor) for x in assert numpy.all([not isinstance(x.op, tensor.IncSubtensor)
f.maker.fgraph.toposort()]) for x in f.maker.fgraph.toposort()])
def test_advancedincsubtensor1_allocs0(self): def test_advancedincsubtensor1_allocs0(self):
x = tensor.matrix() x = tensor.matrix()
...@@ -3001,6 +2989,7 @@ def test_local_IncSubtensor_serialize(): ...@@ -3001,6 +2989,7 @@ def test_local_IncSubtensor_serialize():
tensor.IncSubtensor, tensor.AdvancedIncSubtensor, tensor.IncSubtensor, tensor.AdvancedIncSubtensor,
tensor.AdvancedIncSubtensor1]) tensor.AdvancedIncSubtensor1])
def test_local_set_to_inc_subtensor(): def test_local_set_to_inc_subtensor():
v = theano.tensor.fmatrix() v = theano.tensor.fmatrix()
s = v[[2, 1]] s = v[[2, 1]]
...@@ -3475,8 +3464,6 @@ class Test_local_useless_elemwise_comparison(unittest.TestCase): ...@@ -3475,8 +3464,6 @@ class Test_local_useless_elemwise_comparison(unittest.TestCase):
x_val = 10 x_val = 10
assert f(x_val) == x_val assert f(x_val) == x_val
#def assert_returns
def test_inequality_with_self(self): def test_inequality_with_self(self):
x = T.scalar('x', dtype=config.floatX) x = T.scalar('x', dtype=config.floatX)
mode = theano.compile.get_default_mode().including('local_useless_elemwise_comparison') mode = theano.compile.get_default_mode().including('local_useless_elemwise_comparison')
...@@ -3501,10 +3488,11 @@ class Test_local_useless_elemwise_comparison(unittest.TestCase): ...@@ -3501,10 +3488,11 @@ class Test_local_useless_elemwise_comparison(unittest.TestCase):
def test_shape_inequality_with_self(self): def test_shape_inequality_with_self(self):
x = T.vector('x', dtype=config.floatX) x = T.vector('x', dtype=config.floatX)
mode = theano.compile.get_default_mode().including('local_useless_elemwise_comparison', mode = theano.compile.get_default_mode().including(
'local_shape_to_shape_i', 'local_useless_elemwise_comparison',
'local_track_shape_i', 'local_shape_to_shape_i',
'local_subtensor_make_vector') 'local_track_shape_i',
'local_subtensor_make_vector')
f = theano.function([x], T.lt(x.shape[0], 0), mode=mode) f = theano.function([x], T.lt(x.shape[0], 0), mode=mode)
self.assert_eqs_const(f, 0) self.assert_eqs_const(f, 0)
...@@ -3534,22 +3522,23 @@ class Test_local_useless_elemwise_comparison(unittest.TestCase): ...@@ -3534,22 +3522,23 @@ class Test_local_useless_elemwise_comparison(unittest.TestCase):
assert f(x_val) == 0 assert f(x_val) == 0
f = theano.function([x], T.minimum([0, 0], x.shape[0]), mode=mode) f = theano.function([x], T.minimum([0, 0], x.shape[0]), mode=mode)
# This case isn't optimized. # This case isn't optimized.
# self.assert_eqs_const(f, 0) # self.assert_eqs_const(f, 0)
utt.assert_allclose(f(x_val), [0, 0]) utt.assert_allclose(f(x_val), [0, 0])
def test_shape_add_inequality(self): def test_shape_add_inequality(self):
x = T.vector('x', dtype=config.floatX) x = T.vector('x', dtype=config.floatX)
mode = theano.compile.get_default_mode().including('local_useless_elemwise_comparison', mode = theano.compile.get_default_mode().including(
'local_shape_to_shape_i', 'local_useless_elemwise_comparison',
'local_track_shape_i', 'local_shape_to_shape_i',
'local_subtensor_make_vector') 'local_track_shape_i',
'local_subtensor_make_vector')
y = T.vector('y', dtype=config.floatX) y = T.vector('y', dtype=config.floatX)
f = theano.function([x, y], T.lt(x.shape[0]+y.shape[0], 0), mode=mode) f = theano.function([x, y], T.lt(x.shape[0] + y.shape[0], 0), mode=mode)
self.assert_eqs_const(f, 0) self.assert_eqs_const(f, 0)
f = theano.function([x, y], T.ge(x.shape[0]+y.shape[0], 0), mode=mode) f = theano.function([x, y], T.ge(x.shape[0] + y.shape[0], 0), mode=mode)
self.assert_eqs_const(f, 1) self.assert_eqs_const(f, 1)
def test_equality_shapes(self): def test_equality_shapes(self):
...@@ -3640,8 +3629,8 @@ class Test_local_canonicalize_alloc(unittest.TestCase): ...@@ -3640,8 +3629,8 @@ class Test_local_canonicalize_alloc(unittest.TestCase):
f = function([], a, mode=mode_opt) f = function([], a, mode=mode_opt)
# The optimization should then be applied, and remove Alloc # The optimization should then be applied, and remove Alloc
assert ([node.op for node in f.maker.fgraph.toposort()] assert ([node.op for node in f.maker.fgraph.toposort()] ==
== [deep_copy_op]) [deep_copy_op])
# In DebugMode, the shape mismatch should be detected # In DebugMode, the shape mismatch should be detected
if isinstance(mode_opt, compile.DebugMode): if isinstance(mode_opt, compile.DebugMode):
...@@ -3671,7 +3660,7 @@ class Test_local_canonicalize_alloc(unittest.TestCase): ...@@ -3671,7 +3660,7 @@ class Test_local_canonicalize_alloc(unittest.TestCase):
mode = mode_opt.excluding('local_canonicalize_alloc') mode = mode_opt.excluding('local_canonicalize_alloc')
x = tensor.matrix('x') x = tensor.matrix('x')
y = tensor.tile(x, (1,)*2) y = tensor.tile(x, (1,) * 2)
f = function([x], [y], mode=mode) f = function([x], [y], mode=mode)
op_classes = [node.op.__class__ for node in f.maker.fgraph.toposort()] op_classes = [node.op.__class__ for node in f.maker.fgraph.toposort()]
...@@ -3781,7 +3770,6 @@ class Test_local_useless_inc_subtensor_alloc(unittest.TestCase): ...@@ -3781,7 +3770,6 @@ class Test_local_useless_inc_subtensor_alloc(unittest.TestCase):
self.assertTrue(check_stack_trace(f1, ops_to_check=tensor.AdvancedIncSubtensor)) self.assertTrue(check_stack_trace(f1, ops_to_check=tensor.AdvancedIncSubtensor))
self.assertTrue(check_stack_trace(f2, ops_to_check=tensor.AdvancedIncSubtensor)) self.assertTrue(check_stack_trace(f2, ops_to_check=tensor.AdvancedIncSubtensor))
def test_advanced_inc_subtensor1(self): def test_advanced_inc_subtensor1(self):
if tensor.inplace_increment is None: if tensor.inplace_increment is None:
raise SkipTest('NumPy version >= 1.8 not available') raise SkipTest('NumPy version >= 1.8 not available')
...@@ -3813,7 +3801,7 @@ class Test_local_useless_inc_subtensor_alloc(unittest.TestCase): ...@@ -3813,7 +3801,7 @@ class Test_local_useless_inc_subtensor_alloc(unittest.TestCase):
# Check stacktrace was copied over correctly after opt was applied # Check stacktrace was copied over correctly after opt was applied
self.assertTrue(check_stack_trace( self.assertTrue(check_stack_trace(
f1, ops_to_check=tensor.AdvancedIncSubtensor1)) f1, ops_to_check=tensor.AdvancedIncSubtensor1))
self.assertTrue(check_stack_trace(f2, ops_to_check='all')) self.assertTrue(check_stack_trace(f2, ops_to_check='all'))
def test_incsubtensor(self): def test_incsubtensor(self):
...@@ -4156,6 +4144,7 @@ class test_assert(utt.InferShapeTester): ...@@ -4156,6 +4144,7 @@ class test_assert(utt.InferShapeTester):
self._compile_and_check([admat, adscal, bdscal], [out], self._compile_and_check([admat, adscal, bdscal], [out],
[admat_val, adscal_val, bdscal_val], Assert) [admat_val, adscal_val, bdscal_val], Assert)
def test_local_mul_specialize(): def test_local_mul_specialize():
mode = theano.config.mode mode = theano.config.mode
if mode == 'FAST_COMPILE': if mode == 'FAST_COMPILE':
...@@ -4492,7 +4481,6 @@ class T_func_inverse(unittest.TestCase): ...@@ -4492,7 +4481,6 @@ class T_func_inverse(unittest.TestCase):
mode = theano.compile.get_default_mode() mode = theano.compile.get_default_mode()
self.mode = mode.including('local_func_inv') self.mode = mode.including('local_func_inv')
def assert_func_pair_optimized(self, func1, func2, data, def assert_func_pair_optimized(self, func1, func2, data,
should_copy=True, is_complex=False): should_copy=True, is_complex=False):
""" """
...@@ -4528,7 +4516,7 @@ class T_func_inverse(unittest.TestCase): ...@@ -4528,7 +4516,7 @@ class T_func_inverse(unittest.TestCase):
dx = numpy.random.rand(5, 4).astype("float32") dx = numpy.random.rand(5, 4).astype("float32")
self.assert_func_pair_optimized(T.deg2rad, T.rad2deg, dx) self.assert_func_pair_optimized(T.deg2rad, T.rad2deg, dx)
dx = numpy.random.rand(5, 4).astype("float32")*180 dx = numpy.random.rand(5, 4).astype("float32") * 180
self.assert_func_pair_optimized(T.rad2deg, T.deg2rad, dx) self.assert_func_pair_optimized(T.rad2deg, T.deg2rad, dx)
# Test the other functional inverses # Test the other functional inverses
...@@ -4538,13 +4526,13 @@ class T_func_inverse(unittest.TestCase): ...@@ -4538,13 +4526,13 @@ class T_func_inverse(unittest.TestCase):
self.assert_func_pair_optimized(T.arctanh, T.tanh, dx) self.assert_func_pair_optimized(T.arctanh, T.tanh, dx)
self.assert_func_pair_optimized(T.inv, T.inv, dx) self.assert_func_pair_optimized(T.inv, T.inv, dx)
self.assert_func_pair_optimized(T.neg, T.neg, dx) self.assert_func_pair_optimized(T.neg, T.neg, dx)
cx = dx + complex(0, 1)*(dx + 0.01) cx = dx + complex(0, 1) * (dx + 0.01)
self.assert_func_pair_optimized(T.conj, T.conj, cx, is_complex=True) self.assert_func_pair_optimized(T.conj, T.conj, cx, is_complex=True)
# Test that non-inverse functions are ran normally # Test that non-inverse functions are ran normally
self.assert_func_pair_optimized(T.conj, T.neg, cx, self.assert_func_pair_optimized(T.conj, T.neg, cx,
should_copy=False, is_complex=True) should_copy=False, is_complex=True)
dx = numpy.random.rand(5, 4).astype("float32")+0.01 dx = numpy.random.rand(5, 4).astype("float32") + 0.01
self.assert_func_pair_optimized(T.rad2deg, T.rad2deg, dx, self.assert_func_pair_optimized(T.rad2deg, T.rad2deg, dx,
should_copy=False) should_copy=False)
self.assert_func_pair_optimized(T.rad2deg, T.cosh, dx, self.assert_func_pair_optimized(T.rad2deg, T.cosh, dx,
...@@ -4590,8 +4578,8 @@ def test_constant_get_stabilized(): ...@@ -4590,8 +4578,8 @@ def test_constant_get_stabilized():
f2 = theano.function([x2], y2, mode=mode) f2 = theano.function([x2], y2, mode=mode)
try: try:
assert len(f2.maker.fgraph.toposort()) == 1 assert len(f2.maker.fgraph.toposort()) == 1
assert f2.maker.fgraph.toposort()[0].op == \ assert (f2.maker.fgraph.toposort()[0].op ==
theano.tensor.nnet.sigm.softplus theano.tensor.nnet.sigm.softplus)
assert f2(800) == 800 assert f2(800) == 800
x = T.as_tensor_variable(800) x = T.as_tensor_variable(800)
...@@ -4624,14 +4612,16 @@ class T_local_switch_sink(unittest.TestCase): ...@@ -4624,14 +4612,16 @@ class T_local_switch_sink(unittest.TestCase):
self.xs = 1. self.xs = 1.
# expected results # expected results
self.resm = [numpy.asarray([[1, 0, 1, 0], [0, 0, 0, 0], [1, 1, 1, 1]])]*3 + [numpy.asarray([[1, 0, 1, 0], [1, 0, 1, 0], [1, 0, 1, 0]])] + \ self.resm = (
2*[numpy.asarray([[1, 0, 1, 0]])] + [[numpy.ones((3, 4)), numpy.zeros((3, 4)), numpy.ones((3, 4)), numpy.zeros((3, 4))]] + \ [numpy.asarray([[1, 0, 1, 0], [0, 0, 0, 0], [1, 1, 1, 1]])] * 3 +
[[numpy.ones((4,)), numpy.zeros((4,)), numpy.ones((4,)), numpy.zeros((4,))]] + \ [numpy.asarray([[1, 0, 1, 0], [1, 0, 1, 0], [1, 0, 1, 0]])] +
[[numpy.asarray(1.0), numpy.asarray( 2 * [numpy.asarray([[1, 0, 1, 0]])] +
0.0), numpy.asarray(1.0), numpy.asarray(0.0)]] [[numpy.ones((3, 4)), numpy.zeros((3, 4)), numpy.ones((3, 4)), numpy.zeros((3, 4))]] +
[[numpy.ones((4,)), numpy.zeros((4,)), numpy.ones((4,)), numpy.zeros((4,))]] +
[[numpy.asarray(1.0), numpy.asarray(0.0), numpy.asarray(1.0), numpy.asarray(0.0)]])
self.mode = theano.compile.mode.get_default_mode().including( self.mode = theano.compile.mode.get_default_mode().including(
'canonicalize', 'fast_run').excluding('gpu', 'fusion') 'canonicalize', 'fast_run').excluding('gpu', 'fusion')
self.mode = copy.copy(self.mode) self.mode = copy.copy(self.mode)
self.mode.check_isfinite = False self.mode.check_isfinite = False
...@@ -4640,7 +4630,7 @@ class T_local_switch_sink(unittest.TestCase): ...@@ -4640,7 +4630,7 @@ class T_local_switch_sink(unittest.TestCase):
It disables checking It disables checking
for NaN removed by optimizations in DebugMode (it has false for NaN removed by optimizations in DebugMode (it has false
positives in that case. positives in that case).
""" """
f = theano.function(*args, **kwargs) f = theano.function(*args, **kwargs)
...@@ -4677,8 +4667,8 @@ class T_local_switch_sink(unittest.TestCase): ...@@ -4677,8 +4667,8 @@ class T_local_switch_sink(unittest.TestCase):
self.resm[idx][i])).sum() == self.resm[idx][i].size self.resm[idx][i])).sum() == self.resm[idx][i].size
else: else:
res = f(condition[1], x[1], -1) res = f(condition[1], x[1], -1)
assert (res == numpy.asarray(self. assert ((res == numpy.asarray(self.resm[idx])).sum() ==
resm[idx])).sum() == self.resm[idx].size self.resm[idx].size)
idx += 1 idx += 1
# This case caused a missed optimization in the past. # This case caused a missed optimization in the past.
...@@ -4693,33 +4683,34 @@ class T_local_switch_sink(unittest.TestCase): ...@@ -4693,33 +4683,34 @@ class T_local_switch_sink(unittest.TestCase):
idx = 0 idx = 0
for condition in [(T.dmatrix('cond'), self.condm), (T.dvector('cond'), self.condv), (T.dscalar('cond'), self.conds)]: for condition in [(T.dmatrix('cond'), self.condm), (T.dvector('cond'), self.condv), (T.dscalar('cond'), self.conds)]:
for x in [(T.dmatrix('x'), self.xm), (T.dvector('x'), self.xv), (T.dscalar('x'), self.xs)]: for x in [(T.dmatrix('x'), self.xm), (T.dvector('x'), self.xv), (T.dscalar('x'), self.xs)]:
y = T.true_div(T.switch(condition[0] > 0, 1. * y = T.true_div(
x[0], 0.*x[0]), T.switch(condition[0] > 0, 1.*x[0], T.log(c)*x[0])) T.switch(condition[0] > 0, 1. * x[0], 0. * x[0]),
T.switch(condition[0] > 0, 1. * x[0], T.log(c) * x[0]))
f = self.function_remove_nan([condition[0], x[0], c], f = self.function_remove_nan([condition[0], x[0], c],
[y], mode=self.mode) [y], mode=self.mode)
if type(condition[1]) is list: if type(condition[1]) is list:
for i in xrange(len(condition[1])): for i in xrange(len(condition[1])):
res = f(condition[1][i], x[1], -1) res = f(condition[1][i], x[1], -1)
assert (res == numpy. assert ((res == numpy.asarray(self.resm[idx][i])).sum() ==
asarray(self.resm[idx][i])).sum() == self.resm[idx][i].size self.resm[idx][i].size)
else: else:
res = f(condition[1], x[1], -1) res = f(condition[1], x[1], -1)
assert (res == numpy.asarray(self. assert ((res == numpy.asarray(self.resm[idx])).sum() ==
resm[idx])).sum() == self.resm[idx].size self.resm[idx].size)
idx += 1 idx += 1
class T_local_erf(unittest.TestCase): class T_local_erf(unittest.TestCase):
def setUp(self): def setUp(self):
self.mode = theano.compile.mode.get_default_mode().including( self.mode = theano.compile.mode.get_default_mode().including(
'canonicalize', 'fast_run').excluding('gpu', 'fusion') 'canonicalize', 'fast_run').excluding('gpu', 'fusion')
self.mode._optimizer.position_cutoff = 1.50001 self.mode._optimizer.position_cutoff = 1.50001
if theano.config.cxx == '' and not theano.scalar.basic_scipy.imported_scipy_special: if theano.config.cxx == '' and not theano.scalar.basic_scipy.imported_scipy_special:
raise SkipTest("erf need a c++ compiler or scipy") raise SkipTest("erf need a c++ compiler or scipy")
def test_local_one_plus_erf(self): def test_local_one_plus_erf(self):
val = numpy.asarray([-30, -3, -2, -1, 0, 1, 2, 3, 30], val = numpy.asarray([-30, -3, -2, -1, 0, 1, 2, 3, 30],
dtype=config.floatX) dtype=config.floatX)
x = T.vector() x = T.vector()
f = theano.function([x], 1 + T.erf(x), mode=self.mode) f = theano.function([x], 1 + T.erf(x), mode=self.mode)
...@@ -4746,18 +4737,18 @@ class T_local_erf(unittest.TestCase): ...@@ -4746,18 +4737,18 @@ class T_local_erf(unittest.TestCase):
x = T.vector() x = T.vector()
f = theano.function([x], 1 - T.erf(x), mode=self.mode) f = theano.function([x], 1 - T.erf(x), mode=self.mode)
assert [n.op for n in f.maker.fgraph.toposort()] == [T.erfc]\ assert [n.op for n in f.maker.fgraph.toposort()] == [T.erfc],\
, f.maker.fgraph.toposort() f.maker.fgraph.toposort()
print(f(val)) print(f(val))
f = theano.function([x], 1 + (-T.erf(x)), mode=self.mode) f = theano.function([x], 1 + (-T.erf(x)), mode=self.mode)
assert [n.op for n in f.maker.fgraph.toposort()] == [T.erfc]\ assert [n.op for n in f.maker.fgraph.toposort()] == [T.erfc],\
, f.maker.fgraph.toposort() f.maker.fgraph.toposort()
print(f(val)) print(f(val))
f = theano.function([x], (-T.erf(x)) + 1, mode=self.mode) f = theano.function([x], (-T.erf(x)) + 1, mode=self.mode)
assert [n.op for n in f.maker.fgraph.toposort()] == [T.erfc]\ assert [n.op for n in f.maker.fgraph.toposort()] == [T.erfc],\
, f.maker.fgraph.toposort() f.maker.fgraph.toposort()
print(f(val)) print(f(val))
f = theano.function([x], 2 - T.erf(x), mode=self.mode) f = theano.function([x], 2 - T.erf(x), mode=self.mode)
...@@ -4771,7 +4762,7 @@ class T_local_erf(unittest.TestCase): ...@@ -4771,7 +4762,7 @@ class T_local_erf(unittest.TestCase):
def test_local_erf_minus_one(self): def test_local_erf_minus_one(self):
val = numpy.asarray([-30, -3, -2, -1, 0, 1, 2, 3, 30], val = numpy.asarray([-30, -3, -2, -1, 0, 1, 2, 3, 30],
dtype=config.floatX) dtype=config.floatX)
x = T.vector() x = T.vector()
f = theano.function([x], T.erf(x) - 1, mode=self.mode) f = theano.function([x], T.erf(x) - 1, mode=self.mode)
...@@ -4803,24 +4794,24 @@ class T_local_erfc(unittest.TestCase): ...@@ -4803,24 +4794,24 @@ class T_local_erfc(unittest.TestCase):
self.mode = self.mode_fusion.excluding('fusion') self.mode = self.mode_fusion.excluding('fusion')
self.mode._optimizer.position_cutoff = 1.50001 self.mode._optimizer.position_cutoff = 1.50001
if (theano.config.cxx == '' and if (theano.config.cxx == '' and
not theano.scalar.basic_scipy.imported_scipy_special): not theano.scalar.basic_scipy.imported_scipy_special):
raise SkipTest("erfc need a c++ compiler or scipy") raise SkipTest("erfc need a c++ compiler or scipy")
def test_local_one_minus_erfc(self): def test_local_one_minus_erfc(self):
""" test opt: 1-erfc(x) => erf(x) and -erfc(x)+1 => erf(x) """ test opt: 1-erfc(x) => erf(x) and -erfc(x)+1 => erf(x)
""" """
val = numpy.asarray([-30, -3, -2, -1, 0, 1, 2, 3, 30], val = numpy.asarray([-30, -3, -2, -1, 0, 1, 2, 3, 30],
dtype=config.floatX) dtype=config.floatX)
x = T.vector('x') x = T.vector('x')
f = theano.function([x], 1 - T.erfc(x), mode=self.mode) f = theano.function([x], 1 - T.erfc(x), mode=self.mode)
assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf]\ assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf],\
, f.maker.fgraph.toposort() f.maker.fgraph.toposort()
print(f(val)) print(f(val))
f = theano.function([x], (-T.erfc(x)) + 1, mode=self.mode) f = theano.function([x], (-T.erfc(x)) + 1, mode=self.mode)
assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf]\ assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf],\
, f.maker.fgraph.toposort() f.maker.fgraph.toposort()
print(f(val)) print(f(val))
f = theano.function([x], 2 - T.erfc(x), mode=self.mode) f = theano.function([x], 2 - T.erfc(x), mode=self.mode)
...@@ -4828,34 +4819,34 @@ class T_local_erfc(unittest.TestCase): ...@@ -4828,34 +4819,34 @@ class T_local_erfc(unittest.TestCase):
assert len(topo) == 2, f.maker.fgraph.toposort() assert len(topo) == 2, f.maker.fgraph.toposort()
assert topo[0].op == T.erfc, f.maker.fgraph.toposort() assert topo[0].op == T.erfc, f.maker.fgraph.toposort()
assert isinstance(topo[1].op, T.Elemwise), f.maker.fgraph.toposort() assert isinstance(topo[1].op, T.Elemwise), f.maker.fgraph.toposort()
assert isinstance(topo[1].op.scalar_op, scal.Sub)\ assert isinstance(topo[1].op.scalar_op, scal.Sub),\
, f.maker.fgraph.toposort() f.maker.fgraph.toposort()
print(f(val)) print(f(val))
def test_local_erf_neg_minus_one(self): def test_local_erf_neg_minus_one(self):
""" test opt: (-1)+erfc(-x)=>erf(x)""" """ test opt: (-1)+erfc(-x)=>erf(x)"""
val = numpy.asarray([-30, -3, -2, -1, 0, 1, 2, 3, 30], val = numpy.asarray([-30, -3, -2, -1, 0, 1, 2, 3, 30],
dtype=config.floatX) dtype=config.floatX)
x = T.vector('x') x = T.vector('x')
f = theano.function([x], -1 + T.erfc(-x), mode=self.mode) f = theano.function([x], -1 + T.erfc(-x), mode=self.mode)
assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf]\ assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf],\
, f.maker.fgraph.toposort() f.maker.fgraph.toposort()
print(f(val)) print(f(val))
f = theano.function([x], T.erfc(-x) - 1, mode=self.mode) f = theano.function([x], T.erfc(-x) - 1, mode=self.mode)
assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf]\ assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf],\
, f.maker.fgraph.toposort() f.maker.fgraph.toposort()
print(f(val)) print(f(val))
f = theano.function([x], T.erfc(-x) + (-1), mode=self.mode) f = theano.function([x], T.erfc(-x) + (-1), mode=self.mode)
assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf]\ assert [n.op for n in f.maker.fgraph.toposort()] == [T.erf],\
, f.maker.fgraph.toposort() f.maker.fgraph.toposort()
print(f(val)) print(f(val))
def test_local_log_erfc(self): def test_local_log_erfc(self):
val = [-30, -27, -26, -11, -10, -3, -2, -1, 0, 1, 2, 3, 10, val = [-30, -27, -26, -11, -10, -3, -2, -1, 0, 1, 2, 3, 10,
11, 26, 27, 28, 30] 11, 26, 27, 28, 30]
if theano.config.mode in ["DebugMode", "DEBUG_MODE", "FAST_COMPILE"]: if theano.config.mode in ["DebugMode", "DEBUG_MODE", "FAST_COMPILE"]:
# python mode don't like the inv(0) # python mode don't like the inv(0)
val.remove(0) val.remove(0)
...@@ -4894,9 +4885,11 @@ class T_local_erfc(unittest.TestCase): ...@@ -4894,9 +4885,11 @@ class T_local_erfc(unittest.TestCase):
def test_local_grad_log_erfc_neg(self): def test_local_grad_log_erfc_neg(self):
val = [-100, -30, -27, -26.4, -26.2, -26, -11, -10, -9, -3, -2, -1, 0, val = [-100, -30, -27, -26.4, -26.2, -26, -11, -10, -9, -3, -2, -1, 0,
1, 2, 3, 9, 10, 11, 27, 26.4, 26.2, 26, 28, 30, 100] 1, 2, 3, 9, 10, 11, 27, 26.4, 26.2, 26, 28, 30, 100]
if theano.config.mode in ["DebugMode", "DEBUG_MODE", "FAST_COMPILE"]: if theano.config.mode in ["DebugMode", "DEBUG_MODE", "FAST_COMPILE"]:
# python mode don't like the inv(0) in computation, but the switch don't select this value. So it is computed for no good reason. # python mode don't like the inv(0) in computation,
# but the switch don't select this value.
# So it is computed for no good reason.
val.remove(0) val.remove(0)
if theano.config.mode in ["DebugMode", "DEBUG_MODE"] and theano.config.floatX == 'float32': if theano.config.mode in ["DebugMode", "DEBUG_MODE"] and theano.config.floatX == 'float32':
# In float32 their is a plage of values close to 10 that we stabilize as it give bigger error then the stabilized version. # In float32 their is a plage of values close to 10 that we stabilize as it give bigger error then the stabilized version.
...@@ -4918,8 +4911,10 @@ class T_local_erfc(unittest.TestCase): ...@@ -4918,8 +4911,10 @@ class T_local_erfc(unittest.TestCase):
assert f.maker.fgraph.outputs[0].dtype == theano.config.floatX assert f.maker.fgraph.outputs[0].dtype == theano.config.floatX
# test with a different mul constant # test with a different mul constant
f = theano.function([x], T.mul(T.exp(T.neg(T.sqr(x))), - f = theano.function(
10.12837917) / T.erfc(x), mode=mode) [x],
T.mul(T.exp(T.neg(T.sqr(x))), - 10.12837917) / T.erfc(x),
mode=mode)
assert len(f.maker.fgraph.apply_nodes) == 23, len(f.maker.fgraph.apply_nodes) assert len(f.maker.fgraph.apply_nodes) == 23, len(f.maker.fgraph.apply_nodes)
assert f.maker.fgraph.outputs[0].dtype == theano.config.floatX assert f.maker.fgraph.outputs[0].dtype == theano.config.floatX
assert all(numpy.isfinite(f(val))) assert all(numpy.isfinite(f(val)))
...@@ -4944,14 +4939,12 @@ class T_local_erfc(unittest.TestCase): ...@@ -4944,14 +4939,12 @@ class T_local_erfc(unittest.TestCase):
assert all(numpy.isfinite(f(val))) assert all(numpy.isfinite(f(val)))
# test that it work correctly if x is x*2 in the graph. # test that it work correctly if x is x*2 in the graph.
f = theano.function([x], T.grad(T.log(T.erfc(2 * x)).sum(), f = theano.function([x], T.grad(T.log(T.erfc(2 * x)).sum(), x), mode=mode)
x), mode=mode)
assert len(f.maker.fgraph.apply_nodes) == 23, len(f.maker.fgraph.apply_nodes) assert len(f.maker.fgraph.apply_nodes) == 23, len(f.maker.fgraph.apply_nodes)
assert numpy.isfinite(f(val)).all() assert numpy.isfinite(f(val)).all()
assert f.maker.fgraph.outputs[0].dtype == theano.config.floatX assert f.maker.fgraph.outputs[0].dtype == theano.config.floatX
f = theano.function([x], T.grad(T.log(T.erfc(x)).sum(), x), f = theano.function([x], T.grad(T.log(T.erfc(x)).sum(), x), mode=mode_fusion)
mode=mode_fusion)
assert len(f.maker.fgraph.apply_nodes) == 1, len(f.maker.fgraph.apply_nodes) assert len(f.maker.fgraph.apply_nodes) == 1, len(f.maker.fgraph.apply_nodes)
assert f.maker.fgraph.outputs[0].dtype == theano.config.floatX assert f.maker.fgraph.outputs[0].dtype == theano.config.floatX
...@@ -4973,8 +4966,8 @@ class T_local_erfc(unittest.TestCase): ...@@ -4973,8 +4966,8 @@ class T_local_erfc(unittest.TestCase):
val = numpy.random.rand(1e6) val = numpy.random.rand(1e6)
x = T.vector() x = T.vector()
mode = theano.compile.mode.get_mode("FAST_RUN") mode = theano.compile.mode.get_mode("FAST_RUN")
f1 = theano.function([x], T.log(T.erfc(x)), mode=mode. f1 = theano.function([x], T.log(T.erfc(x)),
excluding("local_log_erfc")) mode=mode.excluding("local_log_erfc"))
f2 = theano.function([x], T.log(T.erfc(x)), mode=mode) f2 = theano.function([x], T.log(T.erfc(x)), mode=mode)
print(f1.maker.fgraph.toposort()) print(f1.maker.fgraph.toposort())
print(f2.maker.fgraph.toposort()) print(f2.maker.fgraph.toposort())
...@@ -4999,9 +4992,9 @@ class test_local_useless_switch(unittest.TestCase): ...@@ -4999,9 +4992,9 @@ class test_local_useless_switch(unittest.TestCase):
z = theano.tensor.switch(0, x, y) z = theano.tensor.switch(0, x, y)
f = theano.function([x, y], z, mode=self.mode) f = theano.function([x, y], z, mode=self.mode)
assert len([node.op for node in f.maker.fgraph.toposort() if assert len([node.op for node in f.maker.fgraph.toposort() if
(isinstance(node.op, theano.tensor.Elemwise) (isinstance(node.op, theano.tensor.Elemwise) and
and isinstance(node.op.scalar_op, isinstance(node.op.scalar_op,
theano.scalar.basic.Switch))]) == 0 theano.scalar.basic.Switch))]) == 0
vx = numpy.array([[1, 2, 3], [4, 5, 6]], dtype=dtype1) vx = numpy.array([[1, 2, 3], [4, 5, 6]], dtype=dtype1)
vy = numpy.array([[7, 8, 9], [10, 11, 12]], dtype=dtype2) vy = numpy.array([[7, 8, 9], [10, 11, 12]], dtype=dtype2)
assert numpy.all(f(vx, vy) == vy) assert numpy.all(f(vx, vy) == vy)
...@@ -5015,9 +5008,9 @@ class test_local_useless_switch(unittest.TestCase): ...@@ -5015,9 +5008,9 @@ class test_local_useless_switch(unittest.TestCase):
z = theano.tensor.switch(1, x, y) z = theano.tensor.switch(1, x, y)
f = theano.function([x, y], z, mode=self.mode) f = theano.function([x, y], z, mode=self.mode)
assert len([node.op for node in f.maker.fgraph.toposort() if assert len([node.op for node in f.maker.fgraph.toposort() if
(isinstance(node.op, theano.tensor.Elemwise) (isinstance(node.op, theano.tensor.Elemwise) and
and isinstance(node.op.scalar_op, isinstance(node.op.scalar_op,
theano.scalar.basic.Switch))]) == 0 theano.scalar.basic.Switch))]) == 0
vx = numpy.array([[1, 2, 3], [4, 5, 6]], dtype=dtype1) vx = numpy.array([[1, 2, 3], [4, 5, 6]], dtype=dtype1)
vy = numpy.array([[7, 8, 9], [10, 11, 12]], dtype=dtype2) vy = numpy.array([[7, 8, 9], [10, 11, 12]], dtype=dtype2)
assert numpy.all(f(vx, vy) == vx) assert numpy.all(f(vx, vy) == vx)
...@@ -5032,7 +5025,7 @@ class test_local_useless_switch(unittest.TestCase): ...@@ -5032,7 +5025,7 @@ class test_local_useless_switch(unittest.TestCase):
z2 = theano.tensor.switch(varc, x, x) z2 = theano.tensor.switch(varc, x, x)
f1 = theano.function([x], z1, mode=self.mode) f1 = theano.function([x], z1, mode=self.mode)
f0 = theano.function([x], z0, mode=self.mode) f0 = theano.function([x], z0, mode=self.mode)
f2 = theano.function([x,varc], z2, mode=self.mode) f2 = theano.function([x, varc], z2, mode=self.mode)
topo = f1.maker.fgraph.toposort() topo = f1.maker.fgraph.toposort()
assert len(topo) == 1 assert len(topo) == 1
...@@ -5050,7 +5043,7 @@ class test_local_useless_switch(unittest.TestCase): ...@@ -5050,7 +5043,7 @@ class test_local_useless_switch(unittest.TestCase):
vc = numpy.array([[1, 2, 3], [4, 5, 6]], dtype=dtype1) vc = numpy.array([[1, 2, 3], [4, 5, 6]], dtype=dtype1)
assert numpy.all(f1(vx) == vx) assert numpy.all(f1(vx) == vx)
assert numpy.all(f0(vx) == vx) assert numpy.all(f0(vx) == vx)
assert numpy.all(f2(vx,vc) == vx) assert numpy.all(f2(vx, vc) == vx)
def test_shape_le_0(self): def test_shape_le_0(self):
...@@ -5064,11 +5057,10 @@ class test_local_useless_switch(unittest.TestCase): ...@@ -5064,11 +5057,10 @@ class test_local_useless_switch(unittest.TestCase):
f1 = theano.function([x], z1, mode=self.mode) f1 = theano.function([x], z1, mode=self.mode)
assert isinstance(f1.maker.fgraph.toposort()[0].op, Shape_i) assert isinstance(f1.maker.fgraph.toposort()[0].op, Shape_i)
vx = numpy.random.randn(0,5).astype(dtype1) vx = numpy.random.randn(0, 5).astype(dtype1)
assert f0(vx) == 0 assert f0(vx) == 0
assert f1(vx) == 5 assert f1(vx) == 5
def test_broadcast1(self): def test_broadcast1(self):
# test switch(cst, matrix, row) # test switch(cst, matrix, row)
x = theano.tensor.matrix('x', dtype='int32') x = theano.tensor.matrix('x', dtype='int32')
...@@ -5197,7 +5189,7 @@ class T_local_sum_prod(unittest.TestCase): ...@@ -5197,7 +5189,7 @@ class T_local_sum_prod(unittest.TestCase):
# ensuring that the optimized graph contains the expected number # ensuring that the optimized graph contains the expected number
# of apply nodes for the sum op # of apply nodes for the sum op
prod_nodes = [n for n in f.maker.fgraph.toposort() prod_nodes = [n for n in f.maker.fgraph.toposort()
if isinstance(n.op, reduction_op)] if isinstance(n.op, reduction_op)]
assert len(prod_nodes) == nb_expected_sum_nodes assert len(prod_nodes) == nb_expected_sum_nodes
# Test sum # Test sum
...@@ -5391,9 +5383,9 @@ class T_local_sum_prod(unittest.TestCase): ...@@ -5391,9 +5383,9 @@ class T_local_sum_prod(unittest.TestCase):
dtype='float64') dtype='float64')
mode = self.mode.including('specialize').excluding('fusion') mode = self.mode.including('specialize').excluding('fusion')
for t_like, n_like, nb_nodes in [(tensor.zeros_like, numpy.zeros_like, (1, 3, 3, 2)), for t_like, n_like, nb_nodes in [
(tensor.ones_like, numpy.ones_like, (5, 5, 5, 6))]: (tensor.zeros_like, numpy.zeros_like, (1, 3, 3, 2)),
(tensor.ones_like, numpy.ones_like, (5, 5, 5, 6))]:
# test sum # test sum
f = theano.function([a], t_like(a).sum(None), mode=mode) f = theano.function([a], t_like(a).sum(None), mode=mode)
utt.assert_allclose(f(input), n_like(input).sum()) utt.assert_allclose(f(input), n_like(input).sum())
...@@ -5421,23 +5413,23 @@ class T_local_sum_prod(unittest.TestCase): ...@@ -5421,23 +5413,23 @@ class T_local_sum_prod(unittest.TestCase):
# test prod # test prod
f = theano.function([a], t_like(a).prod(None), mode=mode) f = theano.function([a], t_like(a).prod(None), mode=mode)
utt.assert_allclose(f(input), n_like(input).prod()) utt.assert_allclose(f(input), n_like(input).prod())
#assert len(f.maker.fgraph.apply_nodes) == nb_nodes[0] # assert len(f.maker.fgraph.apply_nodes) == nb_nodes[0]
f = theano.function([a], t_like(a).prod([0, 1, 2]), mode=mode) f = theano.function([a], t_like(a).prod([0, 1, 2]), mode=mode)
utt.assert_allclose(f(input), n_like(input).prod()) utt.assert_allclose(f(input), n_like(input).prod())
#assert len(f.maker.fgraph.apply_nodes) == nb_nodes[0] # assert len(f.maker.fgraph.apply_nodes) == nb_nodes[0]
for d in range(3): for d in range(3):
f = theano.function([a], t_like(a).prod(d), mode=mode) f = theano.function([a], t_like(a).prod(d), mode=mode)
utt.assert_allclose(f(input), n_like(input).prod(d)) utt.assert_allclose(f(input), n_like(input).prod(d))
#assert len(f.maker.fgraph.apply_nodes) == nb_nodes[1] # assert len(f.maker.fgraph.apply_nodes) == nb_nodes[1]
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
assert topo[-1].op == T.alloc assert topo[-1].op == T.alloc
assert not any([isinstance(node.op, T.elemwise.Prod) for node in topo]) assert not any([isinstance(node.op, T.elemwise.Prod) for node in topo])
for i in range(3): for i in range(3):
f = theano.function([a], t_like(a).prod(i), mode=mode) f = theano.function([a], t_like(a).prod(i), mode=mode)
utt.assert_allclose(f(input), n_like(input).prod(i)) utt.assert_allclose(f(input), n_like(input).prod(i))
#assert len(f.maker.fgraph.apply_nodes) == nb_nodes[2] # assert len(f.maker.fgraph.apply_nodes) == nb_nodes[2]
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
assert topo[-1].op == T.alloc assert topo[-1].op == T.alloc
assert not any([isinstance(node.op, T.elemwise.Prod) for node in topo]) assert not any([isinstance(node.op, T.elemwise.Prod) for node in topo])
...@@ -5449,7 +5441,7 @@ class T_local_sum_prod(unittest.TestCase): ...@@ -5449,7 +5441,7 @@ class T_local_sum_prod(unittest.TestCase):
f = theano.function([a], t_like(a). f = theano.function([a], t_like(a).
sum(d).sum(dd), mode=mode) sum(d).sum(dd), mode=mode)
utt.assert_allclose(f(input), utt.assert_allclose(f(input),
n_like(input).sum(d).sum(dd)) n_like(input).sum(d).sum(dd))
assert len(f.maker.fgraph.apply_nodes) == nb_nodes[3] assert len(f.maker.fgraph.apply_nodes) == nb_nodes[3]
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
assert topo[-1].op == T.alloc assert topo[-1].op == T.alloc
...@@ -5470,7 +5462,7 @@ class T_local_sum_prod(unittest.TestCase): ...@@ -5470,7 +5462,7 @@ class T_local_sum_prod(unittest.TestCase):
config.on_opt_error = 'raise' config.on_opt_error = 'raise'
try: try:
# This compilation would fail prior to fix. # This compilation would fail prior to fix.
f = theano.function([x], y) theano.function([x], y)
finally: finally:
config.on_opt_error = backup config.on_opt_error = backup
...@@ -5484,7 +5476,7 @@ class T_local_sum_prod(unittest.TestCase): ...@@ -5484,7 +5476,7 @@ class T_local_sum_prod(unittest.TestCase):
config.on_opt_error = 'raise' config.on_opt_error = 'raise'
try: try:
# This compilation would fail prior to fix. # This compilation would fail prior to fix.
f = theano.function([x], y) theano.function([x], y)
finally: finally:
config.on_opt_error = backup config.on_opt_error = backup
...@@ -5698,14 +5690,14 @@ class T_local_sum_prod_dimshuffle(unittest.TestCase): ...@@ -5698,14 +5690,14 @@ class T_local_sum_prod_dimshuffle(unittest.TestCase):
for i, s in enumerate(sums): for i, s in enumerate(sums):
print(i) print(i)
f = theano.function([a, b, c, d], s, mode=self.mode, f = theano.function([a, b, c, d], s, mode=self.mode,
on_unused_input='ignore') on_unused_input='ignore')
g = f.maker.fgraph.toposort() g = f.maker.fgraph.toposort()
assert isinstance(g[-1].op.scalar_op, assert isinstance(g[-1].op.scalar_op,
theano.scalar.basic.TrueDiv) theano.scalar.basic.TrueDiv)
f(a_val, b_val, c_val, d_val) f(a_val, b_val, c_val, d_val)
finally: finally:
config.warn.sum_sum_bug, config.warn.sum_div_dimshuffle_bug =\ config.warn.sum_sum_bug, config.warn.sum_div_dimshuffle_bug =\
backup backup
def test_local_prod_div_dimshuffle(self): def test_local_prod_div_dimshuffle(self):
a = T.matrix('a') a = T.matrix('a')
...@@ -5827,19 +5819,18 @@ class TestMakeVector(utt.InferShapeTester): ...@@ -5827,19 +5819,18 @@ class TestMakeVector(utt.InferShapeTester):
# Should work # Should work
for (dtype, inputs) in [("int8", (b, b)), for (dtype, inputs) in [("int8", (b, b)),
("int32", (i, b)), ("int32", (i, b)),
("int32", (b, i)), ("int32", (b, i)),
("float64", (b, i)), ("float64", (b, i)),
("float64", (b, d)), ("float64", (b, d)),
("float64", (d, i)), ("float64", (d, i)),
("float64", ()), ("float64", ()),
("int64", ()), ("int64", ()),
]: ]:
mv = opt.MakeVector(dtype=dtype)(*inputs) mv = opt.MakeVector(dtype=dtype)(*inputs)
assert mv.dtype == dtype assert mv.dtype == dtype
f = theano.function([b, i, d], mv, on_unused_input='ignore') f = theano.function([b, i, d], mv, on_unused_input='ignore')
f_val = f(val[b], val[i], val[d]) f(val[b], val[i], val[d])
# print 'f_val =', f_val
s = mv.sum() s = mv.sum()
gb = T.grad(s, b, disconnected_inputs='ignore') gb = T.grad(s, b, disconnected_inputs='ignore')
...@@ -5888,13 +5879,13 @@ class TestMakeVector(utt.InferShapeTester): ...@@ -5888,13 +5879,13 @@ class TestMakeVector(utt.InferShapeTester):
# should fail # should fail
for (dtype, inputs) in [("int8", (b, i)), for (dtype, inputs) in [("int8", (b, i)),
("int8", (i, b)), ("int8", (i, b)),
("int8", (b, d)), ("int8", (b, d)),
("int8", (i, i)), ("int8", (i, i)),
("int32", (d, i)), ("int32", (d, i)),
("int32", (i, d)), ("int32", (i, d)),
("float32", (i, d)), ("float32", (i, d)),
]: ]:
try: try:
opt.MakeVector(dtype=dtype)(*inputs) opt.MakeVector(dtype=dtype)(*inputs)
raise Exception("Theano should have raised an error") raise Exception("Theano should have raised an error")
...@@ -5915,17 +5906,17 @@ class TestMakeVector(utt.InferShapeTester): ...@@ -5915,17 +5906,17 @@ class TestMakeVector(utt.InferShapeTester):
ciscal_val = numpy.random.randint(10) ciscal_val = numpy.random.randint(10)
discal_val = numpy.random.randint(10) discal_val = numpy.random.randint(10)
self._compile_and_check([adscal, aiscal], self._compile_and_check([adscal, aiscal],
[MakeVector('float64')(adscal, aiscal)], [MakeVector('float64')(adscal, aiscal)],
[adscal_val, aiscal_val], MakeVector) [adscal_val, aiscal_val], MakeVector)
self._compile_and_check([adscal, bdscal, aiscal], self._compile_and_check([adscal, bdscal, aiscal],
[MakeVector('float64')(adscal, bdscal, aiscal)], [MakeVector('float64')(adscal, bdscal, aiscal)],
[adscal_val, bdscal_val, aiscal_val], MakeVector) [adscal_val, bdscal_val, aiscal_val], MakeVector)
self._compile_and_check([aiscal, biscal, ciscal, discal], self._compile_and_check([aiscal, biscal, ciscal, discal],
[MakeVector('int32')(aiscal, biscal, ciscal, discal)], [MakeVector('int32')(aiscal, biscal, ciscal, discal)],
[aiscal_val, biscal_val, ciscal_val, discal_val], [aiscal_val, biscal_val, ciscal_val, discal_val],
MakeVector) MakeVector)
def test_local_join_1(): def test_local_join_1():
...@@ -5982,7 +5973,6 @@ def test_local_join_empty(): ...@@ -5982,7 +5973,6 @@ def test_local_join_empty():
for n in e if isinstance(n.op, Join)]) for n in e if isinstance(n.op, Join)])
assert f.maker.fgraph.outputs[0].dtype == config.floatX assert f.maker.fgraph.outputs[0].dtype == config.floatX
# test for matrix join(1,a) # test for matrix join(1,a)
empty_mat = numpy.asarray([[]], dtype=config.floatX) empty_mat = numpy.asarray([[]], dtype=config.floatX)
m = tensor.matrix('m') m = tensor.matrix('m')
...@@ -6059,10 +6049,10 @@ def test_local_add_specialize(): ...@@ -6059,10 +6049,10 @@ def test_local_add_specialize():
def test_local_tensor_scalar_tensor(): def test_local_tensor_scalar_tensor():
dtypes = ['int8', 'int16', 'int32', 'int64', dtypes = ['int8', 'int16', 'int32', 'int64',
'uint8', 'uint16', 'uint32', 'uint64', 'uint8', 'uint16', 'uint32', 'uint64',
'float32', 'float64', 'float32', 'float64',
'complex64', 'complex128' 'complex64', 'complex128'
] ]
for dtype in dtypes: for dtype in dtypes:
t_type = TensorType(dtype=dtype, broadcastable=()) t_type = TensorType(dtype=dtype, broadcastable=())
...@@ -6073,18 +6063,18 @@ def test_local_tensor_scalar_tensor(): ...@@ -6073,18 +6063,18 @@ def test_local_tensor_scalar_tensor():
f = function([t], t2, mode=mode_opt) f = function([t], t2, mode=mode_opt)
e = f.maker.fgraph.toposort() e = f.maker.fgraph.toposort()
cast_nodes = [n for n in e cast_nodes = [n for n in e
if isinstance(n.op, (tensor.TensorFromScalar, if isinstance(n.op, (tensor.TensorFromScalar,
tensor.ScalarFromTensor))] tensor.ScalarFromTensor))]
assert len(cast_nodes) == 0 assert len(cast_nodes) == 0
f(0) f(0)
def test_local_scalar_tensor_scalar(): def test_local_scalar_tensor_scalar():
dtypes = ['int8', 'int16', 'int32', 'int64', dtypes = ['int8', 'int16', 'int32', 'int64',
'uint8', 'uint16', 'uint32', 'uint64', 'uint8', 'uint16', 'uint32', 'uint64',
'float32', 'float64', 'float32', 'float64',
'complex64', 'complex128' 'complex64', 'complex128'
] ]
for dtype in dtypes: for dtype in dtypes:
s_type = theano.scalar.Scalar(dtype=dtype) s_type = theano.scalar.Scalar(dtype=dtype)
...@@ -6095,8 +6085,8 @@ def test_local_scalar_tensor_scalar(): ...@@ -6095,8 +6085,8 @@ def test_local_scalar_tensor_scalar():
f = function([s], s2, mode=mode_opt) f = function([s], s2, mode=mode_opt)
e = f.maker.fgraph.toposort() e = f.maker.fgraph.toposort()
cast_nodes = [n for n in e cast_nodes = [n for n in e
if isinstance(n.op, (tensor.TensorFromScalar, if isinstance(n.op, (tensor.TensorFromScalar,
tensor.ScalarFromTensor))] tensor.ScalarFromTensor))]
assert len(cast_nodes) == 0 assert len(cast_nodes) == 0
f(0) f(0)
...@@ -6127,13 +6117,13 @@ def test_local_useless_split(): ...@@ -6127,13 +6117,13 @@ def test_local_useless_split():
f_opt = theano.function([x, splits], opt, mode=mode) f_opt = theano.function([x, splits], opt, mode=mode)
f_nonopt = theano.function([x, splits], nonopt, mode=mode) f_nonopt = theano.function([x, splits], nonopt, mode=mode)
f_opt(numpy.random.rand(4,4).astype(config.floatX), [4]) f_opt(numpy.random.rand(4, 4).astype(config.floatX), [4])
f_nonopt(numpy.random.rand(4,4).astype(config.floatX), [1,2,1]) f_nonopt(numpy.random.rand(4, 4).astype(config.floatX), [1, 2, 1])
graph_opt = f_opt.maker.fgraph.toposort() graph_opt = f_opt.maker.fgraph.toposort()
graph_nonopt = f_nonopt.maker.fgraph.toposort() graph_nonopt = f_nonopt.maker.fgraph.toposort()
assert isinstance(graph_opt[-1].op, DeepCopyOp) assert isinstance(graph_opt[-1].op, DeepCopyOp)
assert len(graph_nonopt)==1 assert len(graph_nonopt) == 1
assert isinstance(graph_nonopt[0].op, tensor.Split) assert isinstance(graph_nonopt[0].op, tensor.Split)
assert check_stack_trace(f_opt, ops_to_check=[Assert]) assert check_stack_trace(f_opt, ops_to_check=[Assert])
...@@ -6151,12 +6141,12 @@ def test_local_flatten_lift(): ...@@ -6151,12 +6141,12 @@ def test_local_flatten_lift():
x_np = numpy.random.rand(5, 4, 3, 2).astype(config.floatX) x_np = numpy.random.rand(5, 4, 3, 2).astype(config.floatX)
out_np = f(x_np) out_np = f(x_np)
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
shape_out_np = tuple(x_np.shape[:i-1])+(numpy.prod(x_np.shape[i-1:]),) shape_out_np = tuple(x_np.shape[:i - 1]) + (numpy.prod(x_np.shape[i - 1:]),)
assert shape_out_np == out_np.shape assert shape_out_np == out_np.shape
reshape_nodes = [n for n in topo if isinstance(n.op, tensor.Reshape)] reshape_nodes = [n for n in topo if isinstance(n.op, tensor.Reshape)]
assert (len(reshape_nodes) == 1 and assert (len(reshape_nodes) == 1 and
tensor.is_flat(reshape_nodes[0].outputs[0], outdim=i)) tensor.is_flat(reshape_nodes[0].outputs[0], outdim=i))
assert isinstance(topo[-1].op, tensor.Elemwise) assert isinstance(topo[-1].op, tensor.Elemwise)
...@@ -6182,7 +6172,7 @@ class Test_local_useless_reshape(unittest.TestCase): ...@@ -6182,7 +6172,7 @@ class Test_local_useless_reshape(unittest.TestCase):
mode = theano.compile.get_default_mode().including( mode = theano.compile.get_default_mode().including(
'local_useless_reshape') 'local_useless_reshape')
i = T.iscalar('i') i = T.iscalar('i')
m = theano.tensor.mgrid[0:i,] m = theano.tensor.mgrid[0:i, ]
f = theano.function([i], m, mode=mode) f = theano.function([i], m, mode=mode)
topo = f.maker.fgraph.toposort() topo = f.maker.fgraph.toposort()
assert not any(isinstance(n.op, tensor.basic.Reshape) for n in topo) assert not any(isinstance(n.op, tensor.basic.Reshape) for n in topo)
...@@ -6343,10 +6333,10 @@ class TestShape_i(utt.InferShapeTester): ...@@ -6343,10 +6333,10 @@ class TestShape_i(utt.InferShapeTester):
admat = matrix() admat = matrix()
admat_val = numpy.random.rand(3, 4).astype(config.floatX) admat_val = numpy.random.rand(3, 4).astype(config.floatX)
self._compile_and_check([admat], [Shape_i(0)(admat)], self._compile_and_check([admat], [Shape_i(0)(admat)],
[admat_val], Shape_i) [admat_val], Shape_i)
self._compile_and_check([admat], [Shape_i(1)(admat)], self._compile_and_check([admat], [Shape_i(1)(admat)],
[admat_val], Shape_i) [admat_val], Shape_i)
class TestShapeFeature(unittest.TestCase): class TestShapeFeature(unittest.TestCase):
...@@ -6435,7 +6425,7 @@ class TestIntDivByOne(unittest.TestCase): ...@@ -6435,7 +6425,7 @@ class TestIntDivByOne(unittest.TestCase):
"""Simple test case for removing dividing by 1""" """Simple test case for removing dividing by 1"""
y = T.tensor4('y') y = T.tensor4('y')
z = y // 1 z = y // 1
f = theano.function([y], z, mode = self.mode) f = theano.function([y], z, mode=self.mode)
graph = f.maker.fgraph.toposort() graph = f.maker.fgraph.toposort()
divs = [node for node in graph divs = [node for node in graph
if isinstance(node.op, T.elemwise.Elemwise) and if isinstance(node.op, T.elemwise.Elemwise) and
...@@ -6445,7 +6435,7 @@ class TestIntDivByOne(unittest.TestCase): ...@@ -6445,7 +6435,7 @@ class TestIntDivByOne(unittest.TestCase):
def test3(self): def test3(self):
"""Simple test case for removing dividing by a tensor of ones""" """Simple test case for removing dividing by a tensor of ones"""
y = T.tensor4('y') y = T.tensor4('y')
z = y // numpy.ones((2,2,2,2)) z = y // numpy.ones((2, 2, 2, 2))
f = theano.function([y], z, mode=self.mode) f = theano.function([y], z, mode=self.mode)
graph = f.maker.fgraph.toposort() graph = f.maker.fgraph.toposort()
divs = [node for node in graph divs = [node for node in graph
...@@ -6456,7 +6446,6 @@ class TestIntDivByOne(unittest.TestCase): ...@@ -6456,7 +6446,6 @@ class TestIntDivByOne(unittest.TestCase):
def test_local_zero_div(): def test_local_zero_div():
"""Tests 0/x -> 0""" """Tests 0/x -> 0"""
mode = theano.compile.mode.get_default_mode().including("local_zero_div")
for t in (T.scalar, T.ivector, T.ftensor4): for t in (T.scalar, T.ivector, T.ftensor4):
x = t('x') x = t('x')
for op in (T.int_div, T.true_div): for op in (T.int_div, T.true_div):
......
Markdown 格式
0% 或
您添加了 0 人 到此讨论。请谨慎行事。
请先完成此评论的编辑！
请 注册 或者 后发表评论