Operators can no longer be chained by '*'; you need to use '.chain()' or '@' (in Python3)

demos/Wiener_Filter.ipynb

@@ -429,7 +429,7 @@
     "mask[l:h] = 0\n",
     "mask = ift.Field.from_global_data(s_space, mask)\n",
     "\n",
-    "R = ift.DiagonalOperator(mask)*HT\n",
+    "R = ift.DiagonalOperator(mask).chain(HT)\n",
     "n = n.to_global_data().copy()\n",
     "n[l:h] = 0\n",
     "n = ift.Field.from_global_data(s_space, n)\n",
@@ -585,7 +585,7 @@
     "mask[l:h,l:h] = 0.\n",
     "mask = ift.Field.from_global_data(s_space, mask)\n",
     "\n",
-    "R = ift.DiagonalOperator(mask)*HT\n",
+    "R = ift.DiagonalOperator(mask).chain(HT)\n",
     "n = n.to_global_data().copy()\n",
     "n[l:h, l:h] = 0\n",
     "n = ift.Field.from_global_data(s_space, n)\n",

demos/getting_started_1.py

@@ -78,7 +78,7 @@ if __name__ == '__main__':
     GR = ift.GeometryRemover(position_space)
     mask = ift.Field.from_global_data(position_space, mask)
     Mask = ift.DiagonalOperator(mask)
-    R = GR * Mask * HT
+    R = GR.chain(Mask).chain(HT)
 
     data_space = GR.target
 
@@ -93,7 +93,7 @@ if __name__ == '__main__':
 
     # Build propagator D and information source j
     j = R.adjoint_times(N.inverse_times(data))
-    D_inv = R.adjoint * N.inverse * R + S.inverse
+    D_inv = R.adjoint.chain(N.inverse).chain(R) + S.inverse
     # Make it invertible
     IC = ift.GradientNormController(iteration_limit=500, tol_abs_gradnorm=1e-3)
     D = ift.InversionEnabler(D_inv, IC, approximation=S.inverse).inverse
@@ -112,7 +112,8 @@ if __name__ == '__main__':
                         title="getting_started_1")
     else:
         ift.plot(HT(MOCK_SIGNAL), title='Mock Signal')
-        ift.plot(mask_to_nan(mask, (GR*Mask).adjoint(data)), title='Data')
+        ift.plot(mask_to_nan(mask, (GR.chain(Mask)).adjoint(data)),
+                 title='Data')
         ift.plot(HT(m), title='Reconstruction')
         ift.plot(mask_to_nan(mask, HT(m-MOCK_SIGNAL)))
         ift.plot_finish(nx=2, ny=2, xsize=10, ysize=10,

demos/getting_started_2.py

@@ -75,7 +75,7 @@ if __name__ == '__main__':
     M = ift.DiagonalOperator(exposure)
     GR = ift.GeometryRemover(position_space)
     # Set up instrumental response
-    R = GR * M
+    R = GR.chain(M)
 
     # Generate mock data
     d_space = R.target[0]

nifty5/energies/energy_adapter.py

@@ -23,7 +23,7 @@ class EnergyAdapter(Energy):
     @property
     def value(self):
         if self._val is None:
-            self._val =  self._op(self._position)
+            self._val = self._op(self._position)
         return self._val
 
     @property

nifty5/library/amplitude_model.py

@@ -130,7 +130,7 @@ class AmplitudeModel(Operator):
         cepstrum = create_cepstrum_amplitude_field(dof_space, kern)
 
         ceps = makeOp(sqrt(cepstrum))
-        self._smooth_op = sym * qht * ceps
+        self._smooth_op = sym.chain(qht).chain(ceps)
         self._keys = tuple(keys)
 
     @property

nifty5/linearization.py

@@ -47,8 +47,8 @@ class Linearization(object):
 
     def __neg__(self):
         return Linearization(
-            -self._val, self._jac*(-1),
-            None if self._metric is None else self._metric*(-1))
+            -self._val, self._jac.chain(-1),
+            None if self._metric is None else self._metric.chain(-1))
 
     def __add__(self, other):
         if isinstance(other, Linearization):
@@ -79,47 +79,49 @@ class Linearization(object):
             d2 = makeOp(other._val)
             return Linearization(
                 self._val*other._val,
-                RelaxedSumOperator((d2*self._jac, d1*other._jac)))
+                RelaxedSumOperator((d2.chain(self._jac),
+                                    d1.chain(other._jac))))
         if isinstance(other, (int, float, complex)):
             # if other == 0:
             #     return ...
-            met = None if self._metric is None else self._metric*other
-            return Linearization(self._val*other, self._jac*other, met)
+            met = None if self._metric is None else self._metric.chain(other)
+            return Linearization(self._val*other, self._jac.chain(other), met)
         if isinstance(other, (Field, MultiField)):
             d2 = makeOp(other)
-            return Linearization(self._val*other, d2*self._jac)
+            return Linearization(self._val*other, d2.chain(self._jac))
         raise TypeError
 
     def __rmul__(self, other):
         from .sugar import makeOp
         if isinstance(other, (int, float, complex)):
-            return Linearization(self._val*other, self._jac*other)
+            return Linearization(self._val*other, self._jac.chain(other))
         if isinstance(other, (Field, MultiField)):
             d1 = makeOp(other)
-            return Linearization(self._val*other, d1*self._jac)
+            return Linearization(self._val*other, d1.chain(self._jac))
 
     def sum(self):
         from .sugar import full
         from .operators.vdot_operator import VdotOperator
-        return Linearization(full((), self._val.sum()),
-                             VdotOperator(full(self._jac.target, 1))*self._jac)
+        return Linearization(
+            full((), self._val.sum()),
+            VdotOperator(full(self._jac.target, 1)).chain(self._jac))
 
     def exp(self):
         tmp = self._val.exp()
-        return Linearization(tmp, makeOp(tmp)*self._jac)
+        return Linearization(tmp, makeOp(tmp).chain(self._jac))
 
     def log(self):
         tmp = self._val.log()
-        return Linearization(tmp, makeOp(1./self._val)*self._jac)
+        return Linearization(tmp, makeOp(1./self._val).chain(self._jac))
 
     def tanh(self):
         tmp = self._val.tanh()
-        return Linearization(tmp, makeOp(1.-tmp**2)*self._jac)
+        return Linearization(tmp, makeOp(1.-tmp**2).chain(self._jac))
 
     def positive_tanh(self):
         tmp = self._val.tanh()
         tmp2 = 0.5*(1.+tmp)
-        return Linearization(tmp2, makeOp(0.5*(1.-tmp**2))*self._jac)
+        return Linearization(tmp2, makeOp(0.5*(1.-tmp**2)).chain(self._jac))
 
     def add_metric(self, metric):
         return Linearization(self._val, self._jac, metric)

nifty5/multi/block_diagonal_operator.py

@@ -68,7 +68,7 @@ class BlockDiagonalOperator(EndomorphicOperator):
     def _combine_chain(self, op):
         if self._domain is not op._domain:
             raise ValueError("domain mismatch")
-        res = tuple(v1*v2 for v1, v2 in zip(self._ops, op._ops))
+        res = tuple(v1.chain(v2) for v1, v2 in zip(self._ops, op._ops))
         return BlockDiagonalOperator(self._domain, res)
 
     def _combine_sum(self, op, selfneg, opneg):

nifty5/operators/harmonic_smoothing_operator.py

@@ -67,4 +67,4 @@ def HarmonicSmoothingOperator(domain, sigma, space=None):
     ddom = list(domain)
     ddom[space] = codomain
     diag = DiagonalOperator(kernel, ddom, space)
-    return Hartley.inverse*diag*Hartley
+    return Hartley.inverse.chain(diag).chain(Hartley)

nifty5/operators/linear_operator.py

@@ -117,18 +117,46 @@ class LinearOperator(Operator):
 
     def __mul__(self, other):
         from .chain_operator import ChainOperator
-        if np.isscalar(other) and other == 1.:
+        if not np.isscalar(other):
+            return Operator.__mul__(self, other)
+        if other == 1.:
             return self
-        other = self._toOperator(other, self.domain)
+        from .scaling_operator import ScalingOperator
+        other = ScalingOperator(other, self.domain)
         return ChainOperator.make([self, other])
 
     def __rmul__(self, other):
         from .chain_operator import ChainOperator
-        if np.isscalar(other) and other == 1.:
+        if not np.isscalar(other):
+            return Operator.__rmul__(self, other)
+        if other == 1.:
             return self
-        other = self._toOperator(other, self.target)
+        from .scaling_operator import ScalingOperator
+        other = ScalingOperator(other, self.target)
         return ChainOperator.make([other, self])
 
+    def __matmul__(self, other):
+        if np.isscalar(other) and other == 1.:
+            return self
+        other2 = self._toOperator(other, self.domain)
+        if other2 == NotImplemented:
+            return Operator.__matmul__(self, other)
+        from .chain_operator import ChainOperator
+        return ChainOperator.make([self, other2])
+
+    def chain(self, other):
+        return self.__matmul__(other)
+
+    def __rmatmul__(self, other):
+        if np.isscalar(other) and other == 1.:
+            return self
+        other2 = self._toOperator(other, self.target)
+        if other2 == NotImplemented:
+            from .chain_operator import ChainOperator
+            return Operator.__rmatmul__(self, other)
+        from .chain_operator import ChainOperator
+        return ChainOperator.make([other2, self])
+
     def __add__(self, other):
         from .sum_operator import SumOperator
         if np.isscalar(other) and other == 0.:
@@ -190,7 +218,7 @@ class LinearOperator(Operator):
         """Same as :meth:`times`"""
         from ..linearization import Linearization
         if isinstance(x, Linearization):
-            return Linearization(self(x._val), self*x._jac)
+            return Linearization(self(x._val), self.chain(x._jac))
         return self.apply(x, self.TIMES)
 
     def times(self, x):

nifty5/operators/sandwich_operator.py

@@ -56,9 +56,9 @@ class SandwichOperator(EndomorphicOperator):
             raise TypeError("cheese must be a linear operator")
         if cheese is None:
             cheese = ScalingOperator(1., bun.target)
-            op = bun.adjoint*bun
+            op = bun.adjoint.chain(bun)
         else:
-            op = bun.adjoint*cheese*bun
+            op = bun.adjoint.chain(cheese).chain(bun)
 
         # if our sandwich is diagonal, we can return immediately
         if isinstance(op, (ScalingOperator, DiagonalOperator)):

nifty5/operators/smoothness_operator.py

@@ -54,4 +54,4 @@ def SmoothnessOperator(domain, strength=1., logarithmic=True, space=None):
     if strength == 0.:
         return ScalingOperator(0., domain)
     laplace = LaplaceOperator(domain, logarithmic=logarithmic, space=space)
-    return (strength**2)*laplace.adjoint*laplace
+    return (strength**2)*laplace.adjoint.chain(laplace)

test/test_field.py

@@ -66,7 +66,7 @@ class Test_Functionality(unittest.TestCase):
 
         op1 = ift.create_power_operator((space1, space2), _spec1, 0)
         op2 = ift.create_power_operator((space1, space2), _spec2, 1)
-        opfull = op2*op1
+        opfull = op2.chain(op1)
 
         samples = 500
         sc1 = ift.StatCalculator()
@@ -94,7 +94,7 @@ class Test_Functionality(unittest.TestCase):
 
         S_1 = ift.create_power_operator((space1, space2), _spec1, 0)
         S_2 = ift.create_power_operator((space1, space2), _spec2, 1)
-        S_full = S_2*S_1
+        S_full = S_2.chain(S_1)
 
         samples = 500
         sc1 = ift.StatCalculator()

test/test_models/test_model_gradients.py

@@ -36,33 +36,6 @@ class Model_Tests(unittest.TestCase):
             return ift.Linearization.make_var(s)
         raise ValueError('unknown type passed')
 
-    def make_model(self, type, **kwargs):
-        if type == 'Constant':
-            np.random.seed(kwargs['seed'])
-            S = ift.ScalingOperator(1., kwargs['space'])
-            s = S.draw_sample()
-            return ift.Constant(
-                ift.MultiField.from_dict({kwargs['space_key']: s}),
-                ift.MultiField.from_dict({kwargs['space_key']: s}))
-        elif type == 'Variable':
-            np.random.seed(kwargs['seed'])
-            S = ift.ScalingOperator(1., kwargs['space'])
-            s = S.draw_sample()
-            return ift.Variable(
-                ift.MultiField.from_dict({kwargs['space_key']: s}))
-        elif type == 'LinearModel':
-            return ift.LinearModel(
-                inp=kwargs['model'], lin_op=kwargs['lin_op'])
-        else:
-            raise ValueError('unknown type passed')
-
-    def make_linear_operator(self, type, **kwargs):
-        if type == 'ScalingOperator':
-            lin_op = ift.ScalingOperator(1., kwargs['space'])
-        else:
-            raise ValueError('unknown type passed')
-        return lin_op
-
     @expand(product(
         [ift.GLSpace(15),
          ift.RGSpace(64, distances=.789),
@@ -71,7 +44,7 @@ class Model_Tests(unittest.TestCase):
         ))
     def testBasics(self, space, seed):
         var = self.make_linearization("Variable", space, seed)
-        model = lambda inp: inp
+        model = ift.ScalingOperator(6., var.target)
         ift.extra.check_value_gradient_consistency(model, var.val)
 
     @expand(product(
@@ -89,17 +62,17 @@ class Model_Tests(unittest.TestCase):
         lin2 = self.make_linearization(type2, dom2, seed)
 
         dom = ift.MultiDomain.union((dom1, dom2))
-        model = lambda inp: inp["s1"]*inp["s2"]
+        model = ift.FieldAdapter(dom, "s1")*ift.FieldAdapter(dom, "s2")
         pos = ift.from_random("normal", dom)
         ift.extra.check_value_gradient_consistency(model, pos)
-        model = lambda inp: inp["s1"]+inp["s2"]
+        model = ift.FieldAdapter(dom, "s1")+ift.FieldAdapter(dom, "s2")
         pos = ift.from_random("normal", dom)
         ift.extra.check_value_gradient_consistency(model, pos)
-        model = lambda inp: inp["s1"]*3.
-        pos = ift.from_random("normal", dom1)
+        model = ift.FieldAdapter(dom, "s1")*3.
+        pos = ift.from_random("normal", dom)
         ift.extra.check_value_gradient_consistency(model, pos)
-        model = lambda inp: ift.ScalingOperator(2.456, space)(
-            inp["s1"]*inp["s2"])
+        model = ift.ScalingOperator(2.456, space).chain(
+            ift.FieldAdapter(dom, "s1")*ift.FieldAdapter(dom, "s2"))
         pos = ift.from_random("normal", dom)
         ift.extra.check_value_gradient_consistency(model, pos)
         model = lambda inp: ift.ScalingOperator(2.456, space)(

test/test_multi_field.py

@@ -40,7 +40,7 @@ class Test_Functionality(unittest.TestCase):
     def test_blockdiagonal(self):
         op = ift.BlockDiagonalOperator(
             dom, (ift.ScalingOperator(20., dom["d1"]),))
-        op2 = op*op
+        op2 = op.chain(op)
         ift.extra.consistency_check(op2)
         assert_equal(type(op2), ift.BlockDiagonalOperator)
         f1 = op2(ift.full(dom, 1))

test/test_operators/test_adjoint.py

@@ -53,7 +53,7 @@ class Consistency_Tests(unittest.TestCase):
                                                        dtype=dtype))
         op = ift.SandwichOperator.make(a, b)
         ift.extra.consistency_check(op, dtype, dtype)
-        op = a*b
+        op = a.chain(b)
         ift.extra.consistency_check(op, dtype, dtype)
         op = a+b
         ift.extra.consistency_check(op, dtype, dtype)

test/test_operators/test_composed_operator.py

@@ -37,7 +37,7 @@ class ComposedOperator_Tests(unittest.TestCase):
         op1 = ift.DiagonalOperator(diag1, cspace, spaces=(0,))
         op2 = ift.DiagonalOperator(diag2, cspace, spaces=(1,))
 
-        op = op2*op1
+        op = op2.chain(op1)
 
         rand1 = ift.Field.from_random('normal', domain=(space1, space2))
         rand2 = ift.Field.from_random('normal', domain=(space1, space2))
@@ -54,7 +54,7 @@ class ComposedOperator_Tests(unittest.TestCase):
         op1 = ift.DiagonalOperator(diag1, cspace, spaces=(0,))
         op2 = ift.DiagonalOperator(diag2, cspace, spaces=(1,))
 
-        op = op2*op1
+        op = op2.chain(op1)
 
         rand1 = ift.Field.from_random('normal', domain=(space1, space2))
         tt1 = op.inverse_times(op.times(rand1))
@@ -75,7 +75,8 @@ class ComposedOperator_Tests(unittest.TestCase):
     def test_chain(self, space):
         op1 = ift.makeOp(ift.Field.full(space, 2.))
         op2 = 3.
-        full_op = op1 * op2 * (op2 * op1) * op1 * op1 * op2
+        full_op = (op1.chain(op2).chain(op2).chain(op1).
+                   chain(op1).chain(op1).chain(op2))
         x = ift.Field.full(space, 1.)
         res = full_op(x)
         assert_equal(isinstance(full_op, ift.DiagonalOperator), True)
@@ -85,7 +86,7 @@ class ComposedOperator_Tests(unittest.TestCase):
     def test_mix(self, space):
         op1 = ift.makeOp(ift.Field.full(space, 2.))
         op2 = 3.
-        full_op = op1 * (op2 + op2) * op1 * op1 - op1 * op2
+        full_op = op1.chain(op2 + op2).chain(op1).chain(op1) - op1.chain(op2)
         x = ift.Field.full(space, 1.)
         res = full_op(x)
         assert_equal(isinstance(full_op, ift.DiagonalOperator), True)