Don't substitute ideal gates in target with Optimize1qGatesDecomposition #11351
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This commit fixes an issue in Optimize1qGatesDecomposition where the pass would defer to synthesized gates if the output from synthesis was ideal even if the input gate was also ideal. This typically comes up in simulators where there are no error rates for any gates and everything is listed as ideal. This would cause the transpiler to translate gates unnecessarily which was unexpected. This commit fixes this by adding an additional check to the subsitution condition to ensure we're only substituting a gate if it's not in the target (when they're all ideal). Fixes Qiskit#10568
mtreinish
added
stable backport potential
Pull Request Test Coverage Report for Build 7052106759
💛 - Coveralls |
jakelishman
reviewed
Nov 30, 2023
qiskit/transpiler/passes/optimization/optimize_1q_decomposition.py
Outdated
Show resolved
Hide resolved
Co-authored-by: Jake Lishman <jake@binhbar.com>
jakelishman
approved these changes
Nov 30, 2023
mergify bot
pushed a commit
that referenced
this pull request
Dec 1, 2023
…ion (#11351) * Don't substitute ideal gates in target with Optimize1qGatesDecomposition This commit fixes an issue in Optimize1qGatesDecomposition where the pass would defer to synthesized gates if the output from synthesis was ideal even if the input gate was also ideal. This typically comes up in simulators where there are no error rates for any gates and everything is listed as ideal. This would cause the transpiler to translate gates unnecessarily which was unexpected. This commit fixes this by adding an additional check to the subsitution condition to ensure we're only substituting a gate if it's not in the target (when they're all ideal). Fixes #10568 * Update releasenotes/notes/fix-optimize-1q-sim-407b88e45e6062b6.yaml Co-authored-by: Jake Lishman <jake@binhbar.com> * Update qiskit/transpiler/passes/optimization/optimize_1q_decomposition.py * Fix formatting --------- Co-authored-by: Jake Lishman <jake@binhbar.com> (cherry picked from commit f12db3b)
github-merge-queue bot
pushed a commit
that referenced
this pull request
Dec 1, 2023
…ion (#11351) (#11359) * Don't substitute ideal gates in target with Optimize1qGatesDecomposition This commit fixes an issue in Optimize1qGatesDecomposition where the pass would defer to synthesized gates if the output from synthesis was ideal even if the input gate was also ideal. This typically comes up in simulators where there are no error rates for any gates and everything is listed as ideal. This would cause the transpiler to translate gates unnecessarily which was unexpected. This commit fixes this by adding an additional check to the subsitution condition to ensure we're only substituting a gate if it's not in the target (when they're all ideal). Fixes #10568 * Update releasenotes/notes/fix-optimize-1q-sim-407b88e45e6062b6.yaml Co-authored-by: Jake Lishman <jake@binhbar.com> * Update qiskit/transpiler/passes/optimization/optimize_1q_decomposition.py * Fix formatting --------- Co-authored-by: Jake Lishman <jake@binhbar.com> (cherry picked from commit f12db3b) Co-authored-by: Matthew Treinish <mtreinish@kortar.org>
mtreinish
added a commit
to mtreinish/qiskit-aer
that referenced
this pull request
May 17, 2024
This commit removes the qiskit aer translation plugin. This was added as a workaround for a Qiskit/qiskit#11351. This has been fixed in Qiskit since 0.45.2 and is no longer necessary. The mechanism by which the workaround worked was unsound in practice as it was mutating the target and also explicitly using private attributes of the Target. This is causing real issues now as it only worked by assuming the target wasn't shared between passmanagers which is never guaranteed. Similarly the reliance on internal private attributes of the Target class will cause issues in the future when the target internals change (see Qiskit/qiskit#12292). This commit opts to remove the plugin in its entirity as it's no longer necessary and actively causing issues with Qiskit 1.1 and transpiling targeting aer backends with >1 circuit. As it's private internal detail there isn't a release note. Fixes Qiskit/qiskit#12425 Fixes Qiskit#2141
mtreinish
added a commit
to mtreinish/qiskit-aer
that referenced
this pull request
May 17, 2024
This commit removes the qiskit aer translation plugin. This was added as a workaround for a Qiskit/qiskit#11351. This has been fixed in Qiskit since 0.45.2 and is no longer necessary. The mechanism by which the workaround worked was unsound in practice as it was mutating the target and also explicitly using private attributes of the Target. This is causing real issues now as it only worked by assuming the target wasn't shared between passmanagers which is never guaranteed. Similarly the reliance on internal private attributes of the Target class will cause issues in the future when the target internals change (see Qiskit/qiskit#12292). This commit opts to remove the plugin in its entirity as it's no longer necessary and actively causing issues with Qiskit 1.1 and transpiling targeting aer backends with >1 circuit. As it's private internal detail there isn't a release note. Fixes Qiskit/qiskit#12425 Fixes Qiskit#2141
doichanj
pushed a commit
to Qiskit/qiskit-aer
that referenced
this pull request
May 20, 2024
This commit removes the qiskit aer translation plugin. This was added as a workaround for a Qiskit/qiskit#11351. This has been fixed in Qiskit since 0.45.2 and is no longer necessary. The mechanism by which the workaround worked was unsound in practice as it was mutating the target and also explicitly using private attributes of the Target. This is causing real issues now as it only worked by assuming the target wasn't shared between passmanagers which is never guaranteed. Similarly the reliance on internal private attributes of the Target class will cause issues in the future when the target internals change (see Qiskit/qiskit#12292). This commit opts to remove the plugin in its entirity as it's no longer necessary and actively causing issues with Qiskit 1.1 and transpiling targeting aer backends with >1 circuit. As it's private internal detail there isn't a release note. Fixes Qiskit/qiskit#12425 Fixes #2141
doichanj
pushed a commit
to doichanj/qiskit-aer
that referenced
this pull request
May 23, 2024
This commit removes the qiskit aer translation plugin. This was added as a workaround for a Qiskit/qiskit#11351. This has been fixed in Qiskit since 0.45.2 and is no longer necessary. The mechanism by which the workaround worked was unsound in practice as it was mutating the target and also explicitly using private attributes of the Target. This is causing real issues now as it only worked by assuming the target wasn't shared between passmanagers which is never guaranteed. Similarly the reliance on internal private attributes of the Target class will cause issues in the future when the target internals change (see Qiskit/qiskit#12292). This commit opts to remove the plugin in its entirity as it's no longer necessary and actively causing issues with Qiskit 1.1 and transpiling targeting aer backends with >1 circuit. As it's private internal detail there isn't a release note. Fixes Qiskit/qiskit#12425 Fixes Qiskit#2141
doichanj
added a commit
to Qiskit/qiskit-aer
that referenced
this pull request
May 31, 2024
* release 0.14.2 * Disable test shot_branching on MacOS (#2143) * remove test_shot_branching * disable sho_branching tests on MacOS * fix test * fix test * format * fix test_shot_branching * fix test_runtime_parameterization * fix tests for the latest Qiskit (#2138) * Remove qiskit aer translation stage (#2142) This commit removes the qiskit aer translation plugin. This was added as a workaround for a Qiskit/qiskit#11351. This has been fixed in Qiskit since 0.45.2 and is no longer necessary. The mechanism by which the workaround worked was unsound in practice as it was mutating the target and also explicitly using private attributes of the Target. This is causing real issues now as it only worked by assuming the target wasn't shared between passmanagers which is never guaranteed. Similarly the reliance on internal private attributes of the Target class will cause issues in the future when the target internals change (see Qiskit/qiskit#12292). This commit opts to remove the plugin in its entirity as it's no longer necessary and actively causing issues with Qiskit 1.1 and transpiling targeting aer backends with >1 circuit. As it's private internal detail there isn't a release note. Fixes Qiskit/qiskit#12425 Fixes #2141 * Fix issue 2084 again (#2119) * Fix issue 2084 again * format * fix test * fix test * Always hook omp functions in Mac (#2128) * always hook omp functions in Mac * fix recent test failures with the latest qiskit --------- Co-authored-by: Jun Doi <doichan@jp.ibm.com> * Add simulator_metadata in metadata of SamplerV2 (#2109) Co-authored-by: Hiroshi Horii <hhorii@users.noreply.github.com> * Fix cuStateVec_enable option (#2146) * Fix cuStateVec_enable option * fix function name * Add support for rotation gates (#2147) * add rotation gates * add cr* gates to operations.hpp * fix test for stabilkize/extended-stabilizer * fix test again * Fix deterministic measure of stabilizer (#2132) * Fix deterministic measure of stabilizer * fix accumulation in loop * format --------- Co-authored-by: Hiroshi Horii <hhorii@users.noreply.github.com> * Fix init of EstimatorV2 and SamplerV2 (#2120) * fix init of EstimatorV2 and SamplerV2 to handle method * add release note * fix from_backend and example in README * remove setting density_matrix * Fix deploy.yml (#2110) Co-authored-by: Hiroshi Horii <hhorii@users.noreply.github.com> Co-authored-by: Luciano Bello <bel@zurich.ibm.com> * Add support for ECR gate to MPS (#2137) * Add support for ECR gate * replace is not to != * fix release note * Fix ecr implementation for stabilizer/extended-stabilizer * Handle gates with ctrl_stete=0 (#2148) * handle cx gate with ctrl_stete=0 * add handling *_o0, add test for cx_o0 * format * fix test * fix test * fix test cx * fix to handle multiple control states * fix ctrl_state * format * resolve conflict again * fix aer_compiler * fix random seed (#2151) Co-authored-by: Hiroshi Horii <hhorii@users.noreply.github.com> * Move delay gate to custom instructions (#2153) --------- Co-authored-by: Matthew Treinish <mtreinish@kortar.org> Co-authored-by: Hiroshi Horii <hhorii@users.noreply.github.com> Co-authored-by: Luciano Bello <bel@zurich.ibm.com> Co-authored-by: Takashi Imamichi <31178928+t-imamichi@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
This commit fixes an issue in Optimize1qGatesDecomposition where the pass would defer to synthesized gates if the output from synthesis was ideal even if the input gate was also ideal. This typically comes up in simulators where there are no error rates for any gates and everything is listed as ideal. This would cause the transpiler to translate gates unnecessarily which was unexpected. This commit fixes this by adding an additional check to the substitution condition to ensure we're only substituting a gate if it's not in the target (when they're all ideal).
Details and comments
Fixes #10568