Native gate + stdlib dialect group. #502
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
3 tasks
Codecov Report❌ Patch coverage is
📢 Thoughts on this report? Let us know! |
☂️ Python Coverage
Overall Coverage
New Files
Modified Files
|
There was a problem hiding this comment.
Pull Request Overview
Introduces a new native dialect group and associated standard library of gate kernels (single-qubit and broadcast variants) plus a PyQrack backend implementation and tests. Key additions include dialect statements (CZ, R, Rz), interpreter implementations, stdlib wrappers for common gates, and initial tests for GHZ and selected 1-qubit gates.
- Adds native gate dialect (CZ, R, Rz) and kernel decorator.
- Provides stdlib single-qubit and broadcast gate implementations.
- Implements PyQrack native interpreter methods and initial (but partially ineffective) tests.
Reviewed Changes
Copilot reviewed 12 out of 14 changed files in this pull request and generated 8 comments.
Show a summary per file
| File | Description |
|---|---|
| test/native/test_stdlib.py | Adds GHZ and 1-qubit gate tests (assertion logic currently faulty, some expected states incorrect, limited coverage). |
| src/bloqade/pyqrack/native.py | Implements PyQrack interpreter methods for CZ, R, Rz (minor naming/built-in shadowing issue). |
| src/bloqade/pyqrack/init.py | Exports NativeMethods. |
| src/bloqade/native/stdlib/simple.py | Single-qubit gate kernels (duplication with broadcast; minor naming inconsistency; exposes many untested gates). |
| src/bloqade/native/stdlib/broadcast.py | Broadcast (vectorized) versions of gates (duplicated patterns from simple.py). |
| src/bloqade/native/stdlib/init.py | Re-exports stdlib symbols. |
| src/bloqade/native/dialects/gates/stmts.py | Declares CZ, R, Rz statements. |
| src/bloqade/native/dialects/gates/_interface.py | Provides lowering wrappers (CZ parameter order inconsistent with statement and usage). |
| src/bloqade/native/dialects/gates/_dialect.py | Defines dialect object. |
| src/bloqade/native/dialects/gates/init.py | Re-exports dialect constructs. |
| src/bloqade/native/_prelude.py | Defines kernel dialect group and pass pipeline wrapper. |
| src/bloqade/native/init.py | Exports kernel decorator and gates interface. |
Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
david-pl
reviewed
Sep 22, 2025
david-pl
reviewed
Sep 23, 2025
|
@weinbe58 I really like the general structure here. One small thing I'd vote for is that we re-export the stdlib on the top-level of from bloqade import native
from bloqade.squin import qubit # NOTE: move that to top-level bloqade in other PR
@native.kernel
def main():
q = qubit.new(4)
native.x(q[0]) # calls stdlib.simple
native.broadcast.x(q) # calls stdlib.broadcastI'd also do the same in squin, so users can then just write |
david-pl
reviewed
Sep 23, 2025
david-pl
reviewed
Sep 23, 2025
david-pl
approved these changes
Sep 25, 2025
weinbe58
deleted the
phil/493-native-dialect-and-its-stdlib-that-contains-only-our-native-gates
branch
weinbe58
added a commit
that referenced
this pull request
Sep 25, 2025
This PR implements the native dialect group. In this dialect group we compose with `qubit` and `gates` where `gates` only has Neutral atom specific gate operations. Semanticaly these operators permit explicit parallelism at the statement level, e.g. all qubit arguments are `IListType[QubitType, Any]` and all rotation angles are a single value. Along with the dialect groups there is a standard library of some common gates implemented with this dialect group. There are versions of the gate library `bloqade.native.stdlib` has functions with single qubits as arguments and then there is another library `bloqade.native.stdlib.broadcast` where those arguments are broadcasted like the statements. NOTE: there is no analysis methods required here because all statements have no return value.
david-pl
added a commit
that referenced
this pull request
Sep 29, 2025
This implements the clifford dialect under squin, as discussed in #494. Consistent with #502, all statements take a list of qubits as arguments. We don't expose the wrappers, instead, there is a `simple` stdlib and a `broadcast` stdlib, which contains kernel functions that provide the user-facing API. Some open things: * I'm re-exporting the standard library functions on the top-level of squin. This is for usability, so the user can just write e.g. `squin.x(q[0])`, which calls the `stdlib.simple.x` method. Similarly, you can call `squin.broadcast.x(q)`. The only downside of this, IMO, is that things might be a bit confusing when we backport this. But we can also remove the export and add it back later when we are doing the breaking refactor. * There are a few statements, which I'm not sure we need, so I skipped them for now. Once we settle on which statements should be added, I'll add the rest as stdlib functions. The statements are: * `Shift` * `U3` * `Rot` - this one is actually part of the native dialect, but I'm not sure we need it in clifford, since I already added `Rx` and `Ry`. @weinbe58 what do you think? **Edit**: The way I export stuff here is somewhat related to QuEraComputing/bloqade#287. As @Roger-luo suggests there, we could also move the stdlib here to `gate`, but that is breaking since e.g. `gate.x` is already taken as of now.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR implements the native dialect group. In this dialect group we compose with
qubitandgateswheregatesonly has Neutral atom specific gate operations. Semanticaly these operators permit explicit parallelism at the statement level, e.g. all qubit arguments areIListType[QubitType, Any]and all rotation angles are a single value.Along with the dialect groups there is a standard library of some common gates implemented with this dialect group. There are versions of the gate library
bloqade.native.stdlibhas functions with single qubits as arguments and then there is another librarybloqade.native.stdlib.broadcastwhere those arguments are broadcasted like the statements.NOTE: there is no analysis methods required here because all statements have no return value.