Merge branch 'rework_jactests' into 'NIFTy_7'

Restructure operator tests

See merge request !542

ChangeLog.md

 Changes since NIFTy 6
 =====================
 
+Naming of operator tests
+------------------------
+
+The implementation tests for nonlinear operators are now available in
+`ift.extra.check_operator()` and for linear operators
+`ift.extra.check_linear_operator()`.
+
 MetricGaussianKL interface
 --------------------------
 

demos/polynomial_fit.py

@@ -97,7 +97,7 @@ def main():
     p_space = ift.UnstructuredDomain(N_params)
     params = ift.full(p_space, 0.)
     R = PolynomialResponse(p_space, x)
-    ift.extra.consistency_check(R)
+    ift.extra.check_linear_operator(R)
 
     d_space = R.target
     d = ift.makeField(d_space, y)

src/extra.py

@@ -15,6 +15,8 @@
 #
 # NIFTy is being developed at the Max-Planck-Institut fuer Astrophysik.
 
+from itertools import combinations
+
 import numpy as np
 from numpy.testing import assert_
 
@@ -23,13 +25,103 @@ from .field import Field
 from .linearization import Linearization
 from .multi_domain import MultiDomain
 from .multi_field import MultiField
+from .operators.energy_operators import EnergyOperator
 from .operators.linear_operator import LinearOperator
-from .sugar import from_random
+from .operators.operator import Operator
+from .sugar import from_random, makeDomain
 
-__all__ = ["consistency_check", "check_jacobian_consistency",
+__all__ = ["check_linear_operator", "check_operator",
            "assert_allclose"]
 
 
+def check_linear_operator(op, domain_dtype=np.float64, target_dtype=np.float64,
+                          atol=1e-12, rtol=1e-12, only_r_linear=False):
+    """
+    Checks an operator for algebraic consistency of its capabilities.
+
+    Checks whether times(), adjoint_times(), inverse_times() and
+    adjoint_inverse_times() (if in capability list) is implemented
+    consistently. Additionally, it checks whether the operator is linear.
+
+    Parameters
+    ----------
+    op : LinearOperator
+        Operator which shall be checked.
+    domain_dtype : dtype
+        The data type of the random vectors in the operator's domain. Default
+        is `np.float64`.
+    target_dtype : dtype
+        The data type of the random vectors in the operator's target. Default
+        is `np.float64`.
+    atol : float
+        Absolute tolerance for the check. If rtol is specified,
+        then satisfying any tolerance will let the check pass.
+        Default: 0.
+    rtol : float
+        Relative tolerance for the check. If atol is specified,
+        then satisfying any tolerance will let the check pass.
+        Default: 0.
+    only_r_linear: bool
+        set to True if the operator is only R-linear, not C-linear.
+        This will relax the adjointness test accordingly.
+    """
+    if not isinstance(op, LinearOperator):
+        raise TypeError('This test tests only linear operators.')
+    _domain_check_linear(op, domain_dtype)
+    _domain_check_linear(op.adjoint, target_dtype)
+    _domain_check_linear(op.inverse, target_dtype)
+    _domain_check_linear(op.adjoint.inverse, domain_dtype)
+    _check_linearity(op, domain_dtype, atol, rtol)
+    _check_linearity(op.adjoint, target_dtype, atol, rtol)
+    _check_linearity(op.inverse, target_dtype, atol, rtol)
+    _check_linearity(op.adjoint.inverse, domain_dtype, atol, rtol)
+    _full_implementation(op, domain_dtype, target_dtype, atol, rtol,
+                         only_r_linear)
+    _full_implementation(op.adjoint, target_dtype, domain_dtype, atol, rtol,
+                         only_r_linear)
+    _full_implementation(op.inverse, target_dtype, domain_dtype, atol, rtol,
+                         only_r_linear)
+    _full_implementation(op.adjoint.inverse, domain_dtype, target_dtype, atol,
+                         rtol, only_r_linear)
+
+
+def check_operator(op, loc, tol=1e-12, ntries=100, perf_check=True,
+                   only_r_differentiable=True, metric_sampling=True):
+    """
+    Performs various checks of the implementation of linear and nonlinear
+    operators.
+
+    Computes the Jacobian with finite differences and compares it to the
+    implemented Jacobian.
+
+    Parameters
+    ----------
+    op : Operator
+        Operator which shall be checked.
+    loc : Field or MultiField
+        An Field or MultiField instance which has the same domain
+        as op. The location at which the gradient is checked
+    tol : float
+        Tolerance for the check.
+    perf_check : Boolean
+        Do performance check. May be disabled for very unimportant operators.
+    only_r_differentiable : Boolean
+        Jacobians of C-differentiable operators need to be C-linear.
+        Default: True
+    metric_sampling: Boolean
+        If op is an EnergyOperator, metric_sampling determines whether the
+        test shall try to sample from the metric or not.
+    """
+    if not isinstance(op, Operator):
+        raise TypeError('This test tests only linear operators.')
+    _domain_check_nonlinear(op, loc)
+    _performance_check(op, loc, bool(perf_check))
+    _linearization_value_consistency(op, loc)
+    _jac_vs_finite_differences(op, loc, np.sqrt(tol), ntries, only_r_differentiable)
+    _check_nontrivial_constant(op, loc, tol, ntries, only_r_differentiable,
+                               metric_sampling)
+
+
 def assert_allclose(f1, f2, atol, rtol):
     if isinstance(f1, Field):
         return np.testing.assert_allclose(f1.val, f2.val, atol=atol, rtol=rtol)
@@ -44,6 +136,20 @@ def assert_equal(f1, f2):
         assert_equal(val, f2[key])
 
 
+def _nozero(fld):
+    if isinstance(fld, Field):
+        return np.testing.assert_((fld != 0).s_all())
+    for val in fld.values():
+        _nozero(val)
+
+
+def _allzero(fld):
+    if isinstance(fld, Field):
+        return np.testing.assert_((fld == 0.).s_all())
+    for val in fld.values():
+        _allzero(val)
+
+
 def _adjoint_implementation(op, domain_dtype, target_dtype, atol, rtol,
                             only_r_linear):
     needed_cap = op.TIMES | op.ADJOINT_TIMES
@@ -90,7 +196,8 @@ def _check_linearity(op, domain_dtype, atol, rtol):
     assert_allclose(val1, val2, atol=atol, rtol=rtol)
 
 
-def _actual_domain_check_linear(op, domain_dtype=None, inp=None):
+def _domain_check_linear(op, domain_dtype=None, inp=None):
+    _domain_check(op)
     needed_cap = op.TIMES
     if (op.capability & needed_cap) != needed_cap:
         return
@@ -102,8 +209,9 @@ def _actual_domain_check_linear(op, domain_dtype=None, inp=None):
     assert_(op(inp).domain is op.target)
 
 
-def _actual_domain_check_nonlinear(op, loc):
-    assert isinstance(loc, (Field, MultiField))
+def _domain_check_nonlinear(op, loc):
+    _domain_check(op)
+    assert_(isinstance(loc, (Field, MultiField)))
     assert_(loc.domain is op.domain)
     for wm in [False, True]:
         lin = Linearization.make_var(loc, wm)
@@ -118,8 +226,8 @@ def _actual_domain_check_nonlinear(op, loc):
         assert_(reslin.jac.domain is reslin.domain)
         assert_(reslin.jac.target is reslin.target)
         assert_(lin.want_metric == reslin.want_metric)
-        _actual_domain_check_linear(reslin.jac, inp=loc)
-        _actual_domain_check_linear(reslin.jac.adjoint, inp=reslin.jac(loc))
+        _domain_check_linear(reslin.jac, inp=loc)
+        _domain_check_linear(reslin.jac.adjoint, inp=reslin.jac(loc))
         if reslin.metric is not None:
             assert_(reslin.metric.domain is reslin.metric.target)
             assert_(reslin.metric.domain is op.domain)
@@ -171,58 +279,6 @@ def _performance_check(op, pos, raise_on_fail):
                 raise RuntimeError(s)
 
 
-def consistency_check(op, domain_dtype=np.float64, target_dtype=np.float64,
-                      atol=0, rtol=1e-7, only_r_linear=False):
-    """
-    Checks an operator for algebraic consistency of its capabilities.
-
-    Checks whether times(), adjoint_times(), inverse_times() and
-    adjoint_inverse_times() (if in capability list) is implemented
-    consistently. Additionally, it checks whether the operator is linear.
-
-    Parameters
-    ----------
-    op : LinearOperator
-        Operator which shall be checked.
-    domain_dtype : dtype
-        The data type of the random vectors in the operator's domain. Default
-        is `np.float64`.
-    target_dtype : dtype
-        The data type of the random vectors in the operator's target. Default
-        is `np.float64`.
-    atol : float
-        Absolute tolerance for the check. If rtol is specified,
-        then satisfying any tolerance will let the check pass.
-        Default: 0.
-    rtol : float
-        Relative tolerance for the check. If atol is specified,
-        then satisfying any tolerance will let the check pass.
-        Default: 0.
-    only_r_linear: bool
-        set to True if the operator is only R-linear, not C-linear.
-        This will relax the adjointness test accordingly.
-    """
-    if not isinstance(op, LinearOperator):
-        raise TypeError('This test tests only linear operators.')
-    _domain_check(op)
-    _actual_domain_check_linear(op, domain_dtype)
-    _actual_domain_check_linear(op.adjoint, target_dtype)
-    _actual_domain_check_linear(op.inverse, target_dtype)
-    _actual_domain_check_linear(op.adjoint.inverse, domain_dtype)
-    _check_linearity(op, domain_dtype, atol, rtol)
-    _check_linearity(op.adjoint, target_dtype, atol, rtol)
-    _check_linearity(op.inverse, target_dtype, atol, rtol)
-    _check_linearity(op.adjoint.inverse, domain_dtype, atol, rtol)
-    _full_implementation(op, domain_dtype, target_dtype, atol, rtol,
-                         only_r_linear)
-    _full_implementation(op.adjoint, target_dtype, domain_dtype, atol, rtol,
-                         only_r_linear)
-    _full_implementation(op.inverse, target_dtype, domain_dtype, atol, rtol,
-                         only_r_linear)
-    _full_implementation(op.adjoint.inverse, domain_dtype, target_dtype, atol,
-                         rtol, only_r_linear)
-
-
 def _get_acceptable_location(op, loc, lin):
     if not np.isfinite(lin.val.s_sum()):
         raise ValueError('Initial value must be finite')
@@ -255,34 +311,48 @@ def _linearization_value_consistency(op, loc):
         assert_allclose(fld0, fld1, 0, 1e-7)
 
 
-def check_jacobian_consistency(op, loc, tol=1e-8, ntries=100, perf_check=True,
-                               only_r_differentiable=True):
-    """
-    Checks the Jacobian of an operator against its finite difference
-    approximation.
-
-    Computes the Jacobian with finite differences and compares it to the
-    implemented Jacobian.
-
-    Parameters
-    ----------
-    op : Operator
-        Operator which shall be checked.
-    loc : Field or MultiField
-        An Field or MultiField instance which has the same domain
-        as op. The location at which the gradient is checked
-    tol : float
-        Tolerance for the check.
-    perf_check : Boolean
-        Do performance check. May be disabled for very unimportant operators.
-    only_r_differentiable : Boolean
-        Jacobians of C-differentiable operators need to be C-linear. 
-        Default: True
-    """
-    _domain_check(op)
-    _actual_domain_check_nonlinear(op, loc)
-    _performance_check(op, loc, bool(perf_check))
-    _linearization_value_consistency(op, loc)
+def _check_nontrivial_constant(op, loc, tol, ntries, only_r_differentiable,
+                               metric_sampling):
+    return  # FIXME
+    # Assumes that the operator is not constant
+    if isinstance(op.domain, DomainTuple):
+        return
+    keys = op.domain.keys()
+    for ll in range(0, len(keys)):
+        for cstkeys in combinations(keys, ll):
+            cstdom, vardom = {}, {}
+            for kk, dd in op.domain.items():
+                if kk in cstkeys:
+                    cstdom[kk] = dd
+                else:
+                    vardom[kk] = dd
+            cstdom, vardom = makeDomain(cstdom), makeDomain(vardom)
+            cstloc = loc.extract(cstdom)
+
+            val0 = op(loc)
+            _, op0 = op.simplify_for_constant_input(cstloc)
+            val1 = op0(loc)
+            # MR FIXME: This tests something we don't promise!
+#            val2 = op0(loc.unite(cstloc))
+#            assert_equal(val1, val2)
+            assert_equal(val0, val1)
+
+            lin = Linearization.make_var(loc, want_metric=True)
+            oplin = op0(lin)
+            if isinstance(op, EnergyOperator):
+                _allzero(oplin.gradient.extract(cstdom))
+            # MR FIXME: This tests something we don't promise!
+#            _allzero(oplin.jac(from_random(cstdom).unite(full(vardom, 0))))
+
+            if isinstance(op, EnergyOperator) and metric_sampling:
+                samp0 = oplin.metric.draw_sample()
+                _allzero(samp0.extract(cstdom))
+                _nozero(samp0.extract(vardom))
+
+            _jac_vs_finite_differences(op0, loc, np.sqrt(tol), ntries, only_r_differentiable)
+
+
+def _jac_vs_finite_differences(op, loc, tol, ntries, only_r_differentiable):
     for _ in range(ntries):
         lin = op(Linearization.make_var(loc))
         loc2, lin2 = _get_acceptable_location(op, loc, lin)
@@ -307,8 +377,6 @@ def check_jacobian_consistency(op, loc, tol=1e-8, ntries=100, perf_check=True,
             print(hist)
             raise ValueError("gradient and value seem inconsistent")
         loc = locnext
-
-        ddtype = loc.values()[0].dtype if isinstance(loc, MultiField) else loc.dtype
-        tdtype = dirder.values()[0].dtype if isinstance(dirder, MultiField) else dirder.dtype
-        consistency_check(linmid.jac, domain_dtype=ddtype, target_dtype=tdtype,
-                          only_r_linear=only_r_differentiable)
+        check_linear_operator(linmid.jac, domain_dtype=loc.dtype,
+                              target_dtype=dirder.dtype,
+                              only_r_linear=only_r_differentiable)

src/operators/operator.py

@@ -291,7 +291,6 @@ class Operator(metaclass=NiftyMeta):
                 op = ConstantEnergyOperator(self.domain, self(c_inp))
             else:
                 op = ConstantOperator(self.domain, self(c_inp))
-            op = ConstantOperator(self.domain, self(c_inp))
             return op(c_inp), op
         if not isinstance(dom, MultiDomain):
             raise RuntimeError

src/operators/simplify_for_const.py

@@ -25,23 +25,23 @@ from .simple_linear_operators import NullOperator
 
 class ConstCollector(object):
     def __init__(self):
-        self._const = None
-        self._nc = set()
+        self._const = None  # MultiField on the part of the MultiDomain that could be constant
+        self._nc = set()  # NoConstant - set of keys that we know cannot be constant
 
     def mult(self, const, fulldom):
         if const is None:
-            self._nc |= set(fulldom)
+            self._nc |= set(fulldom.keys())
         else:
-            self._nc |= set(fulldom) - set(const)
+            from ..multi_field import MultiField
+            self._nc |= set(fulldom.keys()) - set(const.keys())
             if self._const is None:
-                from ..multi_field import MultiField
                 self._const = MultiField.from_dict(
-                    {key: const[key] for key in const if key not in self._nc})
-            else:
-                from ..multi_field import MultiField
+                    {key: const[key]
+                     for key in const.keys() if key not in self._nc})
+            else:  # we know that the domains are identical for products
                 self._const = MultiField.from_dict(
                     {key: self._const[key]*const[key]
-                     for key in const if key not in self._nc})
+                     for key in const.keys() if key not in self._nc})
 
     def add(self, const, fulldom):
         if const is None:
@@ -49,15 +49,10 @@ class ConstCollector(object):
         else:
             from ..multi_field import MultiField
             self._nc |= set(fulldom.keys()) - set(const.keys())
-            if self._const is None:
-                self._const = MultiField.from_dict(
-                    {key: const[key]
-                     for key in const.keys() if key not in self._nc})
-            else:
-                self._const = self._const.unite(const)
-                self._const = MultiField.from_dict(
-                    {key: self._const[key]
-                     for key in self._const if key not in self._nc})
+            self._const = const if self._const is None else self._const.unite(const)
+            self._const = MultiField.from_dict(
+                {key: const[key]
+                 for key in const.keys() if key not in self._nc})
 
     @property
     def constfield(self):

test/test_energy_gradients.py

@@ -32,13 +32,13 @@ ntries = 10
 
 def test_gaussian(field):
     energy = ift.GaussianEnergy(domain=field.domain)
-    ift.extra.check_jacobian_consistency(energy, field)
+    ift.extra.check_operator(energy, field)
 
 
 def test_ScaledEnergy(field):
     icov = ift.ScalingOperator(field.domain, 1.2)
     energy = ift.GaussianEnergy(inverse_covariance=icov, sampling_dtype=np.float64)
-    ift.extra.check_jacobian_consistency(energy.scale(0.3), field)
+    ift.extra.check_operator(energy.scale(0.3), field)
 
     lin = ift.Linearization.make_var(field, want_metric=True)
     met1 = energy(lin).metric
@@ -54,17 +54,17 @@ def test_QuadraticFormOperator(field):
     op = ift.ScalingOperator(field.domain, 1.2)
     endo = ift.makeOp(op.draw_sample_with_dtype(dtype=np.float64))
     energy = ift.QuadraticFormOperator(endo)
-    ift.extra.check_jacobian_consistency(energy, field)
+    ift.extra.check_operator(energy, field)
 
 
 def test_studentt(field):
     if isinstance(field.domain, ift.MultiDomain):
         return
     energy = ift.StudentTEnergy(domain=field.domain, theta=.5)
-    ift.extra.check_jacobian_consistency(energy, field, tol=1e-6)
+    ift.extra.check_operator(energy, field)
     theta = ift.from_random(field.domain, 'normal').exp()
     energy = ift.StudentTEnergy(domain=field.domain, theta=theta)
-    ift.extra.check_jacobian_consistency(energy, field, tol=1e-6, ntries=ntries)
+    ift.extra.check_operator(energy, field, ntries=ntries)
 
 
 def test_hamiltonian_and_KL(field):
@@ -72,10 +72,10 @@ def test_hamiltonian_and_KL(field):
     space = field.domain
     lh = ift.GaussianEnergy(domain=space)
     hamiltonian = ift.StandardHamiltonian(lh)
-    ift.extra.check_jacobian_consistency(hamiltonian, field, ntries=ntries)
+    ift.extra.check_operator(hamiltonian, field, ntries=ntries)
     samps = [ift.from_random(space, 'normal') for i in range(2)]
     kl = ift.AveragedEnergy(hamiltonian, samps)
-    ift.extra.check_jacobian_consistency(kl, field, ntries=ntries)
+    ift.extra.check_operator(kl, field, ntries=ntries)
 
 
 def test_variablecovariancegaussian(field):
@@ -84,7 +84,7 @@ def test_variablecovariancegaussian(field):
     dc = {'a': field, 'b': field.ptw("exp")}
     mf = ift.MultiField.from_dict(dc)
     energy = ift.VariableCovarianceGaussianEnergy(field.domain, 'a', 'b', np.float64)
-    ift.extra.check_jacobian_consistency(energy, mf, tol=1e-6, ntries=ntries)
+    ift.extra.check_operator(energy, mf, ntries=ntries)
     energy(ift.Linearization.make_var(mf, want_metric=True)).metric.draw_sample()
 
 
@@ -93,7 +93,7 @@ def test_specialgamma(field):
         return
     energy = ift.operators.energy_operators._SpecialGammaEnergy(field)
     loc = ift.from_random(energy.domain).exp()
-    ift.extra.check_jacobian_consistency(energy, loc, tol=1e-6, ntries=ntries)
+    ift.extra.check_operator(energy, loc, ntries=ntries)
     energy(ift.Linearization.make_var(loc, want_metric=True)).metric.draw_sample()
 
 
@@ -105,7 +105,7 @@ def test_inverse_gamma(field):
     d = ift.random.current_rng().normal(10, size=space.shape)**2
     d = ift.Field(space, d)
     energy = ift.InverseGammaLikelihood(d)
-    ift.extra.check_jacobian_consistency(energy, field, tol=1e-5)
+    ift.extra.check_operator(energy, field, tol=1e-10)
 
 
 def testPoissonian(field):
@@ -116,7 +116,7 @@ def testPoissonian(field):
     d = ift.random.current_rng().poisson(120, size=space.shape)
     d = ift.Field(space, d)
     energy = ift.PoissonianEnergy(d)
-    ift.extra.check_jacobian_consistency(energy, field, tol=1e-6)
+    ift.extra.check_operator(energy, field)
 
 
 def test_bernoulli(field):
@@ -127,4 +127,4 @@ def test_bernoulli(field):
     d = ift.random.current_rng().binomial(1, 0.1, size=space.shape)
     d = ift.Field(space, d)
     energy = ift.BernoulliEnergy(d)
-    ift.extra.check_jacobian_consistency(energy, field, tol=1e-5)
+    ift.extra.check_operator(energy, field, tol=1e-10)

test/test_gaussian_energy.py

@@ -70,7 +70,7 @@ def test_gaussian_energy(space, nonlinearity, noise, seed):
             N = None
 
         energy = ift.GaussianEnergy(d, N) @ d_model()
-        ift.extra.check_jacobian_consistency(
+        ift.extra.check_operator(
             energy, xi0, ntries=ntries, tol=1e-6)
 
 
@@ -99,9 +99,9 @@ def testgaussianenergy_compatibility(cplx):
     np.testing.assert_equal(val0, val1)
     np.testing.assert_equal(val1, val2)
 
-    ift.extra.check_jacobian_consistency(e, loc, ntries=ntries)
-    ift.extra.check_jacobian_consistency(e0, loc, ntries=ntries)
-    ift.extra.check_jacobian_consistency(e1, loc, ntries=ntries)
+    ift.extra.check_operator(e, loc, ntries=ntries)
+    ift.extra.check_operator(e0, loc, ntries=ntries, tol=1e-7)
+    ift.extra.check_operator(e1, loc, ntries=ntries)
 
     # Test jacobian is zero
     lin = ift.Linearization.make_var(loc, want_metric=True)

test/test_multi_field.py

@@ -61,7 +61,7 @@ def test_blockdiagonal():
     op = ift.BlockDiagonalOperator(
         dom, {"d1": ift.ScalingOperator(dom["d1"], 20.)})
     op2 = op(op)
-    ift.extra.consistency_check(op2)
+    ift.extra.check_linear_operator(op2)
     assert_equal(type(op2), ift.BlockDiagonalOperator)
     f1 = op2(ift.full(dom, 1))
     for val in f1.values():

test/test_operators/test_adjoint.py

@@ -42,7 +42,7 @@ def testLOSResponse(sp, dtype):
     sigma_low = 1e-4*ift.random.current_rng().standard_normal(10)
     sigma_ups = 1e-5*ift.random.current_rng().standard_normal(10)
     op = ift.LOSResponse(sp, starts, ends, sigma_low, sigma_ups)
-    ift.extra.consistency_check(op, dtype, dtype)
+    ift.extra.check_linear_operator(op, dtype, dtype)
 
 
 @pmp('sp', _h_spaces + _p_spaces + _pow_spaces)
@@ -50,13 +50,13 @@ def testOperatorCombinations(sp, dtype):
     a = ift.DiagonalOperator(ift.Field.from_random(sp, "normal", dtype=dtype))
     b = ift.DiagonalOperator(ift.Field.from_random(sp, "normal", dtype=dtype))
     op = ift.SandwichOperator.make(a, b)
-    ift.extra.consistency_check(op, dtype, dtype)
+    ift.extra.check_linear_operator(op, dtype, dtype)
     op = a(b)
-    ift.extra.consistency_check(op, dtype, dtype)
+    ift.extra.check_linear_operator(op, dtype, dtype)
     op = a + b
-    ift.extra.consistency_check(op, dtype, dtype)
+    ift.extra.check_linear_operator(op, dtype, dtype)
     op = a - b
-    ift.extra.consistency_check(op, dtype, dtype)
+    ift.extra.check_linear_operator(op, dtype, dtype)
 
 
 def testLinearInterpolator():
@@ -65,60 +65,60 @@ def testLinearInterpolator():
     pos[0, :] *= 0.9
     pos[1, :] *= 7*3.5
     op = ift.LinearInterpolator(sp, pos)
-    ift.extra.consistency_check(op)
+    ift.extra.check_linear_operator(op)
 
 
 @pmp('sp', _h_spaces + _p_spaces + _pow_spaces)
 def testRealizer(sp):
     op = ift.Realizer(sp)
-    ift.extra.consistency_check(op, np.complex128, np.float64,
+    ift.extra.check_linear_operator(op, np.complex128, np.float64,
                                 only_r_linear=True)
 
 
 @pmp('sp', _h_spaces + _p_spaces + _pow_spaces)
 def testImaginizer(sp):
     op = ift.Imaginizer(sp)
-    ift.extra.consistency_check(op, np.complex128, np.float64,
+    ift.extra.check_linear_operator(op, np.complex128, np.float64,
                                 only_r_linear=True)
     loc = ift.from_random(op.domain, dtype=np.complex128)
-    ift.extra.check_jacobian_consistency(op, loc)
+    ift.extra.check_operator(op, loc)
 
 
 @pmp('sp', _h_spaces + _p_spaces + _pow_spaces)
 def testConjugationOperator(sp):
     op = ift.ConjugationOperator(sp)
-    ift.extra.consistency_check(op, np.complex128, np.complex128,
+    ift.extra.check_linear_operator(op, np.complex128, np.complex128,
                                 only_r_linear=True)
 
 
 @pmp('sp', _h_spaces + _p_spaces + _pow_spaces)
 def testOperatorAdaptor(sp, dtype):
     op = ift.DiagonalOperator(ift.Field.from_random(sp, "normal", dtype=dtype))
-    ift.extra.consistency_check(op.adjoint, dtype, dtype)
-    ift.extra.consistency_check(op.inverse, dtype, dtype)
-    ift.extra.consistency_check(op.inverse.adjoint, dtype, dtype)
-    ift.extra.consistency_check(op.adjoint.inverse, dtype, dtype)
+    ift.extra.check_linear_operator(op.adjoint, dtype, dtype)
+    ift.extra.check_linear_operator(op.inverse, dtype, dtype)
+    ift.extra.check_linear_operator(op.inverse.adjoint, dtype, dtype)
+    ift.extra.check_linear_operator(op.adjoint.inverse, dtype, dtype)
 
 
 @pmp('sp1', _h_spaces + _p_spaces + _pow_spaces)
 @pmp('sp2', _h_spaces + _p_spaces + _pow_spaces)