Remove DeprecationWarning of np.int, np.bool, np.complex #5758

ikkoham · 2021-02-01T02:08:39Z

Summary

Resolves: #5756

Numpy 1.20.0 deprecates Using the aliases of builtin types like np.int.
See the release note.

t-imamichi · 2021-02-01T02:59:09Z

qiskit/circuit/register.py

@@ -131,7 +131,7 @@ def __getitem__(self, key):
            CircuitError: if the `key` is not an integer.
            QiskitIndexError: if the `key` is not in the range `(0, self.size)`.
        """
-        if not isinstance(key, (int, np.int, np.int32, np.int64, slice, list)):
+        if not isinstance(key, (int, np.int32, np.int64, slice, list)):


The following might be simpler. If we want reject np.int16 etc., the current code is appropriate.

Suggested change

if not isinstance(key, (int, np.int32, np.int64, slice, list)):

if not isinstance(key, (int, np.integer, slice, list)):

I am not sure why it does not accept np.int8 and np.int16, if the code owner is ok, please commit it.

The original code was committed as part of #4810. if not isinstance(key, (int, np.integer, slice, list)) looks more appropriate to me. @kdk What do you think?

I think this is probably fine, #4810 was a response to the performance regression in #4791 and just selected the types that were in the original issue about numpy types not working. np.integer shouldn't have the overhead that numbers.Integral introduced and would provide better matching for short types so I think it is a better solution. @t-imamichi do you want to push a follow up PR switching the numpy type checking to this instead?

t-imamichi · 2021-02-01T03:00:17Z

qiskit/quantum_info/operators/symplectic/pauli.py

@@ -290,7 +290,7 @@ def __len__(self):
    def __getitem__(self, qubits):
        """Return the unsigned Pauli group Pauli for subset of qubits."""
        # Set group phase to 0 so returned Pauli is always +1 coeff
-        if isinstance(qubits, (int, np.int)):
+        if isinstance(qubits, int):


If qubits can be a numpy integer, we need the following.

Suggested change

if isinstance(qubits, int):

if isinstance(qubits, (int, np.integer)):

I am not sure. If the code owner is ok, please commit it.

@chriseclectic Which did you intend, isinstance(qubits, int) or isinstance(qubits, (int, np.integer))?

qiskit/opflow/primitive_ops/pauli_op.py

Co-authored-by: Takashi Imamichi <31178928+t-imamichi@users.noreply.github.com>

t-imamichi · 2021-02-01T04:48:09Z

qiskit/quantum_info/operators/symplectic/pauli.py

@@ -316,7 +316,7 @@ def delete(self, qubits):
            QiskitError: if ind is out of bounds for the array size or
                         number of qubits.
        """
-        if isinstance(qubits, (int, np.int)):
+        if isinstance(qubits, int):


same as above

Suggested change

if isinstance(qubits, int):

if isinstance(qubits, (int, np.integer)):

t-imamichi · 2021-02-01T04:48:25Z

qiskit/quantum_info/operators/symplectic/pauli.py

-        if isinstance(qubits, (int, np.int)):
+        ret = Pauli((np.zeros(ret_qubits, dtype=bool),
+                     np.zeros(ret_qubits, dtype=bool)))
+        if isinstance(qubits, int):


same as above

Suggested change

if isinstance(qubits, int):

if isinstance(qubits, (int, np.integer)):

t-imamichi · 2021-02-01T04:48:43Z

qiskit/quantum_info/operators/symplectic/pauli_table.py

@@ -242,7 +242,7 @@ def __getitem__(self, key):
        """Return a view of the PauliTable."""
        # Returns a view of specified rows of the PauliTable
        # This supports all slicing operations the underlying array supports.
-        if isinstance(key, (int, np.int)):
+        if isinstance(key, int):


same as above

Suggested change

if isinstance(key, int):

if isinstance(key, (int, np.integer)):

t-imamichi · 2021-02-01T04:49:09Z

qiskit/quantum_info/operators/symplectic/sparse_pauli_op.py

@@ -128,7 +128,7 @@ def __getitem__(self, key):
        """Return a view of the SparsePauliOp."""
        # Returns a view of specified rows of the PauliTable
        # This supports all slicing operations the underlying array supports.
-        if isinstance(key, (int, np.int)):
+        if isinstance(key, int):


same as above

Suggested change

if isinstance(key, int):

if isinstance(key, (int, np.integer)):

ikkoham · 2021-02-01T10:11:04Z

Since this is a blocker for the whole Qiskit, I think it's best to merge it ASAP. So I'm trying to keep the behavior unchanged. (Note that np.int is an alias of int.) In particular, the quantum_info module might be better changed to np.bool_ and np.int64 for performance reasons, but that should be discussed in another issue.

t-imamichi · 2021-02-01T11:16:42Z

Yes, it makes sense to discuss the type of isinstance in another place. I leave my comments for the future discussion, but, I am OK to merge this PR as it is.

1ucian0

Thanks a lot for the quick fix!

* fix deprecation of np.int, np.bool, np.complex * Update qiskit/opflow/primitive_ops/pauli_op.py Co-authored-by: Takashi Imamichi <31178928+t-imamichi@users.noreply.github.com> Co-authored-by: Takashi Imamichi <31178928+t-imamichi@users.noreply.github.com> (cherry picked from commit 5f9da19) # Conflicts: # qiskit/opflow/primitive_ops/pauli_op.py # qiskit/opflow/primitive_ops/pauli_sum_op.py # qiskit/quantum_info/operators/measures.py # qiskit/quantum_info/operators/symplectic/base_pauli.py # qiskit/quantum_info/operators/symplectic/clifford.py # qiskit/quantum_info/operators/symplectic/pauli.py # qiskit/quantum_info/operators/symplectic/random.py # qiskit/visualization/pulse_v2/generators/waveform.py

The latest release of numpy, 1.20.0, deprecated type aliases for python's built in numeric types (complex, int, float, etc). The usage of these aliases has caused a myriad of deprecation warnings on import and during runtime everytime they were used. This commit fixes this so that terra's 0.16.x series is compatible with numpy 1.20.0. This commit is effectively a backport of Qiskit#5758 and Qiskit#5768 but because of significant divergance between master and stable/0.16 it was not a straight backport.

* Fix compatibility with numpy 1.20.0 The latest release of numpy, 1.20.0, deprecated type aliases for python's built in numeric types (complex, int, float, etc). The usage of these aliases has caused a myriad of deprecation warnings on import and during runtime everytime they were used. This commit fixes this so that terra's 0.16.x series is compatible with numpy 1.20.0. This commit is effectively a backport of #5758 and #5768 but because of significant divergance between master and stable/0.16 it was not a straight backport. * Disable pylint rule causing failure with fastjsonschema 2.15.0 (#5767) * Disable pylint rule causing failure with fastjsonschema 2.15.0 On Feb. 1, 2021 fastjsonschema 2.15.0 was released that made some changes to the exception API. [1][2] These changed the inheritance tree used for the exception classes so that 'JsonSchemaException' (which is the class qiskit is catching) was changed to be a base class instead of the class raised by validation errors. A new class 'JsonSchemaValueException' a subclass of that now base class is now raised by validation errors. The attributes used to wrap that exception and raise a custom exception are now defined in `JsonSchemaValueException` and not 'JsonSchemaException'. This is causing pylint to fail because the attributes are not defined in the base class. But in practice this will not raise an error as `JsonSchemaValueException` will always be raised even if we only isinstance check its base class. To still support old versions of fastjsonschema we shouldn't change that isinstance check so this commit just disables the rules. Especially since the qobj jsonschema validation is deprecated and will be removed in a future release as long as the code works pylint's pedantry isn't as important. [1] horejsek/python-fastjsonschema@64a635e [2] https://github.com/horejsek/python-fastjsonschema/blob/master/CHANGELOG.txt#L6-L8 * Fix lint in lint disable comment (cherry picked from commit 56facba) * Fix stray deprecated types

In Qiskit#5617 a single usage of np.int slipped in to the clifford module. This type is deprecated in numpy 1.20.0 and is just an alias for python's stdlib int type (see Qiskit#5788 and Qiskit#5758 for the earlier fixes). This usage is causing downstream failures in other project's CI that use this module in environment's where warnings aren't allowed (mainly ignis's docs builds). This commit fixes this stray np.int usage to unblock CI and remove the deprecation warning from anything using the clifford module.

In #5617 a single usage of np.int slipped in to the clifford module. This type is deprecated in numpy 1.20.0 and is just an alias for python's stdlib int type (see #5788 and #5758 for the earlier fixes). This usage is causing downstream failures in other project's CI that use this module in environment's where warnings aren't allowed (mainly ignis's docs builds). This commit fixes this stray np.int usage to unblock CI and remove the deprecation warning from anything using the clifford module. Co-authored-by: Luciano Bello <bel@zurich.ibm.com>

fix deprecation of np.int, np.bool, np.complex

74e03a9

ikkoham requested review from chriseclectic, DanPuzzuoli, eggerdj, lcapelluto, manoelmarques, nkanazawa1989, nonhermitian, taalexander, woodsp-ibm and a team as code owners February 1, 2021 02:08

ikkoham changed the title ~~Fix deprecation of np.int, np.bool, np.complex~~ Remove DeprecationWarning of np.int, np.bool, np.complex Feb 1, 2021

t-imamichi reviewed Feb 1, 2021

View reviewed changes

qiskit/opflow/primitive_ops/pauli_op.py Outdated Show resolved Hide resolved

Update qiskit/opflow/primitive_ops/pauli_op.py

0c2817a

Co-authored-by: Takashi Imamichi <31178928+t-imamichi@users.noreply.github.com>

t-imamichi reviewed Feb 1, 2021

View reviewed changes

ShellyGarion mentioned this pull request Feb 1, 2021

Improved the time complexity of TensoredFilter.apply() into O(n * 2^n) qiskit-community/qiskit-ignis#551

Merged

1ucian0 approved these changes Feb 1, 2021

View reviewed changes

1ucian0 merged commit 5f9da19 into Qiskit:master Feb 1, 2021

t-imamichi mentioned this pull request Feb 2, 2021

Fix an isintanceof statement to allow all numpy integers #5768

Merged

mtreinish added the stable backport potential The bug might be minimal and/or import enough to be port to stable label Feb 5, 2021

mergify bot mentioned this pull request Feb 5, 2021

Remove DeprecationWarning of np.int, np.bool, np.complex (bp #5758) #5786

Closed

mtreinish mentioned this pull request Feb 5, 2021

Fix compatibility with numpy 1.20.0 #5788

Merged

mtreinish mentioned this pull request Feb 15, 2021

Remove stray np.int usage #5849

Merged

t-imamichi mentioned this pull request Feb 18, 2021

Remove np.bool and np.int in the unit tests #5869

Merged

kdk added the Changelog: None Do not include in changelog label Mar 31, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove DeprecationWarning of np.int, np.bool, np.complex #5758

Remove DeprecationWarning of np.int, np.bool, np.complex #5758

ikkoham commented Feb 1, 2021 •

edited by 1ucian0

Loading

t-imamichi Feb 1, 2021 •

edited

Loading

ikkoham Feb 1, 2021

t-imamichi Feb 1, 2021

mtreinish Feb 1, 2021

t-imamichi Feb 1, 2021

ikkoham Feb 1, 2021

t-imamichi Feb 1, 2021

t-imamichi Feb 1, 2021

t-imamichi Feb 1, 2021

t-imamichi Feb 1, 2021 •

edited

Loading

t-imamichi Feb 1, 2021

ikkoham commented Feb 1, 2021

t-imamichi commented Feb 1, 2021

1ucian0 left a comment

	if not isinstance(key, (int, np.int32, np.int64, slice, list)):
	if not isinstance(key, (int, np.integer, slice, list)):

	if isinstance(qubits, int):
	if isinstance(qubits, (int, np.integer)):

	if isinstance(key, int):
	if isinstance(key, (int, np.integer)):

Remove DeprecationWarning of np.int, np.bool, np.complex #5758

Remove DeprecationWarning of np.int, np.bool, np.complex #5758

Conversation

ikkoham commented Feb 1, 2021 • edited by 1ucian0 Loading

Summary

t-imamichi Feb 1, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

t-imamichi Feb 1, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ikkoham commented Feb 1, 2021

t-imamichi commented Feb 1, 2021

1ucian0 left a comment

Choose a reason for hiding this comment

ikkoham commented Feb 1, 2021 •

edited by 1ucian0

Loading

t-imamichi Feb 1, 2021 •

edited

Loading

t-imamichi Feb 1, 2021 •

edited

Loading