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.

Sign up for GitHub

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

Jump to bottom

Migrate to Target as only backend input for transpiler passes #9256

Open
9 of 12 tasks
mtreinish opened this issue Dec 6, 2022 · 0 comments
Open
9 of 12 tasks

Migrate to Target as only backend input for transpiler passes #9256

mtreinish opened this issue Dec 6, 2022 · 0 comments
Assignees
@mtreinish @ElePT
Labels
mod: transpiler Issues and PRs related to Transpiler priority: medium type: epic A theme of work that contain sub-tasks type: feature request New feature or request
Milestone
2.0.0

Comments

@mtreinish
Copy link
Member

mtreinish commented Dec 6, 2022 •
edited
Loading

What should we add?

Right now we have a mix of the legacy approach to modeling a backend in the transpiler (basis_gates, CouplingMap, etc) and the new approach, the Target class. This causes friction when adding new passes or maintaining existing ones because we have to support dual paths for everything. We unify all of the internals on the new model because in almost every case we can convert from the legacy model to the target without issue.

After the above is done and released then we can start the deprecation process for the legacy inputs to the transpiler passes along with the PassManager config.
@mtreinish mtreinish added type: feature request New feature or request Epic mod: transpiler Issues and PRs related to Transpiler labels Dec 6, 2022
@mtreinish mtreinish added this to the 0.24.0 milestone Dec 6, 2022
mtreinish added a commit to mtreinish/qiskit-core that referenced this issue Dec 7, 2022
@mtreinish
Make all Layout and Routing passes target aware
ed529f1 
This commit updates all the layout and routing passes which leverage
hardware constraints to have a new optional keyword argument, target,
to specify a Target object for modelling the constraints of the target
backend. This option will superscede any other specified arguments for
backend constraints. For most of these passes the target argument is
just internally leveraged to build a CouplingMap (and in 2 cases a
BackendProperties) as most of the algorithms don't need more than the
global connectivity graph and there is minimal overhead to convert
from a target to a CouplingMap. The 2 passes which take backend properties
should be refactored to work natively with the target because the
conversion adds extra overhead which isn't needed, but in those cases
the passes aren't enabled by default (or are slated to be moved out of
tree as a plugin, see Qiskit#8662) so the extra overhead isn't of particular
concern.

This is part of the first step towards making all backend constraints
for the transpiler used the Target internally (see Qiskit#9256). Once all
passes that take hardware constraints parameters as an input are updated
to have a target we can start internally using the Target only for all
preset pass manager construction and start the long process of
deprecating the legacy interface in these passes.

Related to Qiskit#9256
@mtreinish mtreinish mentioned this issue Dec 7, 2022
Merged
mtreinish added a commit to mtreinish/qiskit-core that referenced this issue Jan 4, 2023
@mtreinish
Add target support to basis, optimization, and util passes
42ff7c2 
This commit updates all the basis, optimization, and util passes which
leverage hardware constraints to have a new optional keyword argument,
target, to specify a Target object for modelling the constraints of the
target backend. This option will superscede any other specified arguments
for backend constraints.

With this and Qiskit#9263 all the passes except for the scheduling passes
which take in backend constraints are made target aware. Making all the
passes target aware is the first step towards making all backend
constraints for the transpiler used the Target internally (see Qiskit#9256).
Once all passes that take hardware constraints parameters as an input are
updated to have a target we can start internally using the Target only for
all preset pass manager construction and start the long process of
deprecating the legacy interface in these passes.
@mtreinish mtreinish mentioned this issue Jan 4, 2023
Merged
mergify bot added a commit that referenced this issue Jan 20, 2023
@mtreinish @mergify
Add target support to basis, optimization, scheduling, and util passes (
f061be8 
#9343)

* Add target support to basis, optimization, and util passes

This commit updates all the basis, optimization, and util passes which
leverage hardware constraints to have a new optional keyword argument,
target, to specify a Target object for modelling the constraints of the
target backend. This option will superscede any other specified arguments
for backend constraints.

With this and #9263 all the passes except for the scheduling passes
which take in backend constraints are made target aware. Making all the
passes target aware is the first step towards making all backend
constraints for the transpiler used the Target internally (see #9256).
Once all passes that take hardware constraints parameters as an input are
updated to have a target we can start internally using the Target only for
all preset pass manager construction and start the long process of
deprecating the legacy interface in these passes.

* Add target to scheduling passes too

* Update preset pass manager drawer tests

* Add unit tests for non-scheduling passes

* Add tests for scheduling passes

* Fix lint

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
@mtreinish mtreinish mentioned this issue Mar 13, 2023
Closed
mtreinish added a commit to mtreinish/qiskit-core that referenced this issue Mar 13, 2023
@mtreinish
Fix basis_gates and coupling_map backend override in transpile()
eaf8dca 
This commit fixes an issue in the transpile() function when a user
specified the `backend` argument with a BackendV2 based backend along
with `basis_gates` or `coupling_map`. In this case the `transpile()` was
generating the preset pass manager with a target, coupling map, and
basis gates list. However, most individual passes that take basis gates
and a Target will prefer to use the target if both are specified. This
is generally sane behavior at the pass level because the target contains
more rich data and tighter constraints that a transpiler pass will need
to worry about. To fix this limitation this commit updates transpile()
to not use the backend's target if either basis_gates or coupling_map
are specified.

Longer term this should no longer be an issue when Qiskit#9256 is implemented
and we'll be relying solely on a target internally. But this fix is
needed until Qiskit#9256 is started.

Fixes Qiskit#9781
@mtreinish mtreinish mentioned this issue Mar 13, 2023
Merged
mergify bot added a commit that referenced this issue Mar 14, 2023
@mtreinish @mergify
Fix basis_gates and coupling_map backend override in transpile() (#9789)
bf3bb96 
This commit fixes an issue in the transpile() function when a user
specified the `backend` argument with a BackendV2 based backend along
with `basis_gates` or `coupling_map`. In this case the `transpile()` was
generating the preset pass manager with a target, coupling map, and
basis gates list. However, most individual passes that take basis gates
and a Target will prefer to use the target if both are specified. This
is generally sane behavior at the pass level because the target contains
more rich data and tighter constraints that a transpiler pass will need
to worry about. To fix this limitation this commit updates transpile()
to not use the backend's target if either basis_gates or coupling_map
are specified.

Longer term this should no longer be an issue when #9256 is implemented
and we'll be relying solely on a target internally. But this fix is
needed until #9256 is started.

Fixes #9781

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
mergify bot pushed a commit that referenced this issue Mar 14, 2023
@mtreinish @mergify
Fix basis_gates and coupling_map backend override in transpile() (#9789)
a43c61d 
This commit fixes an issue in the transpile() function when a user
specified the `backend` argument with a BackendV2 based backend along
with `basis_gates` or `coupling_map`. In this case the `transpile()` was
generating the preset pass manager with a target, coupling map, and
basis gates list. However, most individual passes that take basis gates
and a Target will prefer to use the target if both are specified. This
is generally sane behavior at the pass level because the target contains
more rich data and tighter constraints that a transpiler pass will need
to worry about. To fix this limitation this commit updates transpile()
to not use the backend's target if either basis_gates or coupling_map
are specified.

Longer term this should no longer be an issue when #9256 is implemented
and we'll be relying solely on a target internally. But this fix is
needed until #9256 is started.

Fixes #9781

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
(cherry picked from commit bf3bb96)
mergify bot added a commit that referenced this issue Mar 14, 2023
@mergify @mtreinish
Fix basis_gates and coupling_map backend override in transpile() (#9789…
b220d07 
…) (#9792)

This commit fixes an issue in the transpile() function when a user
specified the `backend` argument with a BackendV2 based backend along
with `basis_gates` or `coupling_map`. In this case the `transpile()` was
generating the preset pass manager with a target, coupling map, and
basis gates list. However, most individual passes that take basis gates
and a Target will prefer to use the target if both are specified. This
is generally sane behavior at the pass level because the target contains
more rich data and tighter constraints that a transpiler pass will need
to worry about. To fix this limitation this commit updates transpile()
to not use the backend's target if either basis_gates or coupling_map
are specified.

Longer term this should no longer be an issue when #9256 is implemented
and we'll be relying solely on a target internally. But this fix is
needed until #9256 is started.

Fixes #9781

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
(cherry picked from commit bf3bb96)

Co-authored-by: Matthew Treinish <[email protected]>
adjs added a commit to comp-quantica/qiskit-terra that referenced this issue Mar 29, 2023
@adjs
Squashed commit of the following:
566be43 
commit 43a9dce
Author: Eric Arellano <[email protected]>
Date:   Tue Mar 28 15:27:07 2023 -0600

    Apply new deprecation decorators to algorithms folder (Qiskit#9864)

    * Apply new deprecation decorators to algorithms folder

    * Fix docstring starting on first line

    * Fix test due to positional args not triggering warning

    * Improve test further

commit 67eaa4b
Author: a-matsuo <[email protected]>
Date:   Wed Mar 29 04:50:41 2023 +0900

    Rearrange the gradient result based on the order of the input parameters (Qiskit#9503)

    * finite diff and spsa estimator gradient

    * for param shift, lin_comb, and reverse gradients

    * reverse estimator gradient

    * implemented the new feature

    * fix lint

    * fix test

    * fix and add reno

    * fix

    * fix style

commit 7e09125
Author: Eric Arellano <[email protected]>
Date:   Tue Mar 28 08:33:08 2023 -0600

    Add `@deprecate_func` and `@deprecate_arg` decorators (Qiskit#9676)

    * Add `@deprecate_func` and `@deprecate_arg` decorators

    * Update the release note

    * Apply suggestions from code review

    Co-authored-by: Luciano Bello <[email protected]>

    * lint

    * Use PyPI package_name rather than loose English

    ---------

    Co-authored-by: Luciano Bello <[email protected]>

commit 90b2439
Author: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Date:   Tue Mar 28 07:44:45 2023 -0400

    Bump indexmap from 1.9.2 to 1.9.3 (Qiskit#9858)

    Bumps [indexmap](https://github.com/bluss/indexmap) from 1.9.2 to 1.9.3.
    - [Release notes](https://github.com/bluss/indexmap/releases)
    - [Changelog](https://github.com/bluss/indexmap/blob/master/RELEASES.md)
    - [Commits](indexmap-rs/indexmap@1.9.2...1.9.3)

    ---
    updated-dependencies:
    - dependency-name: indexmap
      dependency-type: direct:production
      update-type: version-update:semver-patch
    ...

    Signed-off-by: dependabot[bot] <[email protected]>
    Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

commit be8eeff
Author: Evgenii Zheltonozhskii <[email protected]>
Date:   Tue Mar 28 12:59:41 2023 +0300

    Fix mypy errors (circuit) (Qiskit#8267)

    * Fix circuit mypy errors

    * Fix 2

    * Remove ignores

    * Remove ignores for sure

    * Remove TODO

    * fix for docs

    * Add annotation

    * Fix control bug, remove unnecessary annotation

    * Add test and fix description

    * Fix test

    * Reformat

    * Fix test description

    * Update releasenotes/notes/fix-control-with-string-parameter-4eb8a308170e08db.yaml

    Co-authored-by: Julien Gacon <[email protected]>

    * More minor fixes

    * Fix callable import

    * Fix Instruction import for sphinx

    * Replace list[Any] with list

    ---------

    Co-authored-by: Julien Gacon <[email protected]>

commit 3118683
Author: Elena Peña Tapia <[email protected]>
Date:   Tue Mar 28 11:51:20 2023 +0200

    Fix examples (Qiskit#9860)

commit a36416b
Author: Elena Peña Tapia <[email protected]>
Date:   Tue Mar 28 10:27:43 2023 +0200

    Remove factorizers and linear solvers from `algorithms` (Qiskit#9832)

    * Remove factorizers and linear solvers

    * Add reno

    * Remove shor from backend tests

    * Fix reno

    * Apply suggestions from code review

    Co-authored-by: Steve Wood <[email protected]>

    * Apply review reno

    ---------

    Co-authored-by: Steve Wood <[email protected]>

commit aacbc66
Author: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Date:   Mon Mar 27 19:13:15 2023 +0000

    Bump pyo3 from 0.18.1 to 0.18.2 (Qiskit#9854)

    Bumps [pyo3](https://github.com/pyo3/pyo3) from 0.18.1 to 0.18.2.
    - [Release notes](https://github.com/pyo3/pyo3/releases)
    - [Changelog](https://github.com/PyO3/pyo3/blob/main/CHANGELOG.md)
    - [Commits](PyO3/pyo3@v0.18.1...v0.18.2)

    ---
    updated-dependencies:
    - dependency-name: pyo3
      dependency-type: direct:production
      update-type: version-update:semver-patch
    ...

    Signed-off-by: dependabot[bot] <[email protected]>
    Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

commit abfb53e
Author: Jake Lishman <[email protected]>
Date:   Mon Mar 27 15:06:28 2023 +0100

    Ensure up-to-date `pip`, `setuptools` and `wheel` in CI (Qiskit#9850)

    Up-to-date Python packages should not require this step, however there
    are several packages, especially those that are optional for Terra
    functionality, that do not yet contain `pyproject.toml` files when
    building them from source.  In these cases, `pip` will begin erroring
    out from version 23.1 if `wheel` is not installed.

    This commit proactively ensures that the minimum build dependencies for
    legacy Python packages is prepared and up-to-date before attempting
    installations; this includes ensuring that these are updated _inside_
    any created venvs as well as outside them.

commit 03473bd
Author: Elena Peña Tapia <[email protected]>
Date:   Mon Mar 27 12:23:12 2023 +0200

    pin symengine to `<0.10.0` (Qiskit#9857)

    * pin symengine

    * Avoid specific version

commit a1a67a1
Author: Matthew Treinish <[email protected]>
Date:   Fri Mar 24 14:34:53 2023 -0400

    Add support to CouplingMap for disjoint qubits (Qiskit#9710)

    * Add support to CouplingMap for disjoint qubits

    Previously the CouplingMap class only supported graphs which were fully
    connected. This prevented us from modeling potential hardware which
    didn't have a path between all qubits. This isn't an inherent limitation
    of the underlying graph data structure but was a limitation put on the
    CouplingMap class because several pieces of the transpiler assume a path
    always exists between 2 qubits (mainly in layout and routing). This
    commit removes this limitation and also adds a method to get a subgraph
    CouplingMap for all the components of the CouplingMap. This enables us
    to model these devices with a CouplingMap, which is the first step
    towards supporting these devices in the transpiler.

    One limitation with this PR is most fo the layout and routing algorithms
    do not support disjoint connectivity. The primary exception being
    TrivialLayout (although the output might be invalid) VF2Layout and
    VF2PostLayout which inherently support this already. This commit lays
    the groundwork to fix this limitation in a follow-up PR but for the time
    being it just raises an error in those passes if a disconnected
    CouplingMap is being used. The intent here is to follow up to this commit
    soon for adding support for SabreLayout, SabreSwap, DenseLayout,
    and StochasticSwap to leverage the method get_component_subgraphs added
    here to make them usable on such coupling maps.

    * Remove coupling map connected check from NoiseAdaptiveLayout

    Noise adaptive layout doesn't use a CouplingMap so we can't check for a
    disconnected coupling map in it.

    * Change DenseLayout guard to only prevent running when it won't work

    * Rename get_component_subgraphs to components and cache result

    This commit renames the get_component_subgraphs() method to components()
    which is much more consise name. At the same time this adds caching to
    the return just in case building the component subgraphs is expensive to
    compute we only need to ever do it once.

    * Drop caching of connected components

    * Fix check for dense layout to do a valid comparison

    * Ensure self loops in CouplingMap.distance() return 0

    In a previous commit the distance() method was updated to handle
    disjoint graphs correctly. Prior to this PR it was expected to raise
    when a path didn't exist between 2 qubits by nature of the distance
    matrix construction failing if there was a disconnected coupling map.
    Since that isn't the case after this PR the error condition was
    changed to check explicitly that there is no path available and then
    error. However, there was an issue in this case and self loops would
    incorrectly error as well when instead they should return 0. This commit
    updates the error check to ignore self loops so they return correctly.

    * Fix lint

    * Update CouplingMap.components() docstring

    Co-authored-by: Kevin Krsulich <[email protected]>

    * Expand test coverage

    * Remove unused option for strongly connected components

    * Expand docstring to explain return list order

    * Use infinity for disconnected nodes in distance matrix

    * Update releasenotes/notes/add-cmap-componets-7ed56cdf294150f1.yaml

    * Rename CouplingMap.components to connected_components()

    THis commit renames the CouplingMap.components() method to
    connected_components(). It also adds an example to the docstring to
    better explain what a connected component is.

    * Fix typo in relaese note

    * Update method name in release note

    * Restore previous reduce() behavior

    The current reduce() error behavior of raising on trying to reduce to a
    disconnected coupling map is being depended on in other locations. To
    avoid a potentially breaking change this commit reverts the removal of
    that limitation in the method. We can look at doing that in the future
    independently of this PR because removing this specific restriction on
    the reduce() method is not 100% tied to generally allowing disconnected
    coupling map objects.

    * Add missing import

    ---------

    Co-authored-by: Kevin Krsulich <[email protected]>

commit 0bd4cfd
Author: Matthew Treinish <[email protected]>
Date:   Fri Mar 24 10:48:21 2023 -0400

    Add option to skip deepcopy on dag_to_circuit (Qiskit#9825)

    * Add option to skip deepcopy on dag_to_circuit

    This commit adds a new option, copy_operations, which allows to disable
    the deepcopy on dag_to_circuit(). Previously the dag to circuit
    converter would always deep copy the operations, however in the
    transpiler we don't need this extra overhead because the DAGCircuits are
    temporary internal artifacts and will not be resused outside of the
    transpiler. So we can avoid the copy overhead in these situations,
    however in general the converter needs to default to copying. This
    new argument enables this workflow by keeping the current behavior by
    default but enabling the transpiler to skip copying operations in the
    output conversion.

    * Fix docs typo

    * Fix docs typo again

    * Use copy_operations=False in more locations

commit dfcfa4a
Author: Jake Lishman <[email protected]>
Date:   Fri Mar 24 13:16:58 2023 +0000

    Remove dependency on sphinx-autodoc-typehints (Qiskit#9705)

    * Remove dependency on sphinx-autodoc-typehints

    This replaces the whole usage of the separate package
    `sphinx-autodoc-typehints` with the built-in support from `autodoc`,
    available (with this combination of options) since Sphinx 5.0.  This
    removes a docs dependency that has caused us significant problems before
    due to its instability, and the built-in handling in `autodoc` is much
    cleaner as well; rather than trying to textually modify the docstring,
    this version (correctly) mutates the parsed rST structure, which also
    naturally it 100% compatible with the Napoleon processing.

    * Maintain previous type-hint behaviour

commit 4a63b17
Author: Jake Lishman <[email protected]>
Date:   Fri Mar 24 13:14:07 2023 +0000

    Move OQ3 exporter tests to qasm3 module (Qiskit#9844)

    * Move OQ3 exporter tests to qasm3 module

    These previously lived in `test/python/circuit` just for historical
    reasons (the OQ2 exporter lives there because it's a method on
    `QuantumCircuit`).  Since we have a dedicated `qasm3` module now,
    including in the test suite, it makes sense to move the file there.

    * Add missing __init__.py

commit 9c8eb06
Author: Julien Gacon <[email protected]>
Date:   Fri Mar 24 08:46:38 2023 +0100

    Improve ``Parameter`` handling in ``SparsePauliOp`` (Qiskit#9796)

    * add reno

    * Add assign_parameters and parameter in init

    * add SPO.parameters and remove utils

    * fix ParameterValueType typehint

    * Update qiskit/quantum_info/operators/symplectic/sparse_pauli_op.py

    Co-authored-by: Ikko Hamamura <[email protected]>

    * remove trailing print

    * Elena's comments

    Co-authored-by: Elena Peña Tapia <[email protected]>

    * fix line length

    ---------

    Co-authored-by: Ikko Hamamura <[email protected]>
    Co-authored-by: Elena Peña Tapia <[email protected]>

commit 7bb4af9
Author: Eric Arellano <[email protected]>
Date:   Thu Mar 23 09:24:43 2023 -0600

    Remove tweedledum from requirements-dev.txt (Qiskit#9759)

commit 5a19f8a
Author: Jake Lishman <[email protected]>
Date:   Thu Mar 23 13:57:23 2023 +0000

    Fix missing handling of no-synthesis case in `UnitarySynthesis` (Qiskit#9843)

    Since moving `Optimize1qDecomposition` down to the Rust level, some of
    its use in the default `UnitarySynthesis` plugin was allowing `None` to
    be passed to an internal function unchecked.  This correctly returns the
    "could not synthesise" value when appropriate.

commit 84a0fdd
Author: Takashi Imamichi <[email protected]>
Date:   Thu Mar 23 10:37:34 2023 +0900

    Performance improvement of `SparsePauliOp.to_matrix` (Qiskit#9620)

    * optimize SparsePauliOp.to_matrix

    * optimize dense

    * proper vectorize

    * update copyright year

    * small opt

    * update a comment

commit 262cf08
Author: Jake Lishman <[email protected]>
Date:   Wed Mar 22 16:11:35 2023 +0000

    Revert "Start deprecation of complex parameters in Qiskit Pulse (Qiskit#9735)" (Qiskit#9838)

    This reverts commit eb4c5cb.

    This is currently causing flaky CI on Windows, so is being temporarily
    reverted while a fix can be made.

commit 789d588
Author: Steve Wood <[email protected]>
Date:   Wed Mar 22 09:28:19 2023 -0400

    Update codeowners (Qiskit#9835)

commit eb4c5cb
Author: TsafrirA <[email protected]>
Date:   Tue Mar 21 03:10:59 2023 +0200

    Start deprecation of complex parameters in Qiskit Pulse (Qiskit#9735)

    * Add warning and update tests

    * Release notes

    * Move warning to all Pulse objects.

    * Fix releasenote typo

    * Move warning to format_parameter_value

commit 6cec912
Author: Matthew Treinish <[email protected]>
Date:   Mon Mar 20 13:26:41 2023 -0400

    Fix non-complex dtypes in OneQubitEulerDecomposer methods (Qiskit#9828)

    * Fix non-complex dtypes in OneQubitEulerDecomposer methods

    This commit fixes a regression introduced in the rewrite of the
    internals of the OneQubitEulerDecomposer class to be in rust. Previously
    the angles() and angles_and_phase() methods of OneQubitEulerDecomposer
    would work with a non-complex dtype in the input, however because the
    rust component of the module is strictly typed to only work with a
    complex128 dtype passing the array to rust to compute the angles and
    phase of the unitary would fail. To address this limitation this commit
    casts the input matrix to be complex before passing it to the rust
    function.

    Fixes Qiskit#9827

    * Cleanup release note

    * Update test/python/quantum_info/test_synthesis.py

    Co-authored-by: Jake Lishman <[email protected]>

    ---------

    Co-authored-by: Jake Lishman <[email protected]>

commit 9500f42
Author: Etienne Wodey <[email protected]>
Date:   Mon Mar 20 15:30:55 2023 +0100

    transpile: narrow the return type depending on the circuits argument (Qiskit#9799)

    * transpile: narrow the return type depending on the circuits argument

    * Fixup release note

    ---------

    Co-authored-by: Jake Lishman <[email protected]>

commit 8ceb57d
Author: Takashi Imamichi <[email protected]>
Date:   Mon Mar 20 22:38:07 2023 +0900

    fix qasm with reset (Qiskit#9819)

commit e2f6606
Author: Eric Arellano <[email protected]>
Date:   Fri Mar 17 11:36:27 2023 -0600

    Consolidate `pip install` calls in CI (Qiskit#9757)

    * Remove unnecessary `pip install` in CI

    * Consolidate `pip install` calls in CI

    * Add back `-U`

    * Go back to editable install so Rust is in debug mode

    * Add RUST_DEBUG env var so we can avoid editable install

    * Remove SETUPTOOLS_ENABLE_FEATURES and also Aer from lint job

    * See if editable install fixes Pylint

    * Give up on not using editable installs in CI

    I don't know why they're causing issues with loading the Rust extension. But it's not a priority
    to figure that out. This PR is still some forward progress.

commit 2afcebc
Author: Eric Arellano <[email protected]>
Date:   Fri Mar 17 11:34:25 2023 -0600

    Fix Tox installing Terra from PyPI (Qiskit#9758)

    * Fix Tox installing Terra from PyPI

    * Don't use editable mode

commit b00a8de
Author: Matthew Treinish <[email protected]>
Date:   Thu Mar 16 18:32:14 2023 -0400

    Stop using mergify merge queue feature (Qiskit#9807)

    Due to recent changes in the mergify configuration/service the merge
    queue feature of mergify is no longer working as expected. It's not
    clear if this just a temporary bug in mergify, a longer term change in
    behavior (which requires all PR authors to be registered on mergify), or
    some incompatibility between the github api and mergify. However,
    because we're no longer able to rely on it we've moved to using github's
    native merge queue feature [1][2] instead. Since this new merge queue
    would conflict with mergify's this commit removes the merge queue
    configuration from the mergify configuration. The configuration for
    stable backporting is left in place because github doesn't natively
    offer that feature and it appears to still work fine.

    [1] https://github.blog/changelog/2023-02-08-pull-request-merge-queue-public-beta/
    [2] https://docs.github.com/en/repositories/configuring-branches-and-merges-in-your-repository/configuring-pull-request-merges/managing-a-merge-queue

commit 07ca200
Author: Julien Gacon <[email protected]>
Date:   Thu Mar 16 20:20:04 2023 +0100

    Clip probabilities in ``QuantumState`` (Qiskit#9762)

    * Clip probabilities

    * clip implictly

    * add reno

    * add tests

    * add renormalization

    * use sum, not norm

    * ensure round still works

    * normalizing seems to re-introduce errors

    * Tighten floating-point tests

    In cases of rounding and clipping, it's important that the
    floating-point output is bit-for-bit correct, so the fuzzy tests weren't
    ideal (some of these were strict, but it was inconsistent).  It's better
    to use `assertEqual` rather than `assertTrue` where possible so we get
    better errors on failure.

    ---------

    Co-authored-by: Jake Lishman <[email protected]>

commit 5a9d041
Author: Jake Lishman <[email protected]>
Date:   Thu Mar 16 18:50:07 2023 +0000

    Remove useless `into_iter()` on `Range` (Qiskit#9806)

    Clippy has started complaining about this on the stable branch, and it's
    probably correct to do so.

commit bbd023b
Author: Matthew Treinish <[email protected]>
Date:   Thu Mar 16 14:44:44 2023 -0400

    Leverage Rust circuit sequence construction for `OneQubitEulerDecomposer` (Qiskit#9583)

    * Leverage Rust circuit sequence construction for OneQubitEulerDecomposer

    This commit is a follow-up to Qiskit#9578 which added a rust implementation
    of the second half of the single qubit euler decomposition routines and
    leveraged them for the Optimize1qGatesDecomposition transpiler pass.
    With that PR the Optimize1qGatesDecomposition no longer was dependent on
    the OneQubitEulerDecomposer class. This commit continues from that PR
    and updates the OneQubitEulerDecomposer to leverage the same Rust
    implementation internally. Calling a decomposer object will internally
    call the rust function to generate a circuit sequence and then return
    either a QuantumCircuit or DAGCircuit (depending on the options).
    Similarly all the angle computation is done directly in Rust.

    * Add missing atol clamping to mod_2pi

    The python version of the OneQubitEulerDecomposer class had atol
    clamping on it's output from mod_2pi, when this function was ported to
    rust this was not included. At the time it was because nothing set the
    atol parameter when calling mod_2pi (the angle calculation in Qiskit#9185 did
    not have atol and the expansion to construct circuits for
    Optimize1qGatesDecomposition always used the default value). However,
    now that we're expanding OneQubitEulerDecomposer to internally do all
    the calculations in rust we need to support an adjustable atol which
    includes the missing endpoint clamping in mod_2pi. This commit adds this
    missing functionality to the function.

    * Add docstring to mod_2pi rust function

    * Remove mod_2pi python function

commit 0d15506
Author: sg495 <[email protected]>
Date:   Thu Mar 16 18:40:33 2023 +0000

    new parameters in `plot_bloch_multivector`: `figsize`, `font_size`, and `title_font_size`  (Qiskit#7264)

    * Update state_visualization.py

    Closes Qiskit#7263

    * Introduced font size, title font size and title padding to plot_bloch_multivector

    * Update state_visualization.py

    Documented new kwargs.

    * Fixed code formatting issues in previous commits of Qiskit#7264.

    * Update qiskit/visualization/bloch.py

    Co-authored-by: Luciano Bello <[email protected]>

    * Update qiskit/visualization/bloch.py

    Co-authored-by: Luciano Bello <[email protected]>

    * Update qiskit/visualization/state_visualization.py

    Co-authored-by: Luciano Bello <[email protected]>

    * Update qiskit/visualization/state_visualization.py

    Co-authored-by: Luciano Bello <[email protected]>

    * Create update-state-visualization-6836bd53e3a24891.yaml

    Added release notes.

    * Update test_graph_matplotlib_drawer.py

    Included a test for the new features.

    * fix test

    * blacking

    ---------

    Co-authored-by: Luciano Bello <[email protected]>

commit a85456d
Author: Jim Garrison <[email protected]>
Date:   Thu Mar 16 14:40:16 2023 -0400

    Remove shared `black`/`lint/`lint-incr` `envdir` in `tox.ini` (Qiskit#9800)

    Under tox3, it was faster to have the `black`, `lint`, and `lint-incr`
    environments share an `envdir`.  However, tox4 recreates the environment
    every time one switches from one to another when they share an `envdir`,
    leading to much slower execution.  Here I've changed these environments
    to all use the default (unique) directory rather than a shared one.

    Closes Qiskit#9782

commit b047a7c
Author: Matthew Treinish <[email protected]>
Date:   Thu Mar 16 17:29:12 2023 -0400

    Attempt to trigger Azure pipelines on merge queue events (Qiskit#9809)

    * Attempt to trigger Azure pipelines on merge queue events

    This commit attempts to add a trigger condition to our azure pipelines
    job to trigger on commits to the gh merge queue branch.

    * Only trigger PR CI in queue, not push CI

    ---------

    Co-authored-by: Jake Lishman <[email protected]>

commit b4bc559
Author: Luciano Bello <[email protected]>
Date:   Thu Mar 16 15:56:55 2023 +0100

    Some doc formatting and link, mostly on quantumcircuit.py file (Qiskit#9771)

    * some doc formating and link, mostly on quantumcircuit.py file

    * Update qiskit/circuit/quantumcircuit.py

    Co-authored-by: Julien Gacon <[email protected]>

    * ident

    * more typos

    * more typos

    * another iteration

    * Bit and Register

    * notes on Register and Bit

    * Update qiskit/circuit/bit.py

    Co-authored-by: Eric Arellano <[email protected]>

    * Update qiskit/circuit/register.py

    Co-authored-by: Eric Arellano <[email protected]>

    ---------

    Co-authored-by: Julien Gacon <[email protected]>
    Co-authored-by: Eric Arellano <[email protected]>
    Co-authored-by: Jake Lishman <[email protected]>

commit c37cec2
Author: Jim Garrison <[email protected]>
Date:   Thu Mar 16 00:26:01 2023 -0400

    Remove implicit broadcasting from `Pauli`/label construction of `PauliList` (Qiskit#9779)

    * Remove implicit broadcasting from `Pauli`/label construction of `PauliList`

    * Update releasenotes/notes/paulilist-do-not-broadcast-from-paulis-96de3832fba21b94.yaml

    Co-authored-by: Jake Lishman <[email protected]>

    ---------

    Co-authored-by: Jake Lishman <[email protected]>
    Co-authored-by: Jake Lishman <[email protected]>

commit b0b5541
Author: Eric Arellano <[email protected]>
Date:   Wed Mar 15 14:09:13 2023 -0600

    `@deprecate_arguments` and `@deprecate_function` add deprecation to docstring (Qiskit#9790)

commit fa4a136
Author: Alexander Ivrii <[email protected]>
Date:   Wed Mar 15 00:21:35 2023 +0200

    Alternative construction mechanism for HLS config (Qiskit#9413)

    * Alternative construction mechanism for HLS config

    * improved how synthesis methods can be specified

    * fixes

    * pass over release notes

    * adding tests for short form

    * Updating the documentation of HLSConfig and HighLevelSynthesis

    * removing unnecessary import in documentation

    ---------

    Co-authored-by: Luciano Bello <[email protected]>

commit 6bb4957
Author: Luciano Bello <[email protected]>
Date:   Tue Mar 14 21:41:02 2023 +0100

    extend the documentation for qiskit.circuit.QuantumCircuit.measure (Qiskit#9698)

    * extend the documentation for qiskit.circuit.QuantumCircuit.measure

    * spelling

    * H

    * qregs and cregs

    * spelling

    * Update qiskit/circuit/quantumcircuit.py

    * Update qiskit/circuit/quantumcircuit.py

    * Update qiskit/circuit/quantumcircuit.py

    * Update qiskit/circuit/quantumcircuit.py

    * examples after Args/Return/Raises

    * Update qiskit/circuit/quantumcircuit.py

    Co-authored-by: Jake Lishman <[email protected]>

    * short description

    * better description

    * as equivalence

    * shorten it a bit

    * ::

    * raw string

    * lvert

    * Tweak grammar

    ---------

    Co-authored-by: Junye Huang <[email protected]>
    Co-authored-by: Jake Lishman <[email protected]>

commit bf3bb96
Author: Matthew Treinish <[email protected]>
Date:   Tue Mar 14 10:18:45 2023 -0400

    Fix basis_gates and coupling_map backend override in transpile() (Qiskit#9789)

    This commit fixes an issue in the transpile() function when a user
    specified the `backend` argument with a BackendV2 based backend along
    with `basis_gates` or `coupling_map`. In this case the `transpile()` was
    generating the preset pass manager with a target, coupling map, and
    basis gates list. However, most individual passes that take basis gates
    and a Target will prefer to use the target if both are specified. This
    is generally sane behavior at the pass level because the target contains
    more rich data and tighter constraints that a transpiler pass will need
    to worry about. To fix this limitation this commit updates transpile()
    to not use the backend's target if either basis_gates or coupling_map
    are specified.

    Longer term this should no longer be an issue when Qiskit#9256 is implemented
    and we'll be relying solely on a target internally. But this fix is
    needed until Qiskit#9256 is started.

    Fixes Qiskit#9781

    Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

commit 648da26
Author: Jake Lishman <[email protected]>
Date:   Tue Mar 14 13:25:52 2023 +0000

    Improve error messages in `Target`-based `GateDirection` pass (Qiskit#9787)

    * Improve error messages in `Target`-based `GateDirection` pass

    The `Target` version of `GateDirection` has more information available
    to make better error messages. It can distinguish cases where the
    failure is because a gate would be valid if flipped but the pass doesn't
    know how to do it, and the case where the gate wouldn't be valid in
    either direction.

    * Reword error message on failed flip

    ---------

    Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

commit 3005806
Author: Eric Arellano <[email protected]>
Date:   Mon Mar 13 13:30:26 2023 -0600

    Add `add_deprecation_to_docstring` for docsite deprecation support (Qiskit#9685)

    * Add deprecations to function docstring

    * Work around issue with multiline strings

    * Better variable name

    * Jake's review feedback

    * Make Napoleon check case-insensitive

    * Promote add_deprecation_to_docstring to be public

    It may be useful for Applications/Ecosystem, when they want this functionality but want to keep using their own deprecation decorators

    * Properly error when metadata line is the first line

    * Tests feedback: don't use helper function and simplify instructions to generate tests

    * Don't use the function with @deprecate_function and @deprecate_arguments yet

    This makes the PR a smaller diff, so easier to review and also less risk when we land that it breaks something. Now, this PR only adds new functionality, the public `add_deprecation_to_docstring` function

    * Update qiskit/utils/deprecation.py

    Co-authored-by: Luciano Bello <[email protected]>

    * Update qiskit/utils/deprecation.py

    Co-authored-by: Luciano Bello <[email protected]>

    * Add release note

    ---------

    Co-authored-by: Luciano Bello <[email protected]>

commit c2affb1
Author: Jake Lishman <[email protected]>
Date:   Mon Mar 13 16:38:36 2023 +0000

    Fix explicitly calibrated gates in `GateDirection` (Qiskit#9786)

    If there is an explicit calibration given, this overrides the generic
    information from the `CouplingMap` or the `Target` for that particular
    instance, since one can use pulse-level control to define gates on a
    circuit-by-circuit basis that are not generically available in a way
    that can be specified in the coupling or target.

commit 2ce129a
Author: Naoki Kanazawa <[email protected]>
Date:   Mon Mar 13 22:08:35 2023 +0900

    Fix unitary synthesis module to get proper error value when it's empty. (Qiskit#9774)

    Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

commit 6829bb1
Author: Kazuki Tsuoka <[email protected]>
Date:   Sun Mar 12 16:27:55 2023 +0900

    Fix `Parameter.is_real()` (Qiskit#9664)

    * fix bug

    * add reno

    * Update releasenotes/notes/fix-parameter-is_real-8b8f99811e58075e.yaml

    ---------

    Co-authored-by: Julien Gacon <[email protected]>

commit 5a62949
Author: Luciano Bello <[email protected]>
Date:   Fri Mar 10 23:40:07 2023 +0100

    Qiskit gates not `qelib1.inc` are now defined when dumped (Qiskit#9777)

    * non-standard gates

    * Qiskit gates not qelib1.inc are now defined when dumped

    ---------

    Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

commit 5ce80ab
Author: Daniel Puzzuoli <[email protected]>
Date:   Fri Mar 10 11:49:56 2023 -0800

    Implementing CouplingMap.__eq__ (Qiskit#9766)

    * implementing Coupling.__eq__

    * Update qiskit/transpiler/coupling.py

    Co-authored-by: Matthew Treinish <[email protected]>

    * adding comments to test, and adding release note

    * black formatting

    * updating CouplingMap.__eq__ to check edge list equality instead of isomorphism

    * updating release notes

    * updating test doc string

    * Update test/python/transpiler/test_coupling.py

    Co-authored-by: Matthew Treinish <[email protected]>

    * Update test/python/transpiler/test_coupling.py

    Co-authored-by: Matthew Treinish <[email protected]>

    * Update test/python/transpiler/test_coupling.py

    Co-authored-by: Matthew Treinish <[email protected]>

    * Update test/python/transpiler/test_coupling.py

    Co-authored-by: Matthew Treinish <[email protected]>

    ---------

    Co-authored-by: Matthew Treinish <[email protected]>

commit 332f4b2
Author: Jake Lishman <[email protected]>
Date:   Fri Mar 10 08:31:36 2023 +0000

    Fix zero-qubit `Pauli` label strings (Qiskit#9726)

    It was previously possible to construct a zero-qubit Pauli operator
    using the array form (`Pauli(([], []))`), or by empty-slicing an
    existing Pauli (`Pauli("IXZ")[[]]`).  This commit completes the set by
    making labels with no qubits work as well, for consistency.

    Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

commit 3284ea0
Author: Eric Arellano <[email protected]>
Date:   Thu Mar 9 17:56:54 2023 -0600

    Fix `tox -e docs` for macOS (Qiskit#9765)

    * Fix `tox -e docs` for macOS

    Co-authored-by: Jake Lishman <[email protected]>

    * Fix two issues. Duh, thanks Jake

    * Update Azure job to set RUST_DEBUG

    Not strictly necessary since Tox already sets it. But makes things consistent and avoids accidentally
    removing this in the future

    ---------

    Co-authored-by: Jake Lishman <[email protected]>

commit d272919
Author: Matthew Treinish <[email protected]>
Date:   Thu Mar 9 18:11:05 2023 -0500

    Suppress PulseSimulator deprecation warning in tests (Qiskit#9767)

    In the recently released qiskit-aer 0.12.0 the PulseSimulator backend
    (used as its name implies to simulate pulse schedules) was deprecated in
    favor of the actively maintained qiskit-dynamcis package. This
    deprecation is causing test failures when running our test suite with
    qiskit-aer installed because a single test was opportunistically using
    the PulseSimulator backend when aer was installed. This commit adds an
    assertion to catch the DeprecationWarning and adds a comment to rewrite
    the test to use dynamics. Although we may not wish to add the optional
    bidirectional dependency for a single test, in which case we should just
    delete the test. But that is a discussion for a follow up PR when CI is
    not blocked by an upstream package's release. This commit is just a
    short term workaround to unblock forward progress.

commit fa7ac68
Author: Eric Arellano <[email protected]>
Date:   Wed Mar 8 12:05:55 2023 -0600

    Don't require Tweedledum for `qiskit.circuit.classicalfunction` on import (Qiskit#9754)

    * Don't require Tweedledum for `qiskit.circuit.classicalfunction` on import

    * Review feedback

commit 9d33618
Author: Jake Lishman <[email protected]>
Date:   Wed Mar 8 15:47:39 2023 +0000

    Fix docs build on case-insensitive systems (Qiskit#9748)

    During the switchover of the `SymbolicPulse` classes in the pulse
    library, there are several objects documented under
    `qiskit.library.pulse` that differ only by capitalisation.  Under
    `autosummary`, these therefore generate files with names that differ
    only by capitalisation, which causes naming clashes on systems that have
    case-insensitive semantics, such as macOS.  This then causes warnings to
    be emitted from the Sphinx build, failing it.

    Autosummary has this setting that can be used as a workaround.  I went
    this route rather than converting the documentation to use more inline
    documentation for the library for now, because I expect the clash to be
    more short-lived, and I didn't want to make a change that meaningful
    impacts the style of the documentation without knowing the full context
    around it.

    Co-authored-by: Eric Arellano <[email protected]>

commit 16f6adb
Author: Junye Huang <[email protected]>
Date:   Wed Mar 8 14:09:39 2023 +0100

    Remove non-existent files from MANIFEST.in (Qiskit#9752)

    * remove qiskit/schemas

    * remove include cython files

    * remove include pickle files

commit fe12c26
Author: Eric Arellano <[email protected]>
Date:   Tue Mar 7 18:44:38 2023 -0600

    Tag Eric in Rust changes (Qiskit#9749)

commit c6cd0d5
Author: Jake Lishman <[email protected]>
Date:   Tue Mar 7 23:12:39 2023 +0000

    Add structure for multiple Rust crates (Qiskit#9742)

    * Add structure for multiplie Rust extension crates

    This is a precursor to adding an entirely separate and self-contained
    crate to handle parsing of OpenQASM 2 (not 3 - this is kind of like a
    trial run for that).  It could conceivably still live within
    `qiskit._accelerate`, but having separate crates helps us enforce more
    sane API barriers, and has improvements in the incremental build time
    when working on only one of the Rust extensions.

    The intent after this commit is still to have `qiskit._accelerate` as an
    easy catch-all for accelerators for Python.  Developers should not need
    to be concerned with defining a new crate for every new feature, and for
    the most part `_accelerate` is still logically interoperative within
    itself (for example, `NLayout` is used all over).  This is just laying
    out the groundwork so more complex crate additions _can_ also be made.

    Some of the niceties to do with Cargo workspaces only became stabilised
    in Rust 1.64.  In particular, inheritance from the workspace root for
    things like the package version, Rust version, and dependencies only
    came in then.  For now, the `workspace.packages` key in the root
    `Cargo.toml` is ignored with a small warning during the build, but I've
    put it in place mostly to keep track of what we can change once the MSRV
    becomes 1.64 or greater (likely not til at least Terra 0.26).

    * Add README.md to crates/accelerate

    * Correct licence metadata

commit 2553376
Author: Julien Gacon <[email protected]>
Date:   Tue Mar 7 13:15:22 2023 +0100

    Fix bitstring padding in the `BackendSampler` (Qiskit#9744)

    * fix padding

    * add reno

commit 5c49d14
Author: Etienne Wodey <[email protected]>
Date:   Mon Mar 6 20:25:02 2023 +0100

    test/transpiler: add regression tests for Qiskit#9701 (Qiskit#9739)

commit 80ca1f3
Author: Shelly Garion <[email protected]>
Date:   Mon Mar 6 19:23:14 2023 +0200

    Add inverse method to ECRGate (Qiskit#9733)

    * add inverse method to ECRGate

    * add release notes

    * Fix sphinx tags

    * Recategorise as bugfix

    ---------

    Co-authored-by: Julien Gacon <[email protected]>
    Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
    Co-authored-by: Jake Lishman <[email protected]>

commit 126c9ba
Author: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Date:   Mon Mar 6 14:12:24 2023 +0000

    Bump rayon from 1.6.1 to 1.7.0 (Qiskit#9737)

    Bumps [rayon](https://github.com/rayon-rs/rayon) from 1.6.1 to 1.7.0.
    - [Release notes](https://github.com/rayon-rs/rayon/releases)
    - [Changelog](https://github.com/rayon-rs/rayon/blob/master/RELEASES.md)
    - [Commits](rayon-rs/rayon@rayon-core-v1.6.1...rayon-core-v1.7.0)

    ---
    updated-dependencies:
    - dependency-name: rayon
      dependency-type: direct:production
      update-type: version-update:semver-minor
    ...

    Signed-off-by: dependabot[bot] <[email protected]>
    Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
    Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

commit 6d12cdd
Author: Junye Huang <[email protected]>
Date:   Mon Mar 6 13:57:14 2023 +0100

    Update slack invite link to https://qisk.it/join-slack (Qiskit#9736)

commit b99ef32
Author: Jake Lishman <[email protected]>
Date:   Sun Mar 5 15:59:19 2023 +0000

    Add docs team to tag list for docs configuation changes (Qiskit#9714)

    This adds the current docs team to the qiskit-bot configuration to get
    automatically tagged when changes to the `docs` folder are made.  For
    the most part, this should just be about the documentation build itself,
    which is under their umbrella, rather than the text of the API docs
    which is largely generated from Terra's Python source code.

    Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

commit dbdf867
Author: John Lapeyre <[email protected]>
Date:   Sat Mar 4 13:43:06 2023 -0500

    Fix typo in docstring (Qiskit#9730)
@kdk kdk modified the milestones: 0.24.0, 0.25.0 Apr 5, 2023
jakelishman pushed a commit that referenced this issue Apr 6, 2023
@mtreinish @ajavadia
Make all Layout and Routing passes target aware (#9263)
5672c70 
* Make all Layout and Routing passes target aware

This commit updates all the layout and routing passes which leverage
hardware constraints to have a new optional keyword argument, target,
to specify a Target object for modelling the constraints of the target
backend. This option will superscede any other specified arguments for
backend constraints. For most of these passes the target argument is
just internally leveraged to build a CouplingMap (and in 2 cases a
BackendProperties) as most of the algorithms don't need more than the
global connectivity graph and there is minimal overhead to convert
from a target to a CouplingMap. The 2 passes which take backend properties
should be refactored to work natively with the target because the
conversion adds extra overhead which isn't needed, but in those cases
the passes aren't enabled by default (or are slated to be moved out of
tree as a plugin, see #8662) so the extra overhead isn't of particular
concern.

This is part of the first step towards making all backend constraints
for the transpiler used the Target internally (see #9256). Once all
passes that take hardware constraints parameters as an input are updated
to have a target we can start internally using the Target only for all
preset pass manager construction and start the long process of
deprecating the legacy interface in these passes.

Related to #9256

* Update pass manager drawer reference dot files

The pass manager drawer tests were comparing the dot output for statically
built pass managers to test that the output visualization is generated
correctly. However, the new arguments for the taget are changing the
visualization (to show the new option) and caused these tests to fail.
This commit updates the reference images to include the new argument.

* Fix lint

* Optimize CheckMap

This commit optimizes the logic of the CheckMap pass. Previously we were
unecessarily creating a distance matrix from the coupling graph, and
with a target unecessarily creating a CouplingMap object. Instead of
doing this instead the pass now creates a list of bidirectional edges
for adjacent nodes in the coupling graph and then a set lookup is done
for each check instead of checking for a distance of 1 in the distance
matrix.

* Fix test failures

* Update qiskit/transpiler/passes/layout/full_ancilla_allocation.py

* Add test coverage for passes not run in preset pass managers

* Fix lint

* Fix reference dot for new preset pass managers

* Rework arguments to take CouplingMap or Target

This commit updates the pass constructors to use a single positional
argument for both the coupling map or a target. This replaces the
previous behavior where a new keyword argument was added to pass in the
target. The exception to this is passes where the target was used
differently than the coupling map for extra functionality.

* Expand test coverage

* Small typo fixes from review

* Fix lint and test failure

* Update qiskit/transpiler/preset_passmanagers/level0.py

Co-authored-by: Ali Javadi-Abhari <[email protected]>

* Update logic to favor target in all optimization levels

In an earlier commit we updated the logic in optimization level 0 to
favor target instead of a coupling map. But, that ignored the other
optimization levels. This commit corrects this oversight and adds the
same logically change to all the preset pass managers.

* Update release note for api changes in earlier iterations

* Update docstrings

* Use a single positional argument for embed pm function

---------

Co-authored-by: Ali Javadi-Abhari <[email protected]>
giacomoRanieri pushed a commit to giacomoRanieri/qiskit-terra that referenced this issue Apr 16, 2023
@mtreinish @ajavadia @giacomoRanieri
Make all Layout and Routing passes target aware (Qiskit#9263)
84f7201 
* Make all Layout and Routing passes target aware

This commit updates all the layout and routing passes which leverage
hardware constraints to have a new optional keyword argument, target,
to specify a Target object for modelling the constraints of the target
backend. This option will superscede any other specified arguments for
backend constraints. For most of these passes the target argument is
just internally leveraged to build a CouplingMap (and in 2 cases a
BackendProperties) as most of the algorithms don't need more than the
global connectivity graph and there is minimal overhead to convert
from a target to a CouplingMap. The 2 passes which take backend properties
should be refactored to work natively with the target because the
conversion adds extra overhead which isn't needed, but in those cases
the passes aren't enabled by default (or are slated to be moved out of
tree as a plugin, see Qiskit#8662) so the extra overhead isn't of particular
concern.

This is part of the first step towards making all backend constraints
for the transpiler used the Target internally (see Qiskit#9256). Once all
passes that take hardware constraints parameters as an input are updated
to have a target we can start internally using the Target only for all
preset pass manager construction and start the long process of
deprecating the legacy interface in these passes.

Related to Qiskit#9256

* Update pass manager drawer reference dot files

The pass manager drawer tests were comparing the dot output for statically
built pass managers to test that the output visualization is generated
correctly. However, the new arguments for the taget are changing the
visualization (to show the new option) and caused these tests to fail.
This commit updates the reference images to include the new argument.

* Fix lint

* Optimize CheckMap

This commit optimizes the logic of the CheckMap pass. Previously we were
unecessarily creating a distance matrix from the coupling graph, and
with a target unecessarily creating a CouplingMap object. Instead of
doing this instead the pass now creates a list of bidirectional edges
for adjacent nodes in the coupling graph and then a set lookup is done
for each check instead of checking for a distance of 1 in the distance
matrix.

* Fix test failures

* Update qiskit/transpiler/passes/layout/full_ancilla_allocation.py

* Add test coverage for passes not run in preset pass managers

* Fix lint

* Fix reference dot for new preset pass managers

* Rework arguments to take CouplingMap or Target

This commit updates the pass constructors to use a single positional
argument for both the coupling map or a target. This replaces the
previous behavior where a new keyword argument was added to pass in the
target. The exception to this is passes where the target was used
differently than the coupling map for extra functionality.

* Expand test coverage

* Small typo fixes from review

* Fix lint and test failure

* Update qiskit/transpiler/preset_passmanagers/level0.py

Co-authored-by: Ali Javadi-Abhari <[email protected]>

* Update logic to favor target in all optimization levels

In an earlier commit we updated the logic in optimization level 0 to
favor target instead of a coupling map. But, that ignored the other
optimization levels. This commit corrects this oversight and adds the
same logically change to all the preset pass managers.

* Update release note for api changes in earlier iterations

* Update docstrings

* Use a single positional argument for embed pm function

---------

Co-authored-by: Ali Javadi-Abhari <[email protected]>
king-p3nguin pushed a commit to king-p3nguin/qiskit-terra that referenced this issue May 22, 2023
@mtreinish @mergify @king-p3nguin
Fix basis_gates and coupling_map backend override in transpile() (Qis…
c2ccac1 
…kit#9789)

This commit fixes an issue in the transpile() function when a user
specified the `backend` argument with a BackendV2 based backend along
with `basis_gates` or `coupling_map`. In this case the `transpile()` was
generating the preset pass manager with a target, coupling map, and
basis gates list. However, most individual passes that take basis gates
and a Target will prefer to use the target if both are specified. This
is generally sane behavior at the pass level because the target contains
more rich data and tighter constraints that a transpiler pass will need
to worry about. To fix this limitation this commit updates transpile()
to not use the backend's target if either basis_gates or coupling_map
are specified.

Longer term this should no longer be an issue when Qiskit#9256 is implemented
and we'll be relying solely on a target internally. But this fix is
needed until Qiskit#9256 is started.

Fixes Qiskit#9781

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
king-p3nguin pushed a commit to king-p3nguin/qiskit-terra that referenced this issue May 22, 2023
@mtreinish @ajavadia @king-p3nguin
Make all Layout and Routing passes target aware (Qiskit#9263)
d1d272c 
* Make all Layout and Routing passes target aware

This commit updates all the layout and routing passes which leverage
hardware constraints to have a new optional keyword argument, target,
to specify a Target object for modelling the constraints of the target
backend. This option will superscede any other specified arguments for
backend constraints. For most of these passes the target argument is
just internally leveraged to build a CouplingMap (and in 2 cases a
BackendProperties) as most of the algorithms don't need more than the
global connectivity graph and there is minimal overhead to convert
from a target to a CouplingMap. The 2 passes which take backend properties
should be refactored to work natively with the target because the
conversion adds extra overhead which isn't needed, but in those cases
the passes aren't enabled by default (or are slated to be moved out of
tree as a plugin, see Qiskit#8662) so the extra overhead isn't of particular
concern.

This is part of the first step towards making all backend constraints
for the transpiler used the Target internally (see Qiskit#9256). Once all
passes that take hardware constraints parameters as an input are updated
to have a target we can start internally using the Target only for all
preset pass manager construction and start the long process of
deprecating the legacy interface in these passes.

Related to Qiskit#9256

* Update pass manager drawer reference dot files

The pass manager drawer tests were comparing the dot output for statically
built pass managers to test that the output visualization is generated
correctly. However, the new arguments for the taget are changing the
visualization (to show the new option) and caused these tests to fail.
This commit updates the reference images to include the new argument.

* Fix lint

* Optimize CheckMap

This commit optimizes the logic of the CheckMap pass. Previously we were
unecessarily creating a distance matrix from the coupling graph, and
with a target unecessarily creating a CouplingMap object. Instead of
doing this instead the pass now creates a list of bidirectional edges
for adjacent nodes in the coupling graph and then a set lookup is done
for each check instead of checking for a distance of 1 in the distance
matrix.

* Fix test failures

* Update qiskit/transpiler/passes/layout/full_ancilla_allocation.py

* Add test coverage for passes not run in preset pass managers

* Fix lint

* Fix reference dot for new preset pass managers

* Rework arguments to take CouplingMap or Target

This commit updates the pass constructors to use a single positional
argument for both the coupling map or a target. This replaces the
previous behavior where a new keyword argument was added to pass in the
target. The exception to this is passes where the target was used
differently than the coupling map for extra functionality.

* Expand test coverage

* Small typo fixes from review

* Fix lint and test failure

* Update qiskit/transpiler/preset_passmanagers/level0.py

Co-authored-by: Ali Javadi-Abhari <[email protected]>

* Update logic to favor target in all optimization levels

In an earlier commit we updated the logic in optimization level 0 to
favor target instead of a coupling map. But, that ignored the other
optimization levels. This commit corrects this oversight and adds the
same logically change to all the preset pass managers.

* Update release note for api changes in earlier iterations

* Update docstrings

* Use a single positional argument for embed pm function

---------

Co-authored-by: Ali Javadi-Abhari <[email protected]>
@kdk kdk added type: epic A theme of work that contain sub-tasks and removed Epic labels Jun 6, 2023
mtreinish added a commit to mtreinish/qiskit-core that referenced this issue Jun 15, 2023
@mtreinish
Remove list argument broadcasting and simplify transpile()
094fb0f 
This commit updates the transpile() function to no longer support
broadcast of lists of arguments. This functionality was deprecated in
the 0.23.0 release. As part of this removal the internals of the
transpile() function are simplified so we don't need to handle
broadcasting, building preset pass managers, parallel dispatch, etc
anymore as this functionality (without broadcasting) already exists
through the transpiler API. Besides greatly simplifying the transpile()
code and using more aspects of the public APIs that exist in the
qiskit.transpiler module, this commit also should fix the overhead we
have around parallel execution due to the complexity of supporting
broadcasting. This overhead was partially addressed before in Qiskit#7789
which leveraged shared memory to minimize the serialization time
necessary for IPC but by using `PassManager.run()` internally now all of
that overhead is removed as the initial fork will have all the necessary
context in each process from the start.

Three seemingly unrelated changes made here were necessary to support our
current transpile() API without building custom pass manager
construction.

The first is the handling of layout from intlist. The
current Layout class is dependent on a circuit because it maps Qubit
objects to a physical qubit index. Ideally the layout structure would
just map virtual indices to physical indices (see Qiskit#8060 for a similar
issue, also it's worth noting this is how the internal NLayout and QPY
represent layout), but because of the existing API the construction of
a Layout is dependent on a circuit. For the initial_layout argument when
running with multiple circuits to avoid the need to broadcasting the
layout construction for supported input types that need the circuit to
lookup the Qubit objects the SetLayout pass now supports taking in an
int list and will construct a Layout object at run time. This
effectively defers the Layout object creation for initial_layout to
run time so it can be built as a function of the circuit as the API
demands.

The second is the FakeBackend class used in some tests was constructing
invalid backends in some cases. This wasn't caught in the previous
structure because the backends were not actually being parsed by
transpile() previously which masked this issue. This commit fixes that
issue because PassManagerConfig.from_backend() was failing because of
the invalid backend construction.

The third issue is a new _skip_target private argument to
generate_preset_pass_manager() and PassManagerConfig. This was necessary
to recreate the behavior of transpile() when a user provides a BackendV2
and either `basis_gates` or `coupling_map` arguments. In general the
internals of the transpiler treat a target as higher priority because it
has more complete and restrictive constraints than the
basis_gates/coupling map objects. However, for transpile() if a
backendv2 is passed in for backend paired with coupling_map and/or
basis_gates the expected workflow is that the basis_gates and
coupling_map arguments take priority and override the equivalent
attributes from the backend. To facilitate this we need to block pulling
the target from the backend This should only be needed for a short
period of time as when Qiskit#9256 is implemented we'll just build a single
target from the arguments as needed.

Fixes Qiskit#7741
@mtreinish mtreinish mentioned this issue Jun 15, 2023
Merged
1 task
@mtreinish mtreinish self-assigned this Jun 21, 2023
@mtreinish mtreinish added this to the 0.45.0 milestone Jul 5, 2023
@alexanderivrii alexanderivrii mentioned this issue Jul 12, 2023
Closed
@mtreinish mtreinish mentioned this issue Jul 19, 2023
Closed
github-merge-queue bot pushed a commit that referenced this issue Jul 19, 2023
@mtreinish @jlapeyre @jakelishman
Remove list argument broadcasting and simplify transpile() (#10291)
346ca77 
* Remove list argument broadcasting and simplify transpile()

This commit updates the transpile() function to no longer support
broadcast of lists of arguments. This functionality was deprecated in
the 0.23.0 release. As part of this removal the internals of the
transpile() function are simplified so we don't need to handle
broadcasting, building preset pass managers, parallel dispatch, etc
anymore as this functionality (without broadcasting) already exists
through the transpiler API. Besides greatly simplifying the transpile()
code and using more aspects of the public APIs that exist in the
qiskit.transpiler module, this commit also should fix the overhead we
have around parallel execution due to the complexity of supporting
broadcasting. This overhead was partially addressed before in #7789
which leveraged shared memory to minimize the serialization time
necessary for IPC but by using `PassManager.run()` internally now all of
that overhead is removed as the initial fork will have all the necessary
context in each process from the start.

Three seemingly unrelated changes made here were necessary to support our
current transpile() API without building custom pass manager
construction.

The first is the handling of layout from intlist. The
current Layout class is dependent on a circuit because it maps Qubit
objects to a physical qubit index. Ideally the layout structure would
just map virtual indices to physical indices (see #8060 for a similar
issue, also it's worth noting this is how the internal NLayout and QPY
represent layout), but because of the existing API the construction of
a Layout is dependent on a circuit. For the initial_layout argument when
running with multiple circuits to avoid the need to broadcasting the
layout construction for supported input types that need the circuit to
lookup the Qubit objects the SetLayout pass now supports taking in an
int list and will construct a Layout object at run time. This
effectively defers the Layout object creation for initial_layout to
run time so it can be built as a function of the circuit as the API
demands.

The second is the FakeBackend class used in some tests was constructing
invalid backends in some cases. This wasn't caught in the previous
structure because the backends were not actually being parsed by
transpile() previously which masked this issue. This commit fixes that
issue because PassManagerConfig.from_backend() was failing because of
the invalid backend construction.

The third issue is a new _skip_target private argument to
generate_preset_pass_manager() and PassManagerConfig. This was necessary
to recreate the behavior of transpile() when a user provides a BackendV2
and either `basis_gates` or `coupling_map` arguments. In general the
internals of the transpiler treat a target as higher priority because it
has more complete and restrictive constraints than the
basis_gates/coupling map objects. However, for transpile() if a
backendv2 is passed in for backend paired with coupling_map and/or
basis_gates the expected workflow is that the basis_gates and
coupling_map arguments take priority and override the equivalent
attributes from the backend. To facilitate this we need to block pulling
the target from the backend This should only be needed for a short
period of time as when #9256 is implemented we'll just build a single
target from the arguments as needed.

Fixes #7741

* Fix _skip_target logic

* Fix InstructionScheduleMap handling with backendv2

* Fix test failure caused by exception being raised later

* Fix indentation error

* Update qiskit/providers/fake_provider/fake_backend.py

Co-authored-by: John Lapeyre <[email protected]>

* Fix standalone dt argument handling

* Remove unused code

* Fix lint

* Remove duplicate import in set_layout.py

A duplicate import slipped through in the most recent rebase.
This commit fixes that oversight and removes the duplicate.

* Update release notes

Co-authored-by: Jake Lishman <[email protected]>

* Adjust logic for _skip_transpile to check if None

* Simplify check cmap code

* Only check backend if it exists

---------

Co-authored-by: John Lapeyre <[email protected]>
Co-authored-by: Jake Lishman <[email protected]>
to24toro pushed a commit to to24toro/qiskit-terra that referenced this issue Aug 3, 2023
@mtreinish @jlapeyre
@jakelishman @to24toro
Remove list argument broadcasting and simplify transpile() (Qiskit#10291
62084b9 
)

* Remove list argument broadcasting and simplify transpile()

This commit updates the transpile() function to no longer support
broadcast of lists of arguments. This functionality was deprecated in
the 0.23.0 release. As part of this removal the internals of the
transpile() function are simplified so we don't need to handle
broadcasting, building preset pass managers, parallel dispatch, etc
anymore as this functionality (without broadcasting) already exists
through the transpiler API. Besides greatly simplifying the transpile()
code and using more aspects of the public APIs that exist in the
qiskit.transpiler module, this commit also should fix the overhead we
have around parallel execution due to the complexity of supporting
broadcasting. This overhead was partially addressed before in Qiskit#7789
which leveraged shared memory to minimize the serialization time
necessary for IPC but by using `PassManager.run()` internally now all of
that overhead is removed as the initial fork will have all the necessary
context in each process from the start.

Three seemingly unrelated changes made here were necessary to support our
current transpile() API without building custom pass manager
construction.

The first is the handling of layout from intlist. The
current Layout class is dependent on a circuit because it maps Qubit
objects to a physical qubit index. Ideally the layout structure would
just map virtual indices to physical indices (see Qiskit#8060 for a similar
issue, also it's worth noting this is how the internal NLayout and QPY
represent layout), but because of the existing API the construction of
a Layout is dependent on a circuit. For the initial_layout argument when
running with multiple circuits to avoid the need to broadcasting the
layout construction for supported input types that need the circuit to
lookup the Qubit objects the SetLayout pass now supports taking in an
int list and will construct a Layout object at run time. This
effectively defers the Layout object creation for initial_layout to
run time so it can be built as a function of the circuit as the API
demands.

The second is the FakeBackend class used in some tests was constructing
invalid backends in some cases. This wasn't caught in the previous
structure because the backends were not actually being parsed by
transpile() previously which masked this issue. This commit fixes that
issue because PassManagerConfig.from_backend() was failing because of
the invalid backend construction.

The third issue is a new _skip_target private argument to
generate_preset_pass_manager() and PassManagerConfig. This was necessary
to recreate the behavior of transpile() when a user provides a BackendV2
and either `basis_gates` or `coupling_map` arguments. In general the
internals of the transpiler treat a target as higher priority because it
has more complete and restrictive constraints than the
basis_gates/coupling map objects. However, for transpile() if a
backendv2 is passed in for backend paired with coupling_map and/or
basis_gates the expected workflow is that the basis_gates and
coupling_map arguments take priority and override the equivalent
attributes from the backend. To facilitate this we need to block pulling
the target from the backend This should only be needed for a short
period of time as when Qiskit#9256 is implemented we'll just build a single
target from the arguments as needed.

Fixes Qiskit#7741

* Fix _skip_target logic

* Fix InstructionScheduleMap handling with backendv2

* Fix test failure caused by exception being raised later

* Fix indentation error

* Update qiskit/providers/fake_provider/fake_backend.py

Co-authored-by: John Lapeyre <[email protected]>

* Fix standalone dt argument handling

* Remove unused code

* Fix lint

* Remove duplicate import in set_layout.py

A duplicate import slipped through in the most recent rebase.
This commit fixes that oversight and removes the duplicate.

* Update release notes

Co-authored-by: Jake Lishman <[email protected]>

* Adjust logic for _skip_transpile to check if None

* Simplify check cmap code

* Only check backend if it exists

---------

Co-authored-by: John Lapeyre <[email protected]>
Co-authored-by: Jake Lishman <[email protected]>
ElePT pushed a commit to ElePT/qiskit-ibm-provider that referenced this issue Oct 4, 2023
@mtreinish @jlapeyre @jakelishman
Remove list argument broadcasting and simplify transpile() (Qiskit/qi…
4dd6d7b 
…skit#10291)

* Remove list argument broadcasting and simplify transpile()

This commit updates the transpile() function to no longer support
broadcast of lists of arguments. This functionality was deprecated in
the 0.23.0 release. As part of this removal the internals of the
transpile() function are simplified so we don't need to handle
broadcasting, building preset pass managers, parallel dispatch, etc
anymore as this functionality (without broadcasting) already exists
through the transpiler API. Besides greatly simplifying the transpile()
code and using more aspects of the public APIs that exist in the
qiskit.transpiler module, this commit also should fix the overhead we
have around parallel execution due to the complexity of supporting
broadcasting. This overhead was partially addressed before in Qiskit/qiskit#7789
which leveraged shared memory to minimize the serialization time
necessary for IPC but by using `PassManager.run()` internally now all of
that overhead is removed as the initial fork will have all the necessary
context in each process from the start.

Three seemingly unrelated changes made here were necessary to support our
current transpile() API without building custom pass manager
construction.

The first is the handling of layout from intlist. The
current Layout class is dependent on a circuit because it maps Qubit
objects to a physical qubit index. Ideally the layout structure would
just map virtual indices to physical indices (see Qiskit/qiskit#8060 for a similar
issue, also it's worth noting this is how the internal NLayout and QPY
represent layout), but because of the existing API the construction of
a Layout is dependent on a circuit. For the initial_layout argument when
running with multiple circuits to avoid the need to broadcasting the
layout construction for supported input types that need the circuit to
lookup the Qubit objects the SetLayout pass now supports taking in an
int list and will construct a Layout object at run time. This
effectively defers the Layout object creation for initial_layout to
run time so it can be built as a function of the circuit as the API
demands.

The second is the FakeBackend class used in some tests was constructing
invalid backends in some cases. This wasn't caught in the previous
structure because the backends were not actually being parsed by
transpile() previously which masked this issue. This commit fixes that
issue because PassManagerConfig.from_backend() was failing because of
the invalid backend construction.

The third issue is a new _skip_target private argument to
generate_preset_pass_manager() and PassManagerConfig. This was necessary
to recreate the behavior of transpile() when a user provides a BackendV2
and either `basis_gates` or `coupling_map` arguments. In general the
internals of the transpiler treat a target as higher priority because it
has more complete and restrictive constraints than the
basis_gates/coupling map objects. However, for transpile() if a
backendv2 is passed in for backend paired with coupling_map and/or
basis_gates the expected workflow is that the basis_gates and
coupling_map arguments take priority and override the equivalent
attributes from the backend. To facilitate this we need to block pulling
the target from the backend This should only be needed for a short
period of time as when Qiskit/qiskit#9256 is implemented we'll just build a single
target from the arguments as needed.

Fixes Qiskit/qiskit#7741

* Fix _skip_target logic

* Fix InstructionScheduleMap handling with backendv2

* Fix test failure caused by exception being raised later

* Fix indentation error

* Update qiskit/providers/fake_provider/fake_backend.py

Co-authored-by: John Lapeyre <[email protected]>

* Fix standalone dt argument handling

* Remove unused code

* Fix lint

* Remove duplicate import in set_layout.py

A duplicate import slipped through in the most recent rebase.
This commit fixes that oversight and removes the duplicate.

* Update release notes

Co-authored-by: Jake Lishman <[email protected]>

* Adjust logic for _skip_transpile to check if None

* Simplify check cmap code

* Only check backend if it exists

---------

Co-authored-by: John Lapeyre <[email protected]>
Co-authored-by: Jake Lishman <[email protected]>
@mtreinish mtreinish modified the milestones: 0.45.0, 1.0.0pre1 Oct 6, 2023
ElePT pushed a commit to ElePT/qiskit-ibm-runtime that referenced this issue Oct 10, 2023
@mtreinish @jlapeyre @jakelishman
Remove list argument broadcasting and simplify transpile() (Qiskit/qi…
74f3933 
…skit#10291)

* Remove list argument broadcasting and simplify transpile()

This commit updates the transpile() function to no longer support
broadcast of lists of arguments. This functionality was deprecated in
the 0.23.0 release. As part of this removal the internals of the
transpile() function are simplified so we don't need to handle
broadcasting, building preset pass managers, parallel dispatch, etc
anymore as this functionality (without broadcasting) already exists
through the transpiler API. Besides greatly simplifying the transpile()
code and using more aspects of the public APIs that exist in the
qiskit.transpiler module, this commit also should fix the overhead we
have around parallel execution due to the complexity of supporting
broadcasting. This overhead was partially addressed before in Qiskit/qiskit#7789
which leveraged shared memory to minimize the serialization time
necessary for IPC but by using `PassManager.run()` internally now all of
that overhead is removed as the initial fork will have all the necessary
context in each process from the start.

Three seemingly unrelated changes made here were necessary to support our
current transpile() API without building custom pass manager
construction.

The first is the handling of layout from intlist. The
current Layout class is dependent on a circuit because it maps Qubit
objects to a physical qubit index. Ideally the layout structure would
just map virtual indices to physical indices (see Qiskit/qiskit#8060 for a similar
issue, also it's worth noting this is how the internal NLayout and QPY
represent layout), but because of the existing API the construction of
a Layout is dependent on a circuit. For the initial_layout argument when
running with multiple circuits to avoid the need to broadcasting the
layout construction for supported input types that need the circuit to
lookup the Qubit objects the SetLayout pass now supports taking in an
int list and will construct a Layout object at run time. This
effectively defers the Layout object creation for initial_layout to
run time so it can be built as a function of the circuit as the API
demands.

The second is the FakeBackend class used in some tests was constructing
invalid backends in some cases. This wasn't caught in the previous
structure because the backends were not actually being parsed by
transpile() previously which masked this issue. This commit fixes that
issue because PassManagerConfig.from_backend() was failing because of
the invalid backend construction.

The third issue is a new _skip_target private argument to
generate_preset_pass_manager() and PassManagerConfig. This was necessary
to recreate the behavior of transpile() when a user provides a BackendV2
and either `basis_gates` or `coupling_map` arguments. In general the
internals of the transpiler treat a target as higher priority because it
has more complete and restrictive constraints than the
basis_gates/coupling map objects. However, for transpile() if a
backendv2 is passed in for backend paired with coupling_map and/or
basis_gates the expected workflow is that the basis_gates and
coupling_map arguments take priority and override the equivalent
attributes from the backend. To facilitate this we need to block pulling
the target from the backend This should only be needed for a short
period of time as when Qiskit/qiskit#9256 is implemented we'll just build a single
target from the arguments as needed.

Fixes Qiskit/qiskit#7741

* Fix _skip_target logic

* Fix InstructionScheduleMap handling with backendv2

* Fix test failure caused by exception being raised later

* Fix indentation error

* Update qiskit/providers/fake_provider/fake_backend.py

Co-authored-by: John Lapeyre <[email protected]>

* Fix standalone dt argument handling

* Remove unused code

* Fix lint

* Remove duplicate import in set_layout.py

A duplicate import slipped through in the most recent rebase.
This commit fixes that oversight and removes the duplicate.

* Update release notes

Co-authored-by: Jake Lishman <[email protected]>

* Adjust logic for _skip_transpile to check if None

* Simplify check cmap code

* Only check backend if it exists

---------

Co-authored-by: John Lapeyre <[email protected]>
Co-authored-by: Jake Lishman <[email protected]>
@mtreinish mtreinish mentioned this issue Oct 11, 2023
Merged
ElePT pushed a commit to ElePT/qiskit that referenced this issue Oct 12, 2023
@mtreinish @jlapeyre @jakelishman
Remove list argument broadcasting and simplify transpile() (Qiskit#10291
78ebc61 
)

* Remove list argument broadcasting and simplify transpile()

This commit updates the transpile() function to no longer support
broadcast of lists of arguments. This functionality was deprecated in
the 0.23.0 release. As part of this removal the internals of the
transpile() function are simplified so we don't need to handle
broadcasting, building preset pass managers, parallel dispatch, etc
anymore as this functionality (without broadcasting) already exists
through the transpiler API. Besides greatly simplifying the transpile()
code and using more aspects of the public APIs that exist in the
qiskit.transpiler module, this commit also should fix the overhead we
have around parallel execution due to the complexity of supporting
broadcasting. This overhead was partially addressed before in Qiskit#7789
which leveraged shared memory to minimize the serialization time
necessary for IPC but by using `PassManager.run()` internally now all of
that overhead is removed as the initial fork will have all the necessary
context in each process from the start.

Three seemingly unrelated changes made here were necessary to support our
current transpile() API without building custom pass manager
construction.

The first is the handling of layout from intlist. The
current Layout class is dependent on a circuit because it maps Qubit
objects to a physical qubit index. Ideally the layout structure would
just map virtual indices to physical indices (see Qiskit#8060 for a similar
issue, also it's worth noting this is how the internal NLayout and QPY
represent layout), but because of the existing API the construction of
a Layout is dependent on a circuit. For the initial_layout argument when
running with multiple circuits to avoid the need to broadcasting the
layout construction for supported input types that need the circuit to
lookup the Qubit objects the SetLayout pass now supports taking in an
int list and will construct a Layout object at run time. This
effectively defers the Layout object creation for initial_layout to
run time so it can be built as a function of the circuit as the API
demands.

The second is the FakeBackend class used in some tests was constructing
invalid backends in some cases. This wasn't caught in the previous
structure because the backends were not actually being parsed by
transpile() previously which masked this issue. This commit fixes that
issue because PassManagerConfig.from_backend() was failing because of
the invalid backend construction.

The third issue is a new _skip_target private argument to
generate_preset_pass_manager() and PassManagerConfig. This was necessary
to recreate the behavior of transpile() when a user provides a BackendV2
and either `basis_gates` or `coupling_map` arguments. In general the
internals of the transpiler treat a target as higher priority because it
has more complete and restrictive constraints than the
basis_gates/coupling map objects. However, for transpile() if a
backendv2 is passed in for backend paired with coupling_map and/or
basis_gates the expected workflow is that the basis_gates and
coupling_map arguments take priority and override the equivalent
attributes from the backend. To facilitate this we need to block pulling
the target from the backend This should only be needed for a short
period of time as when Qiskit#9256 is implemented we'll just build a single
target from the arguments as needed.

Fixes Qiskit#7741

* Fix _skip_target logic

* Fix InstructionScheduleMap handling with backendv2

* Fix test failure caused by exception being raised later

* Fix indentation error

* Update qiskit/providers/fake_provider/fake_backend.py

Co-authored-by: John Lapeyre <[email protected]>

* Fix standalone dt argument handling

* Remove unused code

* Fix lint

* Remove duplicate import in set_layout.py

A duplicate import slipped through in the most recent rebase.
This commit fixes that oversight and removes the duplicate.

* Update release notes

Co-authored-by: Jake Lishman <[email protected]>

* Adjust logic for _skip_transpile to check if None

* Simplify check cmap code

* Only check backend if it exists

---------

Co-authored-by: John Lapeyre <[email protected]>
Co-authored-by: Jake Lishman <[email protected]>
@mtreinish mtreinish mentioned this issue Nov 21, 2023
Merged
@mtreinish mtreinish modified the milestones: 1.0.0pre1, 1.0.0 Nov 27, 2023
ElePT pushed a commit to ElePT/qiskit-ibm-runtime that referenced this issue Dec 8, 2023
@mtreinish @jlapeyre @jakelishman
Remove list argument broadcasting and simplify transpile() (Qiskit/qi…
27f563a 
…skit#10291)

* Remove list argument broadcasting and simplify transpile()

This commit updates the transpile() function to no longer support
broadcast of lists of arguments. This functionality was deprecated in
the 0.23.0 release. As part of this removal the internals of the
transpile() function are simplified so we don't need to handle
broadcasting, building preset pass managers, parallel dispatch, etc
anymore as this functionality (without broadcasting) already exists
through the transpiler API. Besides greatly simplifying the transpile()
code and using more aspects of the public APIs that exist in the
qiskit.transpiler module, this commit also should fix the overhead we
have around parallel execution due to the complexity of supporting
broadcasting. This overhead was partially addressed before in Qiskit/qiskit#7789
which leveraged shared memory to minimize the serialization time
necessary for IPC but by using `PassManager.run()` internally now all of
that overhead is removed as the initial fork will have all the necessary
context in each process from the start.

Three seemingly unrelated changes made here were necessary to support our
current transpile() API without building custom pass manager
construction.

The first is the handling of layout from intlist. The
current Layout class is dependent on a circuit because it maps Qubit
objects to a physical qubit index. Ideally the layout structure would
just map virtual indices to physical indices (see Qiskit/qiskit#8060 for a similar
issue, also it's worth noting this is how the internal NLayout and QPY
represent layout), but because of the existing API the construction of
a Layout is dependent on a circuit. For the initial_layout argument when
running with multiple circuits to avoid the need to broadcasting the
layout construction for supported input types that need the circuit to
lookup the Qubit objects the SetLayout pass now supports taking in an
int list and will construct a Layout object at run time. This
effectively defers the Layout object creation for initial_layout to
run time so it can be built as a function of the circuit as the API
demands.

The second is the FakeBackend class used in some tests was constructing
invalid backends in some cases. This wasn't caught in the previous
structure because the backends were not actually being parsed by
transpile() previously which masked this issue. This commit fixes that
issue because PassManagerConfig.from_backend() was failing because of
the invalid backend construction.

The third issue is a new _skip_target private argument to
generate_preset_pass_manager() and PassManagerConfig. This was necessary
to recreate the behavior of transpile() when a user provides a BackendV2
and either `basis_gates` or `coupling_map` arguments. In general the
internals of the transpiler treat a target as higher priority because it
has more complete and restrictive constraints than the
basis_gates/coupling map objects. However, for transpile() if a
backendv2 is passed in for backend paired with coupling_map and/or
basis_gates the expected workflow is that the basis_gates and
coupling_map arguments take priority and override the equivalent
attributes from the backend. To facilitate this we need to block pulling
the target from the backend This should only be needed for a short
period of time as when Qiskit/qiskit#9256 is implemented we'll just build a single
target from the arguments as needed.

Fixes Qiskit/qiskit#7741

* Fix _skip_target logic

* Fix InstructionScheduleMap handling with backendv2

* Fix test failure caused by exception being raised later

* Fix indentation error

* Update qiskit/providers/fake_provider/fake_backend.py

Co-authored-by: John Lapeyre <[email protected]>

* Fix standalone dt argument handling

* Remove unused code

* Fix lint

* Remove duplicate import in set_layout.py

A duplicate import slipped through in the most recent rebase.
This commit fixes that oversight and removes the duplicate.

* Update release notes

Co-authored-by: Jake Lishman <[email protected]>

* Adjust logic for _skip_transpile to check if None

* Simplify check cmap code

* Only check backend if it exists

---------

Co-authored-by: John Lapeyre <[email protected]>
Co-authored-by: Jake Lishman <[email protected]>
@mtreinish mtreinish modified the milestones: 1.0.0, 1.1.0 Jan 23, 2024
This was referenced Mar 12, 2024
Closed
Merged
This was referenced Apr 15, 2024
Merged
Merged
@mtreinish mtreinish modified the milestones: 1.1.0, 1.2.0 May 2, 2024
@mtreinish mtreinish modified the milestones: 1.2.0, 2.0.0 Jul 26, 2024
@ElePT ElePT mentioned this issue Jul 30, 2024
Draft
3 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@mtreinish mtreinish

@ElePT ElePT

Labels
mod: transpiler Issues and PRs related to Transpiler priority: medium type: epic A theme of work that contain sub-tasks type: feature request New feature or request
Projects
Transpiler
Status: in progress
Milestone
2.0.0
Development

No branches or pull requests
5 participants
@1ucian0 @kdk @mtreinish @jakelishman @ElePT