Device deprecations #4744

MichaelBroughton · 2021-12-10T19:43:33Z

Summarize the task
As a part of the devices p1 and p2 we plan to move some behaviors and functionalities out of devices (in order to ensure the class is focused on validating). This issue is to track the deprecation progress of device functionality.

Acceptance criteria - when is the task considered done?

Deprecate the ability to construct a circuit with a device attached.
Deprecate the QidPairs and qid_pairs functionality. (Replacement logic will be in MetaData from the nx.Graph)
Deprecate the can_add_operation_into_moment functionality ("online" validation was decided that it was not worth trying to support in addition to existing validation paths)
Deprecate decompose_operation.
Deprecate qubit_set and move it over to MetaData.

Related

Related issues: #4743

The text was updated successfully, but these errors were encountered:

…cuit.device deprecation (#4757) First devices removal as a part of #4744 . The high level plan for deprecation is to: 1. pull all components of the library off of `.device` and onto `._device` to keep existing behavior working. 2. Deprecate dependent functionality that relies on `.device`. 3. After all dependency deprecations are done, deprecate the `.device` property in `cirq-core` circuits.py. 4. After the deprecation cycle is over, delete .device and _device properties from circuit, abstractcircuit, frozencircuit etc. This PR is tackles `cirq-pasqal`.

Next step in #4744 . Removes use of device attachment ability in cirq-google. This is a breaking change in the sense that some methods become "more forgiving" which I've highlighted below. Happy to add the tag if people think it's appropriate to label it.

Another sub-component for #4744 . This removes the use of `circuit.device` from all contrib modules. BREAKING_CHANGE=removing the `device` property from the `SwapNetwork` class. It also removes some type checking done by the code (all logic in the bulk still works), which makes it more forgiving. With the amount of meaningful maintenance contrib has had (not much), I'm not a huge fan of going through a deprecation process for these mild behavior widenings with no API changes. Are others alright with this ?

Second part of #4744 As far as I can tell there was just one use of `.device` in cirq-aqt in the form of an assertion that the attached device to a circuit was in fact a `cirq.IonDevice` where no call to validate ever took place. Also somewhat confusingly the number of qubits used in simulation would be dictated by the device qubit count instead of the circuit qubit count which could lead to some confusion on the users end. I've gone ahead and just removed the use of device to revert this confusing behavior (which does make this a breaking change).

Last item in step 1 of #4744 .

Adds metadata property default to None to all devices. I'm going to leave this open to vendors to add to this as they see fit. Part of #4743 . Note that if in the process of deprecating other things for #4744 , I will pull vendor devices onto using metadata to keep behavior working.

…cuit.device deprecation (quantumlib#4757) First devices removal as a part of quantumlib#4744 . The high level plan for deprecation is to: 1. pull all components of the library off of `.device` and onto `._device` to keep existing behavior working. 2. Deprecate dependent functionality that relies on `.device`. 3. After all dependency deprecations are done, deprecate the `.device` property in `cirq-core` circuits.py. 4. After the deprecation cycle is over, delete .device and _device properties from circuit, abstractcircuit, frozencircuit etc. This PR is tackles `cirq-pasqal`.

Next step in quantumlib#4744 . Removes use of device attachment ability in cirq-google. This is a breaking change in the sense that some methods become "more forgiving" which I've highlighted below. Happy to add the tag if people think it's appropriate to label it.

Another sub-component for quantumlib#4744 . This removes the use of `circuit.device` from all contrib modules. BREAKING_CHANGE=removing the `device` property from the `SwapNetwork` class. It also removes some type checking done by the code (all logic in the bulk still works), which makes it more forgiving. With the amount of meaningful maintenance contrib has had (not much), I'm not a huge fan of going through a deprecation process for these mild behavior widenings with no API changes. Are others alright with this ?

Second part of quantumlib#4744 As far as I can tell there was just one use of `.device` in cirq-aqt in the form of an assertion that the attached device to a circuit was in fact a `cirq.IonDevice` where no call to validate ever took place. Also somewhat confusingly the number of qubits used in simulation would be dictated by the device qubit count instead of the circuit qubit count which could lead to some confusion on the users end. I've gone ahead and just removed the use of device to revert this confusing behavior (which does make this a breaking change).

Last item in step 1 of quantumlib#4744 .

Adds metadata property default to None to all devices. I'm going to leave this open to vendors to add to this as they see fit. Part of quantumlib#4743 . Note that if in the process of deprecating other things for quantumlib#4744 , I will pull vendor devices onto using metadata to keep behavior working.

Deprecate `qid_pairs`. Next step on #4744 . This also makes a small update to the DeviceMetadata to make optional fields required (since the metadata property itself is optional this seems fine).

More parts for #4744

This deprecates qubit_set on `UnconstrainedDevice`, `Device` and `HyperGraphDevice` in contrib. Note that device already has metadata. UnconstrainedDevice probably shouldn't have a metadata and to add Metadata for HyperGraphDevice, might take a non-trivial amount of work. Maybe we open an issue for this ? xref: #4744 .

@mpharrigan

Yet more of quantumlib#4744 . It also looks like now we can get rid of the device shim @mpharrigan , @dstrain115 (will leave to seperate PR).

More of quantumlib#4744

This deprecates qubit_set on `UnconstrainedDevice`, `Device` and `HyperGraphDevice` in contrib. Note that device already has metadata. UnconstrainedDevice probably shouldn't have a metadata and to add Metadata for HyperGraphDevice, might take a non-trivial amount of work. Maybe we open an issue for this ? xref: quantumlib#4744 .

MichaelBroughton · 2022-04-20T18:39:35Z

Closing now since this is finished. Now just a matter of removing the deprecations themselves once the right version comes around.

…cuit.device deprecation (quantumlib#4757) First devices removal as a part of quantumlib#4744 . The high level plan for deprecation is to: 1. pull all components of the library off of `.device` and onto `._device` to keep existing behavior working. 2. Deprecate dependent functionality that relies on `.device`. 3. After all dependency deprecations are done, deprecate the `.device` property in `cirq-core` circuits.py. 4. After the deprecation cycle is over, delete .device and _device properties from circuit, abstractcircuit, frozencircuit etc. This PR is tackles `cirq-pasqal`.

Next step in quantumlib#4744 . Removes use of device attachment ability in cirq-google. This is a breaking change in the sense that some methods become "more forgiving" which I've highlighted below. Happy to add the tag if people think it's appropriate to label it.

Another sub-component for quantumlib#4744 . This removes the use of `circuit.device` from all contrib modules. BREAKING_CHANGE=removing the `device` property from the `SwapNetwork` class. It also removes some type checking done by the code (all logic in the bulk still works), which makes it more forgiving. With the amount of meaningful maintenance contrib has had (not much), I'm not a huge fan of going through a deprecation process for these mild behavior widenings with no API changes. Are others alright with this ?

Second part of quantumlib#4744 As far as I can tell there was just one use of `.device` in cirq-aqt in the form of an assertion that the attached device to a circuit was in fact a `cirq.IonDevice` where no call to validate ever took place. Also somewhat confusingly the number of qubits used in simulation would be dictated by the device qubit count instead of the circuit qubit count which could lead to some confusion on the users end. I've gone ahead and just removed the use of device to revert this confusing behavior (which does make this a breaking change).

Last item in step 1 of quantumlib#4744 .

Adds metadata property default to None to all devices. I'm going to leave this open to vendors to add to this as they see fit. Part of quantumlib#4743 . Note that if in the process of deprecating other things for quantumlib#4744 , I will pull vendor devices onto using metadata to keep behavior working.

Deprecate `qid_pairs`. Next step on quantumlib#4744 . This also makes a small update to the DeviceMetadata to make optional fields required (since the metadata property itself is optional this seems fine).

More parts for quantumlib#4744

More of quantumlib#4744 .

Yet more of quantumlib#4744

Yet more of quantumlib#4744 .

Yet more of quantumlib#4744

Yet another deprecation for quantumlib#4744 . Vendor deprecations of this fxn to follow.

Also adds the metadata field to devices. More of quantumlib#4744 .

@mpharrigan

Yet more of quantumlib#4744 . It also looks like now we can get rid of the device shim @mpharrigan , @dstrain115 (will leave to seperate PR).

More of quantumlib#4744

This deprecates qubit_set on `UnconstrainedDevice`, `Device` and `HyperGraphDevice` in contrib. Note that device already has metadata. UnconstrainedDevice probably shouldn't have a metadata and to add Metadata for HyperGraphDevice, might take a non-trivial amount of work. Maybe we open an issue for this ? xref: quantumlib#4744 .

MichaelBroughton added kind/task for tracking progress on larger efforts triage/accepted there is consensus amongst maintainers that this is a real bug or a reasonable feature to add labels Dec 10, 2021

MichaelBroughton self-assigned this Dec 10, 2021

viathor added the area/devices label Dec 14, 2021

MichaelBroughton mentioned this issue Dec 15, 2021

Remove uses of .device in cirq-pasqal in preparation for larger circuit.device deprecation #4757

Merged

This was referenced Jan 7, 2022

Remove use of .device from cirq-aqt #4812

Merged

Remove use of .device in cirq-google. #4820

Merged

Remove use of .device in contrib. #4821

Merged

MichaelBroughton mentioned this issue Jan 15, 2022

Deprecate ciruit.device everywhere. #4845

Merged

CirqBot pushed a commit that referenced this issue Jan 19, 2022

Deprecate ciruit.device everywhere. (#4845)

bce3002

Last item in step 1 of #4744 .

This was referenced Jan 19, 2022

[cirq-google] Proposal: Remove XmonDevice, Bristlecone, Foxtail and rename optimize_for_xmon. #4856

Closed

Deprecate XmonDevice, Foxtail and Bristlecone. #4864

Merged

Add metadata property to device. #4868

Merged

MichaelBroughton mentioned this issue Jan 21, 2022

Add metadata property for cirq_google devices. #4869

Merged

MichaelBroughton added a commit to MichaelBroughton/Cirq that referenced this issue Jan 22, 2022

Deprecate ciruit.device everywhere. (quantumlib#4845)

cc182ee

Last item in step 1 of quantumlib#4744 .

This was referenced Jan 27, 2022

Deprecate qid_pairs #4900

Merged

Deprecate can_add_operation_into_moment. #4902

Merged

CirqBot pushed a commit that referenced this issue Jan 31, 2022

Deprecate can_add_operation_into_moment. (#4902)

35e6d21

More parts for #4744

MichaelBroughton mentioned this issue Jan 31, 2022

Deprecate device.decompose_operation in cirq-core. #4922

Merged

95-martin-orion pushed a commit to 95-martin-orion/Cirq that referenced this issue Mar 2, 2022

Deprecate device.qubit_set in cirq_ionq. (quantumlib#4942)

4d10e5c

More of quantumlib#4744

95-martin-orion pushed a commit to 95-martin-orion/Cirq that referenced this issue Mar 2, 2022

Deperecate {ion,neutral_atom}.qubit_set. (quantumlib#4943)

9a85a8d

More of quantumlib#4744

This was referenced Mar 22, 2022

[AQT] Vendor Transformer adoption #5128

Closed

[IONQ] Vendor Transformer adoption #5129

Closed

[Pasqal] Vendor Transformer Adoption #5130

Closed

tanujkhattar added the time/before-1.0 label Mar 28, 2022

95-martin-orion mentioned this issue Apr 6, 2022

Circuit as simple container #2775

Closed

MichaelBroughton closed this as completed Apr 20, 2022

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deprecate ciruit.device everywhere. (quantumlib#4845)

6c98693

Last item in step 1 of quantumlib#4744 .

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deprecate can_add_operation_into_moment. (quantumlib#4902)

6821d9e

More parts for quantumlib#4744

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deprecate decompose_operation in cirq_google. (quantumlib#4924)

e91be48

More of quantumlib#4744 .

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deprecate device.decompose_operation in cirq_pasqal. (quantumlib#4927)

6a83679

Yet more of quantumlib#4744

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deprecate device.decompose_operation in cirq_ionq. (quantumlib#4925)

567b25c

Yet more of quantumlib#4744 .

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deprecate {ion,neutral_atom}.decompose_operation. (quantumlib#4930)

e77c11c

Yet more of quantumlib#4744

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deprecate device.decompose_operation in cirq-core. (quantumlib#4922)

41ddc5c

Yet another deprecation for quantumlib#4744 . Vendor deprecations of this fxn to follow.

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

deprecate device.qubit_set in cirq_pasqal. (quantumlib#4938)

51252ad

Also adds the metadata field to devices. More of quantumlib#4744 .

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deprecate device.qubit_set in cirq_google. (quantumlib#4940)

683551c

Yet more of quantumlib#4744 . It also looks like now we can get rid of the device shim @mpharrigan , @dstrain115 (will leave to seperate PR).

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deprecate device.qubit_set in cirq_ionq. (quantumlib#4942)

796b259

More of quantumlib#4744

rht pushed a commit to rht/Cirq that referenced this issue May 1, 2023

Deperecate {ion,neutral_atom}.qubit_set. (quantumlib#4943)

95e1a05

More of quantumlib#4744

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Device deprecations #4744

Device deprecations #4744

MichaelBroughton commented Dec 10, 2021

MichaelBroughton commented Apr 20, 2022

Device deprecations #4744

Device deprecations #4744

Comments

MichaelBroughton commented Dec 10, 2021

MichaelBroughton commented Apr 20, 2022