Make sure all cirq gates decompose to XPow/YPow/ZPow/CZPow + Measurement #4858

tanujkhattar · 2022-01-19T19:19:54Z

As part of https://tinyurl.com/cirq-organize-decompose-rfc, we want to make sure that all cirq gates decompose to the XPow/YPow/ZPow/CZPow + Measuremet gateset.

This issue is to track all the gates where we need to explicitly add a _decompose_ method to make sure all decompositions end on the above target gateset.
We should also discuss what happens for gates where trivial decomposition is not present?

MatrixGates : Should we add an analytical decomposition to the _decompose_ method of matrix gates?
Same question as above for ControlledGates.

P1 - I need this no later than the next release (end of quarter)

Part of #3242

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

…inate in `CZPowGate`. (#4862) Part of #4858 Decompositions are tested as part of `assert_eigengate_implements_consistent_protocols`, which is already tested for both gates.

vtomole · 2022-01-20T15:35:17Z

Should we add an analytical decomposition to the decompose method of matrix gates

How would this work? Is it possible to decompose any arbitrary MatrixGate into a target gateset?

tanujkhattar · 2022-01-20T16:43:33Z

@vtomole Yes. We have utilities in Cirq to decompose any arbitrary 2q Matrix Gate into the CZ gateset (see See https://github.com/quantumlib/Cirq/blob/master/cirq-core/cirq/transformers/analytical_decompositions/two_qubit_to_cz.py) and there's an open issue #4022 to add compilation of any arbitrary unitary to 1/2q gates

…inate in `CZPowGate`. (quantumlib#4862) Part of quantumlib#4858 Decompositions are tested as part of `assert_eigengate_implements_consistent_protocols`, which is already tested for both gates.

daxfohl · 2022-01-22T21:10:29Z

I also think we should have MeasurementGate with invert_mask decompose into a series of pre- and post-measurement Xs that represent the invert mask (and the subsequent uninversion), so that "measurement" can be a uniquely defined thing.

…5083) - Adds support for decomposing parameterized `cirq.PhasedXPowGate`s. - Also modifies the decomposition to respect global phase, so that the decomposition is valid for controlled variants as well. - Part of #4858

…5085) - Adds support for decomposing parameterized `cirq.DiagonalGate`. - Global phase is ignored for parameterized version because `cirq.GlobalPhaseGate` doesn't yet support symbols. - Part of #4858

…olledOperation` to end in X/Y/Z/CZ target gateset (#5091) When decomposed, controlled gates and operations simply fall back on the decomposition of underlying sub_gate / sub_operation and return apply appropriate controls to each decomposed operation. If we can ensure that all underlying gates / operations decompose to X/Y/Z/CZ target gateset, then their controlled versions will decompose to: - Multi controlled single qubit rotations (corresponding to (X/Y/Z).controlled_by(...)) OR - Multi controlled CZs, which is also equivalent to a multi controlled single qubit rotation (Z.controlled_by(...)) In Cirq, we have an analytical method to decompose a multi controlled rotation into X/Y/Z/CZ - `cirq.decompose_multi_controlled_rotation`, which is now used in the `_decompose_` method of controlled gates. However, there are many corner cases and limitations of the current approach, which are dealt appropriately in this PR to enable a "best-effort" decomposition of controlled gates to the cirq target gateset. Some of the limitations are: - If decomposition of sub_gate / sub_operation ignores global phase, then the controlled operation cannot directly rely on decomposing the sub operation. An explicit check is added to not fallback on sub_gate if sub_gate is a MatrixGate. - `decompose_multi_controlled_rotation` works only for qubits (doesn't work for qudits) and when all control_values are 1. Appropriate logic is added to extend its functionality to handle control_values which are 0 or (0, 1). - We have explicit types for a few important controlled gates, like `CCZ`, `CZ`, `CCX`, `CX` etc. in cirq. Appropriate type conversion logic is added to smartly infer the types of equivalent gates (eg: Controlled(sub_gate=CZ) should be inferred as CCZ) such that their decompositions can be used for decomposing the controlled gates. This is definitely the most tricky one to get right and I've added appropriate tests to cover the different cases. Part of #4858

…djacent swaps (#5093) - Adds decomposition to `cirq.QubitPermutationGate` in terms of minimum number of adjacent swap operations on qubits. - Part of #4858 Closes #5090

…olledOperation` to end in X/Y/Z/CZ target gateset (quantumlib#5091) When decomposed, controlled gates and operations simply fall back on the decomposition of underlying sub_gate / sub_operation and return apply appropriate controls to each decomposed operation. If we can ensure that all underlying gates / operations decompose to X/Y/Z/CZ target gateset, then their controlled versions will decompose to: - Multi controlled single qubit rotations (corresponding to (X/Y/Z).controlled_by(...)) OR - Multi controlled CZs, which is also equivalent to a multi controlled single qubit rotation (Z.controlled_by(...)) In Cirq, we have an analytical method to decompose a multi controlled rotation into X/Y/Z/CZ - `cirq.decompose_multi_controlled_rotation`, which is now used in the `_decompose_` method of controlled gates. However, there are many corner cases and limitations of the current approach, which are dealt appropriately in this PR to enable a "best-effort" decomposition of controlled gates to the cirq target gateset. Some of the limitations are: - If decomposition of sub_gate / sub_operation ignores global phase, then the controlled operation cannot directly rely on decomposing the sub operation. An explicit check is added to not fallback on sub_gate if sub_gate is a MatrixGate. - `decompose_multi_controlled_rotation` works only for qubits (doesn't work for qudits) and when all control_values are 1. Appropriate logic is added to extend its functionality to handle control_values which are 0 or (0, 1). - We have explicit types for a few important controlled gates, like `CCZ`, `CZ`, `CCX`, `CX` etc. in cirq. Appropriate type conversion logic is added to smartly infer the types of equivalent gates (eg: Controlled(sub_gate=CZ) should be inferred as CCZ) such that their decompositions can be used for decomposing the controlled gates. This is definitely the most tricky one to get right and I've added appropriate tests to cover the different cases. Part of quantumlib#4858

…djacent swaps (quantumlib#5093) - Adds decomposition to `cirq.QubitPermutationGate` in terms of minimum number of adjacent swap operations on qubits. - Part of quantumlib#4858 Closes quantumlib#5090

…inate in `CZPowGate`. (quantumlib#4862) Part of quantumlib#4858 Decompositions are tested as part of `assert_eigengate_implements_consistent_protocols`, which is already tested for both gates.

…uantumlib#5083) - Adds support for decomposing parameterized `cirq.PhasedXPowGate`s. - Also modifies the decomposition to respect global phase, so that the decomposition is valid for controlled variants as well. - Part of quantumlib#4858

…uantumlib#5085) - Adds support for decomposing parameterized `cirq.DiagonalGate`. - Global phase is ignored for parameterized version because `cirq.GlobalPhaseGate` doesn't yet support symbols. - Part of quantumlib#4858

…olledOperation` to end in X/Y/Z/CZ target gateset (quantumlib#5091) When decomposed, controlled gates and operations simply fall back on the decomposition of underlying sub_gate / sub_operation and return apply appropriate controls to each decomposed operation. If we can ensure that all underlying gates / operations decompose to X/Y/Z/CZ target gateset, then their controlled versions will decompose to: - Multi controlled single qubit rotations (corresponding to (X/Y/Z).controlled_by(...)) OR - Multi controlled CZs, which is also equivalent to a multi controlled single qubit rotation (Z.controlled_by(...)) In Cirq, we have an analytical method to decompose a multi controlled rotation into X/Y/Z/CZ - `cirq.decompose_multi_controlled_rotation`, which is now used in the `_decompose_` method of controlled gates. However, there are many corner cases and limitations of the current approach, which are dealt appropriately in this PR to enable a "best-effort" decomposition of controlled gates to the cirq target gateset. Some of the limitations are: - If decomposition of sub_gate / sub_operation ignores global phase, then the controlled operation cannot directly rely on decomposing the sub operation. An explicit check is added to not fallback on sub_gate if sub_gate is a MatrixGate. - `decompose_multi_controlled_rotation` works only for qubits (doesn't work for qudits) and when all control_values are 1. Appropriate logic is added to extend its functionality to handle control_values which are 0 or (0, 1). - We have explicit types for a few important controlled gates, like `CCZ`, `CZ`, `CCX`, `CX` etc. in cirq. Appropriate type conversion logic is added to smartly infer the types of equivalent gates (eg: Controlled(sub_gate=CZ) should be inferred as CCZ) such that their decompositions can be used for decomposing the controlled gates. This is definitely the most tricky one to get right and I've added appropriate tests to cover the different cases. Part of quantumlib#4858

…djacent swaps (quantumlib#5093) - Adds decomposition to `cirq.QubitPermutationGate` in terms of minimum number of adjacent swap operations on qubits. - Part of quantumlib#4858 Closes quantumlib#5090

…inate in `CZPowGate`. (quantumlib#4862) Part of quantumlib#4858 Decompositions are tested as part of `assert_eigengate_implements_consistent_protocols`, which is already tested for both gates.

…uantumlib#5083) - Adds support for decomposing parameterized `cirq.PhasedXPowGate`s. - Also modifies the decomposition to respect global phase, so that the decomposition is valid for controlled variants as well. - Part of quantumlib#4858

…uantumlib#5085) - Adds support for decomposing parameterized `cirq.DiagonalGate`. - Global phase is ignored for parameterized version because `cirq.GlobalPhaseGate` doesn't yet support symbols. - Part of quantumlib#4858

…olledOperation` to end in X/Y/Z/CZ target gateset (quantumlib#5091) When decomposed, controlled gates and operations simply fall back on the decomposition of underlying sub_gate / sub_operation and return apply appropriate controls to each decomposed operation. If we can ensure that all underlying gates / operations decompose to X/Y/Z/CZ target gateset, then their controlled versions will decompose to: - Multi controlled single qubit rotations (corresponding to (X/Y/Z).controlled_by(...)) OR - Multi controlled CZs, which is also equivalent to a multi controlled single qubit rotation (Z.controlled_by(...)) In Cirq, we have an analytical method to decompose a multi controlled rotation into X/Y/Z/CZ - `cirq.decompose_multi_controlled_rotation`, which is now used in the `_decompose_` method of controlled gates. However, there are many corner cases and limitations of the current approach, which are dealt appropriately in this PR to enable a "best-effort" decomposition of controlled gates to the cirq target gateset. Some of the limitations are: - If decomposition of sub_gate / sub_operation ignores global phase, then the controlled operation cannot directly rely on decomposing the sub operation. An explicit check is added to not fallback on sub_gate if sub_gate is a MatrixGate. - `decompose_multi_controlled_rotation` works only for qubits (doesn't work for qudits) and when all control_values are 1. Appropriate logic is added to extend its functionality to handle control_values which are 0 or (0, 1). - We have explicit types for a few important controlled gates, like `CCZ`, `CZ`, `CCX`, `CX` etc. in cirq. Appropriate type conversion logic is added to smartly infer the types of equivalent gates (eg: Controlled(sub_gate=CZ) should be inferred as CCZ) such that their decompositions can be used for decomposing the controlled gates. This is definitely the most tricky one to get right and I've added appropriate tests to cover the different cases. Part of quantumlib#4858

…djacent swaps (quantumlib#5093) - Adds decomposition to `cirq.QubitPermutationGate` in terms of minimum number of adjacent swap operations on qubits. - Part of quantumlib#4858 Closes quantumlib#5090

tanujkhattar added the kind/feature-request Describes new functionality label Jan 19, 2022

vtomole changed the title ~~Make sure all cirq gates decompose to XPow/YPow/ZPow/CZPow + Measuremet~~ Make sure all cirq gates decompose to XPow/YPow/ZPow/CZPow + Measurement Jan 19, 2022

tanujkhattar added area/decompose area/gates priority/p2 Next release should contain it triage/discuss Needs decision / discussion, bring these up during Cirq Cynque labels Jan 19, 2022

tanujkhattar mentioned this issue Jan 20, 2022

Add default decompositions for XXPowGate and YYPowGate which terminate in CZPowGate. #4862

Merged

daxfohl mentioned this issue Jan 22, 2022

Deferred measurements transformer #4849

Merged

tanujkhattar added triage/accepted A consensus emerged that this bug report, feature request, or other action should be worked on and removed triage/discuss Needs decision / discussion, bring these up during Cirq Cynque labels Jan 26, 2022

MichaelBroughton mentioned this issue Feb 11, 2022

Organization (and deprecation) of cirq-core/optimizers in cirq-core/transformers #4722

Closed

24 tasks

tanujkhattar mentioned this issue Mar 5, 2022

Make sure cirq gates and operations decompose to XPow/YPow/ZPow/CZPow + Measurement #5057

Closed

This was referenced Mar 16, 2022

Support decomposition of parameterized gates in cirq.PhasedXPowGate #5083

Merged

Add default decomposition for cirq.TwoQubitDiagonalGate #5084

Merged

tanujkhattar mentioned this issue Mar 16, 2022

Add support for decompositions of parameterized cirq.DiagonalGate #5085

Merged

tanujkhattar mentioned this issue Mar 21, 2022

Add assert_decompose_ends_at_default_gateset to consistent protocols test to ensure all cirq gates decompose to default gateset #5107

Merged

tanujkhattar closed this as completed in #5107 Mar 21, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Make sure all cirq gates decompose to XPow/YPow/ZPow/CZPow + Measurement #4858

Make sure all cirq gates decompose to XPow/YPow/ZPow/CZPow + Measurement #4858

tanujkhattar commented Jan 19, 2022 •

edited

Loading

vtomole commented Jan 20, 2022

tanujkhattar commented Jan 20, 2022

daxfohl commented Jan 22, 2022

Make sure all cirq gates decompose to XPow/YPow/ZPow/CZPow + Measurement #4858

Make sure all cirq gates decompose to XPow/YPow/ZPow/CZPow + Measurement #4858

Comments

tanujkhattar commented Jan 19, 2022 • edited Loading

vtomole commented Jan 20, 2022

tanujkhattar commented Jan 20, 2022

daxfohl commented Jan 22, 2022

tanujkhattar commented Jan 19, 2022 •

edited

Loading