Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open v0.3 API + debiasing #75

Merged
merged 34 commits into from
Jan 2, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
34 commits
Select commit Hold shift + click to select a range
fb07c8a
create job with debias
splch Jul 24, 2023
0d01740
get sharpened results
splch Jul 24, 2023
ac9d407
add error_mitigation and sharpen to the end
splch Jul 24, 2023
d54f871
use userwarning
splch Jul 24, 2023
543c885
fix undefined params and sharpen
splch Jul 24, 2023
2386290
reorder args
splch Jul 24, 2023
bdfa904
final arg reorder
splch Jul 24, 2023
2fc6d9b
Merge branch 'master' into v0.3-and-symmetrization
timmysilv Jul 24, 2023
f5503b2
remove unnecessary kwargs pop
splch Jul 24, 2023
a0a77a7
use new fields for tests
splch Aug 3, 2023
e8ad837
add to contributors
splch Aug 7, 2023
1efeb0d
format code + add params
splch Oct 19, 2023
8daacfe
Merge branch 'master' into v0.3-and-symmetrization
splch Oct 19, 2023
ef57f5f
Merge branch 'PennyLaneAI:master' into v0.3-and-symmetrization
splch Oct 30, 2023
be25d2e
Merge branch 'master' into v0.3-and-symmetrization
splch Nov 4, 2023
d4b5d7c
Merge branch 'master' into v0.3-and-symmetrization
lillian542 Dec 15, 2023
075f551
Merge branch 'master' into v0.3-and-symmetrization
lillian542 Dec 18, 2023
10116db
Update CHANGELOG.md
lillian542 Dec 18, 2023
f28f2a9
increase test coverage
lillian542 Dec 18, 2023
ba6dd7e
update changelog
lillian542 Dec 18, 2023
01f6042
add error_mitigation testing
lillian542 Dec 18, 2023
c9750f1
expand on docstrings
lillian542 Dec 18, 2023
6c288f7
black formatting
lillian542 Dec 18, 2023
b3c6f61
update overview doc to refer to docstrings for details
lillian542 Dec 18, 2023
100d6bf
more black formatting
lillian542 Dec 18, 2023
c2a9793
update hyperlink
lillian542 Dec 18, 2023
0bfef67
troubleshoot docstring formatting problem
lillian542 Dec 18, 2023
bed8239
troubleshoot docstring formatting problem 2
lillian542 Dec 18, 2023
27c332f
update changelog and docstrings
lillian542 Dec 18, 2023
b37d18e
black formatting for tests
lillian542 Dec 18, 2023
33ab2da
correct wording
lillian542 Dec 18, 2023
a4712a6
default backend is harmony
timmysilv Jan 2, 2024
7bd4a44
try to please codecov
timmysilv Jan 2, 2024
36b8c96
revert and simplify default target
timmysilv Jan 2, 2024
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 17 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -2,8 +2,24 @@

### New features since last release

* Application of debiasing and sharpening for error mitigation is made available, with parameters set on device initialization. Error mitigation strategies that
need to be set at runtime are defined in the `error_mitigation` dictionary (currently a single strategy, `debias`, is available). Whether or not to
apply sharpening to the returned results is set via the parameter `sharpen`. A device using debiasing and sharpening to mitigate errors can be initialized as:

```python
import pennylane as qml

dev = qml.device("ionq.qpu", wires=2, error_mitigation={"debias": True}, sharpen=True)
```

For more details, see the [IonQ Guide on sharpening and debiasing](https://ionq.com/resources/debiasing-and-sharpening), or refer to the publication <https://arxiv.org/pdf/2301.07233.pdf>
[(#75)](https://github.com/PennyLaneAI/PennyLane-IonQ/pull/75)

### Improvements 🛠

* The IonQ API version accessed via the plugin is updated from 0.1 to 0.3
[(#75)](https://github.com/PennyLaneAI/PennyLane-IonQ/pull/75)

* Use new `backend` field to specify `qpu`.
[(#81)](https://github.com/PennyLaneAI/PennyLane-IonQ/pull/81)

Expand All @@ -20,6 +36,7 @@
This release contains contributions from (in alphabetical order):

Spencer Churchill
Lillian Frederiksen

---
# Release 0.32.0
Expand Down
14 changes: 6 additions & 8 deletions doc/devices.rst
Original file line number Diff line number Diff line change
Expand Up @@ -4,18 +4,17 @@ IonQ Devices
The PennyLane-IonQ plugin provides the ability for PennyLane to access
devices available via IonQ's online API.

Currently, access is available to two remote devices: an ideal and
a noisy trapped-ion simulator.
Currently, access is available to two remote devices: one to access an ideal
trapped-ion simulator and another to access to IonQ's trapped-ion QPUs.

.. raw::html
<section id="simulator">

Ideal trapped-ion simulator
------------------------

This device provides an ideal noiseless trapped-ion simulation.
Once the plugin has been installed, you can use this device
directly in PennyLane by specifying ``"ionq.simulator"``:
The :class:`~.pennylane_ionq.SimulatorDevice` provides an ideal noiseless trapped-ion simulation.
Once the plugin has been installed, you can use this device directly in PennyLane by specifying ``"ionq.simulator"``:

.. code-block:: python

Expand All @@ -38,9 +37,8 @@ directly in PennyLane by specifying ``"ionq.simulator"``:
Trapped-Ion QPU
---------------

This device provides access to IonQ's trapped-ion QPUs.
Once the plugin has been installed, you can use this device
directly in PennyLane by specifying ``"ionq.qpu"`` with a
The :class:`~.pennylane_ionq.QPUDevice` provides access to IonQ's trapped-ion QPUs. Once the plugin has been
installed, you can use this device directly in PennyLane by specifying ``"ionq.qpu"`` with a
``"backend"`` from `available backends <https://docs.ionq.com/#tag/jobs>`_:

.. code-block:: python
Expand Down
21 changes: 13 additions & 8 deletions pennylane_ionq/api_client.py
Original file line number Diff line number Diff line change
Expand Up @@ -112,12 +112,16 @@
api_key (str): IonQ cloud platform API key
"""

USER_AGENT = "pennylane-ionq-api-client/0.1"
HOSTNAME = "api.ionq.co/v0.1"
USER_AGENT = "pennylane-ionq-api-client/0.3"
HOSTNAME = "api.ionq.co/v0.3"
BASE_URL = "https://{}".format(HOSTNAME)

def __init__(self, **kwargs):
self.AUTHENTICATION_TOKEN = os.getenv("IONQ_API_KEY") or kwargs.get("api_key", None)
self.AUTHENTICATION_TOKEN = (
kwargs.get("api_key", None)
or os.getenv("PENNYLANE_IONQ_API_KEY")
or os.getenv("IONQ_API_KEY")
)
timmysilv marked this conversation as resolved.
Show resolved Hide resolved
self.DEBUG = False

if "IONQ_DEBUG" in os.environ:
Expand Down Expand Up @@ -196,17 +200,18 @@

return response

def get(self, path):
def get(self, path, params=None):
"""
Sends a GET request to the provided path. Returns a response object.

Args:
path (str): path to send the GET request to
params (dict): parameters to include in the request

Returns:
requests.Response: A response object, or None if no response could be fetched
"""
return self.request(requests.get, url=self.join_path(path))
return self.request(requests.get, url=self.join_path(path), params=params)

def post(self, path, payload):
"""
Expand Down Expand Up @@ -249,7 +254,7 @@
"""
return join_path(self.resource.PATH, path)

def get(self, resource_id=None):
def get(self, resource_id=None, params=None):
"""
Attempts to retrieve a particular record by sending a GET
request to the appropriate endpoint. If successful, the resource
Expand All @@ -262,9 +267,9 @@
raise MethodNotSupportedException("GET method on this resource is not supported")

if resource_id is not None:
response = self.client.get(self.join_path(str(resource_id)))
response = self.client.get(self.join_path(str(resource_id)), params=params)
else:
response = self.client.get(self.resource.PATH)
response = self.client.get(self.resource.PATH, params=params)
lillian542 marked this conversation as resolved.
Show resolved Hide resolved
self.handle_response(response)

def create(self, **params):
Expand Down Expand Up @@ -337,7 +342,7 @@
def refresh_data(self, data):
"""
Refreshes the instance's attributes with the provided data and
converts it to the correct type.

Check notice on line 345 in pennylane_ionq/api_client.py

View check run for this annotation

codefactor.io / CodeFactor

pennylane_ionq/api_client.py#L345

Raising too general exception: Exception (broad-exception-raised)

Args:
data (dict): A dictionary containing keys and values of data to be stored on the object.
Expand Down
90 changes: 72 additions & 18 deletions pennylane_ionq/device.py
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@
"""
This module contains the device class for constructing IonQ devices for PennyLane.
"""
import os, warnings
import warnings
from time import sleep

import numpy as np
Expand Down Expand Up @@ -63,16 +63,27 @@ class IonQDevice(QubitDevice):
r"""IonQ device for PennyLane.

Args:
target (str): the target device, either ``"simulator"`` or ``"qpu"``
wires (int or Iterable[Number, str]]): Number of wires to initialize the device with,
or iterable that contains unique labels for the subsystems as numbers (i.e., ``[-1, 0, 2]``)
or strings (``['ancilla', 'q1', 'q2']``).
gateset (str): the target gateset, either ``"qis"`` or ``"native"``.

Kwargs:
target (str): the target device, either ``"simulator"`` or ``"qpu"``. Defaults to ``simulator``.
gateset (str): the target gateset, either ``"qis"`` or ``"native"``. Defaults to ``qis``.
shots (int, list[int]): Number of circuit evaluations/random samples used to estimate
expectation values of observables.
expectation values of observables. Defaults to 1024.
If a list of integers is passed, the circuit evaluations are batched over the list of shots.
api_key (str): The IonQ API key. If not provided, the environment
variable ``IONQ_API_KEY`` is used.
error_mitigation (dict): settings for error mitigation when creating a job. Defaults to None.
Not available on all backends. Set by default on some hardware systems. See
`IonQ API Job Creation <https://docs.ionq.com/#tag/jobs/operation/createJob>`_ and
`IonQ Debiasing and Sharpening <https://ionq.com/resources/debiasing-and-sharpening>`_ for details.
Valid keys include: ``debias`` (bool).
sharpen (bool): whether to use sharpening when accessing the results of an executed job. Defaults to None
(no value passed at job retrieval). Will generally return more accurate results if your expected output
distribution has peaks. See `IonQ Debiasing and Sharpening
<https://ionq.com/resources/debiasing-and-sharpening>`_ for details.
"""
# pylint: disable=too-many-instance-attributes
name = "IonQ PennyLane plugin"
Expand All @@ -91,14 +102,26 @@ class IonQDevice(QubitDevice):
# and therefore does not support the Hermitian observable.
observables = {"PauliX", "PauliY", "PauliZ", "Hadamard", "Identity"}

def __init__(self, wires, *, target="simulator", gateset="qis", shots=1024, api_key=None):
def __init__(
self,
wires,
*,
target="simulator",
gateset="qis",
shots=1024,
api_key=None,
error_mitigation=None,
sharpen=False,
):
if shots is None:
raise ValueError("The ionq device does not support analytic expectation values.")

super().__init__(wires=wires, shots=shots)
self.target = target
self.api_key = api_key
self.gateset = gateset
self.error_mitigation = error_mitigation
self.sharpen = sharpen
self._operation_map = _GATESET_OPS[gateset]
self.reset()

Expand All @@ -107,16 +130,25 @@ def reset(self):
self._prob_array = None
self.histogram = None
self.circuit = {
"format": "ionq.circuit.v0",
"qubits": self.num_wires,
"circuit": [],
"gateset": self.gateset,
}
self.job = {
"lang": "json",
"body": self.circuit,
"input": self.circuit,
"target": self.target,
"shots": self.shots,
}
if self.error_mitigation is not None:
self.job["error_mitigation"] = self.error_mitigation
lillian542 marked this conversation as resolved.
Show resolved Hide resolved
if self.job["target"] == "qpu":
self.job["target"] = "qpu.harmony"
warnings.warn(
"The ionq_qpu backend is deprecated. Defaulting to ionq_qpu.harmony.",
UserWarning,
stacklevel=2,
)

@property
def operations(self):
Expand Down Expand Up @@ -190,7 +222,9 @@ def _submit_job(self):
if job.is_failed:
raise JobExecutionError("Job failed")

job.manager.get(job.id.value)
params = {} if self.sharpen is None else {"sharpen": self.sharpen}

job.manager.get(resource_id=job.id.value, params=params)

# The returned job histogram is of the form
# dict[str, float], and maps the computational basis
Expand Down Expand Up @@ -242,19 +276,26 @@ class SimulatorDevice(IonQDevice):
wires (int or Iterable[Number, str]]): Number of wires to initialize the device with,
or iterable that contains unique labels for the subsystems as numbers (i.e., ``[-1, 0, 2]``)
or strings (``['ancilla', 'q1', 'q2']``).
gateset (str): the target gateset, either ``"qis"`` or ``"native"``.
shots (int, list[int]): Number of circuit evaluations/random samples used to estimate
gateset (str): the target gateset, either ``"qis"`` or ``"native"``. Defaults to ``qis``.
shots (int, list[int], None): Number of circuit evaluations/random samples used to estimate
expectation values of observables. If ``None``, the device calculates probability, expectation values,
and variances analytically. If an integer, it specifies the number of samples to estimate these quantities.
If a list of integers is passed, the circuit evaluations are batched over the list of shots.
Defaults to 1024.
api_key (str): The IonQ API key. If not provided, the environment
variable ``IONQ_API_KEY`` is used.
"""
name = "IonQ Simulator PennyLane plugin"
short_name = "ionq.simulator"

def __init__(self, wires, *, target="simulator", gateset="qis", shots=1024, api_key=None):
super().__init__(wires=wires, target=target, gateset=gateset, shots=shots, api_key=api_key)
def __init__(self, wires, *, gateset="qis", shots=1024, api_key=None):
super().__init__(
wires=wires,
target="simulator",
gateset=gateset,
shots=shots,
api_key=api_key,
)

def generate_samples(self):
"""Generates samples by random sampling with the probabilities returned by the simulator."""
Expand All @@ -270,14 +311,23 @@ class QPUDevice(IonQDevice):
wires (int or Iterable[Number, str]]): Number of wires to initialize the device with,
or iterable that contains unique labels for the subsystems as numbers (i.e., ``[-1, 0, 2]``)
or strings (``['ancilla', 'q1', 'q2']``).
gateset (str): the target gateset, either ``"qis"`` or ``"native"``.
gateset (str): the target gateset, either ``"qis"`` or ``"native"``. Defaults to ``qis``.
backend (str): Optional specifier for an IonQ backend. Can be ``"harmony"``, ``"aria-1"``, etc.
Default to ``harmony``.
shots (int, list[int]): Number of circuit evaluations/random samples used to estimate
expectation values of observables. If ``None``, the device calculates probability, expectation values,
and variances analytically. If an integer, it specifies the number of samples to estimate these quantities.
If a list of integers is passed, the circuit evaluations are batched over the list of shots.
expectation values of observables. Defaults to 1024. If a list of integers is passed, the
circuit evaluations are batched over the list of shots.
api_key (str): The IonQ API key. If not provided, the environment
variable ``IONQ_API_KEY`` is used.
error_mitigation (dict): settings for error mitigation when creating a job. Defaults to None.
Not available on all backends. Set by default on some hardware systems. See
`IonQ API Job Creation <https://docs.ionq.com/#tag/jobs/operation/createJob>`_ and
`IonQ Debiasing and Sharpening <https://ionq.com/resources/debiasing-and-sharpening>`_ for details.
Valid keys include: ``debias`` (bool).
sharpen (bool): whether to use sharpening when accessing the results of an executed job.
Defaults to None (no value passed at job retrieval). Will generally return more accurate results if
your expected output distribution has peaks. See `IonQ Debiasing and Sharpening
<https://ionq.com/resources/debiasing-and-sharpening>`_ for details.
"""
name = "IonQ QPU PennyLane plugin"
short_name = "ionq.qpu"
Expand All @@ -286,12 +336,14 @@ def __init__(
self,
wires,
*,
target="qpu",
backend=None,
gateset="qis",
shots=1024,
backend="harmony",
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

changed the default backend to "harmony" because every job submission was raising a warning unless the user specifies "harmony" (or some other backend). Since I believe this was effectively the default before, this will keep that behaviour and not give users warnings (unless they specify None, in which case the warning is justified)

error_mitigation=None,
sharpen=None,
api_key=None,
):
target = "qpu"
self.backend = backend
if self.backend is not None:
target += "." + self.backend
Expand All @@ -301,6 +353,8 @@ def __init__(
gateset=gateset,
shots=shots,
api_key=api_key,
error_mitigation=error_mitigation,
sharpen=sharpen,
)

def generate_samples(self):
Expand Down
4 changes: 1 addition & 3 deletions tests/conftest.py
Original file line number Diff line number Diff line change
Expand Up @@ -36,9 +36,7 @@
U2 = np.array([[0, 1, 1, 1], [1, 0, 1, -1], [1, -1, 0, 1], [1, 1, -1, 0]]) / np.sqrt(3)

# single qubit Hermitian observable
A = np.array(
[[1.02789352, 1.61296440 - 0.3498192j], [1.61296440 + 0.3498192j, 1.23920938 + 0j]]
)
A = np.array([[1.02789352, 1.61296440 - 0.3498192j], [1.61296440 + 0.3498192j, 1.23920938 + 0j]])


# ==========================================================
Expand Down
Loading
Loading