LogInverseGammaOperator: treat scaling correctly

src/library/special_distributions.py

@@ -23,6 +23,7 @@ from .. import random
 from ..domain_tuple import DomainTuple
 from ..domains.unstructured_domain import UnstructuredDomain
 from ..field import Field
+from ..operators.adder import Adder
 from ..operators.operator import Operator
 from ..sugar import makeOp
 
@@ -133,9 +134,8 @@ def LogInverseGammaOperator(domain, alpha, q, delta=1e-2):
     """
     op = _InterpolationOperator(domain, lambda x: np.log(invgamma.ppf(norm._cdf(x), float(alpha))),
                                 -8.2, 8.2, delta)
-    if np.isscalar(q):
-        return op.scale(q)
-    return makeOp(q) @ op
+    q = np.log(q) if np.isscalar(q) else q.log()
+    return Adder(q, domain=op.target) @ op
 
 
 class UniformOperator(Operator):

test/test_operators/test_interpolated.py

@@ -11,7 +11,7 @@
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
-# Copyright(C) 2013-2019 Max-Planck-Society
+# Copyright(C) 2013-2021 Max-Planck-Society
 #
 # NIFTy is being developed at the Max-Planck-Institut fuer Astrophysik.
 
@@ -41,6 +41,9 @@ def testInterpolationAccuracy(space, seed):
         if not np.isscalar(q):
             qfld = ift.makeField(space, q)
         op = ift.InverseGammaOperator(space, alpha, qfld)
+        op1 = ift.LogInverseGammaOperator(space, alpha, qfld).exp()
         arr1 = op(pos).val
+        arr2 = op1(pos).val
         arr0 = invgamma.ppf(norm.cdf(pos.val), alpha, scale=q)
         assert_allclose(arr0, arr1)
+        assert_allclose(arr0, arr2)