feat: Drop linear bits, improve pytket encoding/decoding #420

aborgna-q · 2024-06-21T13:49:21Z

Removes the ad-hoc LINEAR_BIT used for decoding pytket circuits, and uses non-linear BOOL_Ts instead.
This now lets us encode guppy circuits with measurements;

module = GuppyModule("test")
module.load(quantum)

@guppy(module)
def my_func(q0: qubit, q1: qubit) -> tuple[bool,]:
    q0 = phased_x(q0, py(math.pi / 2), py(-math.pi / 2))
    q0 = rz(q0, py(math.pi))
    q1 = phased_x(q1, py(math.pi / 2), py(-math.pi / 2))
    q1 = rz(q1, py(math.pi))
    q0, q1 = zz_max(q0, q1)
    _ = measure(q0)
    return (measure(q1),)

circ = guppy_to_circuit(my_func)
print(to_hugr_mermaid(circ))

tk1 = circ.to_tket1()
render_circuit_jupyter(tk1)

circ2 = Tk2Circuit(tk1)
print(to_hugr_mermaid(circ2))

Mermaid diagram (rooted on the DataflowBlock):

graph LR
    subgraph 0 ["(0) Module"]
        direction LR
        subgraph 7 ["(7) FuncDefn"]
            direction LR
            3["(3) Input"]
            3--"0:0<br>qubit"-->8
            3--"1:1<br>qubit"-->8
            6["(6) Output"]
            subgraph 8 ["(8) CFG"]
                direction LR
                subgraph 1 ["(1) DataflowBlock"]
                    direction LR
                    4["(4) Input"]
                    4--"0:0<br>qubit"-->13
                    4--"1:0<br>qubit"-->21
                    5["(5) Output"]
                    9["(9) const:custom:f64(1.5707963267948966)"]
                    9--"0:0<br>float64"-->10
                    10["(10) LoadConstant"]
                    10--"0:1<br>float64"-->13
                    11["(11) const:custom:f64(-1.5707963267948966)"]
                    11--"0:0<br>float64"-->12
                    12["(12) LoadConstant"]
                    12--"0:2<br>float64"-->13
                    13["(13) quantum.tket2.PhasedX"]
                    13--"0:0<br>qubit"-->16
                    14["(14) const:custom:f64(3.141592653589793)"]
                    14--"0:0<br>float64"-->15
                    15["(15) LoadConstant"]
                    15--"0:1<br>float64"-->16
                    16["(16) quantum.tket2.RzF64"]
                    16--"0:0<br>qubit"-->25
                    17["(17) const:custom:f64(1.5707963267948966)"]
                    17--"0:0<br>float64"-->18
                    18["(18) LoadConstant"]
                    18--"0:1<br>float64"-->21
                    19["(19) const:custom:f64(-1.5707963267948966)"]
                    19--"0:0<br>float64"-->20
                    20["(20) LoadConstant"]
                    20--"0:2<br>float64"-->21
                    21["(21) quantum.tket2.PhasedX"]
                    21--"0:0<br>qubit"-->24
                    22["(22) const:custom:f64(3.141592653589793)"]
                    22--"0:0<br>float64"-->23
                    23["(23) LoadConstant"]
                    23--"0:1<br>float64"-->24
                    24["(24) quantum.tket2.RzF64"]
                    24--"0:1<br>qubit"-->25
                    25["(25) quantum.tket2.ZZMax"]
                    25--"0:0<br>qubit"-->26
                    25--"1:1<br>qubit"-->26
                    26["(26) MakeTuple"]
                    26--"0:0<br>[qubit, qubit]"-->27
                    27["(27) UnpackTuple"]
                    27--"0:0<br>qubit"-->28
                    27--"1:0<br>qubit"-->30
                    28["(28) quantum.tket2.Measure"]
                    28--"0:0<br>qubit"-->29
                    29["(29) quantum.tket2.QFree"]
                    30["(30) quantum.tket2.Measure"]
                    30--"0:0<br>qubit"-->31
                    30--"1:0<br>[]+[]"-->32
                    31["(31) quantum.tket2.QFree"]
                    32["(32) MakeTuple"]
                    32--"0:0<br>[[]+[]]"-->33
                    33["(33) UnpackTuple"]
                    33--"0:1<br>[]+[]"-->5
                    34["(34) Tag"]
                    34--"0:0<br>[]"-->5
                end
                1-."0:0".->2
                2["(2) ExitBlock"]
            end
            8--"0:0<br>[]+[]"-->6
        end
    end

tket1 circuit:

Re-extracted circuit:

graph LR
    subgraph 0 ["(0) FuncDefn"]
        direction LR
        1["(1) Input"]
        1--"0:0<br>qubit"-->7
        1--"1:0<br>qubit"-->12
        2["(2) Output"]
        3["(3) const:custom:f64(1.5707963267948966)"]
        3--"0:0<br>float64"-->4
        4["(4) LoadConstant"]
        4--"0:1<br>float64"-->7
        5["(5) const:custom:f64(-1.5707963267948966)"]
        5--"0:0<br>float64"-->6
        6["(6) LoadConstant"]
        6--"0:2<br>float64"-->7
        7["(7) quantum.tket2.PhasedX"]
        7--"0:0<br>qubit"-->15
        8["(8) const:custom:f64(1.5707963267948966)"]
        8--"0:0<br>float64"-->9
        9["(9) LoadConstant"]
        9--"0:1<br>float64"-->12
        10["(10) const:custom:f64(-1.5707963267948966)"]
        10--"0:0<br>float64"-->11
        11["(11) LoadConstant"]
        11--"0:2<br>float64"-->12
        12["(12) quantum.tket2.PhasedX"]
        12--"0:0<br>qubit"-->18
        13["(13) const:custom:f64(3.141592653589793)"]
        13--"0:0<br>float64"-->14
        14["(14) LoadConstant"]
        14--"0:1<br>float64"-->15
        15["(15) quantum.tket2.RzF64"]
        15--"0:0<br>qubit"-->19
        16["(16) const:custom:f64(3.141592653589793)"]
        16--"0:0<br>float64"-->17
        17["(17) LoadConstant"]
        17--"0:1<br>float64"-->18
        18["(18) quantum.tket2.RzF64"]
        18--"0:1<br>qubit"-->19
        19["(19) quantum.tket2.ZZMax"]
        19--"0:0<br>qubit"-->21
        19--"1:0<br>qubit"-->20
        20["(20) quantum.tket2.Measure"]
        20--"0:1<br>qubit"-->2
        20--"1:2<br>[]+[]"-->2
        21["(21) quantum.tket2.Measure"]
        21--"0:0<br>qubit"-->2
        21--"1:3<br>[]+[]"-->2
    end

This required multiple improvements to the encoder/decoder logic, including

Tk1Op::Native operations (backed by a Tk2Op) can now have different number of input/output qubit/bits.
Added support for tket2 circuits with different input and output signatures.
Added support for QAlloc/QFree operations (generated by guppy) by adding extra input/outputs to the circuit.
Added support for pytket's implicit permutations, and recalculates the value when encoding a tket2 circuit.
Preserve the opgroup value from decoded pytket operations.
Improved error reporting.

Closes #379

codecov · 2024-06-21T13:54:15Z

Codecov Report

Attention: Patch coverage is 83.56589% with 106 lines in your changes missing coverage. Please review.

Project coverage is 81.26%. Comparing base (5499817) to head (4649e17).

Files	Patch %	Lines
tket2/src/serialize/pytket/encoder.rs	82.86%	47 Missing and 14 partials ⚠️
tket2/src/serialize/pytket/decoder.rs	79.47%	24 Missing and 7 partials ⚠️
tket2/src/serialize/pytket/op/native.rs	88.63%	2 Missing and 3 partials ⚠️
tket2/src/serialize/pytket/tests.rs	88.23%	4 Missing ⚠️
tket2/src/serialize/pytket.rs	72.72%	0 Missing and 3 partials ⚠️
tket2/src/serialize/pytket/op.rs	94.87%	0 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #420      +/-   ##
==========================================
- Coverage   81.41%   81.26%   -0.16%     
==========================================
  Files          59       59              
  Lines        5423     5838     +415     
  Branches     4938     5349     +411     
==========================================
+ Hits         4415     4744     +329     
- Misses        790      854      +64     
- Partials      218      240      +22

Flag	Coverage Δ
python	`97.54% <100.00%> (+0.02%)`	⬆️
rust	`79.77% <83.41%> (-0.06%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

aborgna-q · 2024-06-21T14:19:28Z

The reduced coverage is mostly related to error paths and functionality that's still unused (e.g. input bools in operations; there's no classically controlled ops in tket2 yet so we cannot test that).

ss2165

much more flexible than i anticipated

main concerns are with use of default "q" and "c" registers

ss2165 · 2024-06-24T10:18:26Z

tket2/src/serialize/pytket/tests.rs

+        |p: &Vec<circuit_json::Permutation>| -> HashMap<circuit_json::Register, circuit_json::Register> {p.iter().map(|p| (p.0.clone(), p.1.clone())).collect()};
+    let _permutation_a = get_permutation(&a.implicit_permutation);
+    let _permutation_b = get_permutation(&b.implicit_permutation);
+


is there supposed to be an assert here that the permutations are equal?

We cannot unsure that anymore, as circuits with different permutations may still be equal.

I removed this bit and added a simple validation instead.

tket2/src/serialize/pytket/decoder.rs

ss2165 · 2024-06-24T10:42:37Z

tket2/src/serialize/pytket/decoder.rs

+            .map(|p| (p.1.clone(), p.0.clone()))
+            .collect();
+        for qubit in &serialcirc.qubits {
+            output_qubits.push(output_to_input.get(qubit).unwrap_or(qubit).clone());


i find this confusing because you're pushing the result of an "output->input" mapping to "output_qubits", I don't think its wrong but can you add some clarifying comments or renames?

I added better comments, with an example.

We are storing the input name for each output register, in the order that outputs appear in a circuit.

ss2165 · 2024-06-24T10:45:53Z

tket2/src/serialize/pytket/decoder.rs

+            let wire = self.register_wires.remove(&register).unwrap();
+            outputs.push(wire);
+        }
+        for wire in self.register_wires.into_values() {


how does this case come about? I note it is also not covered by test

It shouldn't, since ordered_wires always matches the qubit_registers keys.
I replaced the loop with an assert.

ss2165 · 2024-06-24T10:47:49Z

tket2/src/serialize/pytket/decoder.rs

+        }
+
+        // Assign the new output wires to some register, if needed.
+        for (register, wire) in output_registers.into_iter().zip_eq(wires) {


so this reuses existing registers rather than creating a scratch/temp one?

We're just following the registers indicated by the serial circuit, we don't need temp registers here since they all have been declared in SerialCircuit::{qu,}bit already.

tket2/src/serialize/pytket/encoder.rs

ss2165 · 2024-06-24T12:12:00Z

tket2/src/serialize/pytket/encoder.rs

+
+    /// Add a new register unit for a bit wire.
+    pub fn add_qubit_register(&mut self, unit_id: usize) -> &RegisterUnit {
+        let reg = RegisterUnit("q".to_string(), vec![self.qubit_to_reg.len() as i64]);


if "q" register isn't already being used this could suddenly create a very large "q" register with mostly unused qubits right? Not too much of a problem for tket but if then lowered to something that creates whole registers would be annoying. Maybe worth maintaining a "next free qubit"?

ss2165 · 2024-06-24T12:14:06Z

tket2/src/serialize/pytket/encoder.rs

+            }
+        }
+
+        // TODO: Look at the circuit outputs to determine the final permutation.


now or later?

Added a comment. It needs some extra plumbing, so I'll add it as an issue.

ss2165 · 2024-06-24T12:16:28Z

tket2/src/serialize/pytket/encoder.rs

+            None => RegisterUnit("c".to_string(), vec![self.bit_to_reg.len() as i64]),
+        };


not covered, also might be a problem for targets that enforce bit register max sizes for the problems mentioned in the qubit version

ss2165 · 2024-06-24T12:17:33Z

tket2/src/serialize/pytket/encoder.rs

+            .zip(circuit_output_order)
+            .map(|(out, circ_out)| (RegisterHash::from(out), circ_out))
+            .collect();
+        // The final permutation is the composition of these two mappings.


the comments are very helpful

Co-authored-by: Seyon Sivarajah <[email protected]>

ss2165

down to just two questions!

ss2165 · 2024-06-25T10:51:12Z

tket2/src/serialize/pytket/encoder.rs

+        let mut last_unit: Option<u16> = None;
+        for reg in existing {
+            if reg.0 != register {
+                continue;


suggests there should be some tests with multiple register names?

ss2165 · 2024-06-25T10:52:00Z

tket2/src/serialize/pytket/encoder.rs

+            if reg.0 != register {
+                continue;
+            }
+            last_unit = Some(reg.1[0] as u16);


does this assume the existing are always returned in order (i.e. q[0] is checked before q[1])?

ss2165

## 🤖 New release * `tket2`: 0.1.0-alpha.2 -> 0.1.0 * `tket2-hseries`: 0.1.0 <details><summary>Changelog</summary> ## `tket2` <blockquote> ## [0.1.0](tket2-v0.1.0-alpha.2...tket2-v0.1.0) - 2024-08-01 ### Bug Fixes - Single source of truth for circuit names, and better circuit errors ([#390](#390)) - Support non-DFG circuits ([#391](#391)) - Portmatching not matching const edges ([#444](#444)) - Pattern matcher discriminating on opaqueOp description ([#441](#441)) - `extract_dfg` inserting the output node with an invalid child order ([#442](#442)) - Recompile ecc sets after [#441](#441) ([#484](#484)) ### Documentation - Update tket2-py readme ([#431](#431)) - Better error reporting in portmatching ([#437](#437)) - Improved multi-threading docs for Badger ([#495](#495)) ### New Features - `Circuit::operations` ([#395](#395)) - tuple unpack rewrite ([#406](#406)) - guppy → pytket conversion ([#407](#407)) - Drop linear bits, improve pytket encoding/decoding ([#420](#420)) - *(py)* Allow using `Tk2Op`s in the builder ([#436](#436)) - Initial support for `TailLoop` as circuit parent ([#417](#417)) - Support tuple unpacking with multiple unpacks ([#470](#470)) - Partial tuple unpack ([#475](#475)) - [**breaking**] Compress binary ECCs using zlib ([#498](#498)) - Add timeout options and stats to Badger ([#496](#496)) - Expose advanced Badger timeout options to tket2-py ([#506](#506)) ### Refactor - [**breaking**] Simplify tket1 conversion errors ([#408](#408)) - Cleanup tket1 serialized op structures ([#419](#419)) ### Testing - Add coverage for Badger split circuit multi-threading ([#505](#505)) </blockquote> ## `tket2-hseries` <blockquote> ## [0.1.0](https://github.com/CQCL/tket2/releases/tag/tket2-hseries-v0.1.0) - 2024-08-01 ### New Features - [**breaking**] init tket2-hseries ([#368](#368)) - *(tket2-hseries)* Add `tket2.futures` Hugr extension ([#471](#471)) - Add lazify-measure pass ([#482](#482)) - add results extensions ([#494](#494)) - *(tket2-hseries)* [**breaking**] Add `HSeriesPass` ([#487](#487)) </blockquote> </details> --- This PR was generated with [release-plz](https://github.com/MarcoIeni/release-plz/). --------- Co-authored-by: Douglas Wilson <[email protected]>

## 🤖 New release * `tket2`: 0.1.0 -> 0.1.1 * `tket2-hseries`: 0.1.0 -> 0.1.1 <details><summary>Changelog</summary> ## `tket2` <blockquote> ## [0.1.0](tket2-v0.1.0-alpha.2...tket2-v0.1.0) - 2024-08-01 ### Bug Fixes - Single source of truth for circuit names, and better circuit errors ([#390](#390)) - Support non-DFG circuits ([#391](#391)) - Portmatching not matching const edges ([#444](#444)) - Pattern matcher discriminating on opaqueOp description ([#441](#441)) - `extract_dfg` inserting the output node with an invalid child order ([#442](#442)) - Recompile ecc sets after [#441](#441) ([#484](#484)) ### Documentation - Update tket2-py readme ([#431](#431)) - Better error reporting in portmatching ([#437](#437)) - Improved multi-threading docs for Badger ([#495](#495)) ### New Features - `Circuit::operations` ([#395](#395)) - tuple unpack rewrite ([#406](#406)) - guppy → pytket conversion ([#407](#407)) - Drop linear bits, improve pytket encoding/decoding ([#420](#420)) - *(py)* Allow using `Tk2Op`s in the builder ([#436](#436)) - Initial support for `TailLoop` as circuit parent ([#417](#417)) - Support tuple unpacking with multiple unpacks ([#470](#470)) - Partial tuple unpack ([#475](#475)) - [**breaking**] Compress binary ECCs using zlib ([#498](#498)) - Add timeout options and stats to Badger ([#496](#496)) - Expose advanced Badger timeout options to tket2-py ([#506](#506)) ### Refactor - [**breaking**] Simplify tket1 conversion errors ([#408](#408)) - Cleanup tket1 serialized op structures ([#419](#419)) ### Testing - Add coverage for Badger split circuit multi-threading ([#505](#505)) </blockquote> ## `tket2-hseries` <blockquote> ## [0.1.1](tket2-hseries-v0.1.0...tket2-hseries-v0.1.1) - 2024-08-15 ### New Features - *(tket2-hseries)* make result operation internals public ([#542](#542)) </blockquote> </details> --- This PR was generated with [release-plz](https://github.com/MarcoIeni/release-plz/). --------- Co-authored-by: Seyon Sivarajah <[email protected]>

aborgna-q added 8 commits June 20, 2024 14:05

feat: Remove linear bit type

76b0715

refactor: Cleanup the Tk1Encoder

77fbe52

feat: Handle bits wires in the encoder

ba4ddbc

Use unused input bits as outputs

f45f73e

feat: Encode/decode pytket's opgroup

768133d

feat: Handle bools in the decoder

b2c557e

feat: Implicit permutation support, reroute renamed outputs,

f2a3bdb

Enable guppy test with measurements

fa2d30d

Fix import in rust 1.75

6bd4932

aborgna-q marked this pull request as ready for review June 21, 2024 14:19

aborgna-q requested a review from ss2165 June 21, 2024 14:19

aborgna-q mentioned this pull request Jun 21, 2024

Publish alpha version of tket2-py #387

Closed

aborgna-q changed the title ~~feat: Support non-linear bits, improve pytket encoding/decoding~~ feat: Drop linear bits, improve pytket encoding/decoding Jun 21, 2024

ss2165 reviewed Jun 24, 2024

View reviewed changes

aborgna-q and others added 8 commits June 24, 2024 14:14

Apply suggestions from code review

5b59833

Co-authored-by: Seyon Sivarajah <[email protected]>

docs

7d8f142

Test properties for generated serial circuits

df05590

Clarify docs.

02bc3e2

Drop unreachable code.

39c1e19

Better "next fresh qubit/bit" generators

55e6467

Better handling of QAlloc, and cleanup some error code

f099656

Add test with ancillae

9e31fdf

aborgna-q requested a review from a team as a code owner June 25, 2024 09:57

aborgna-q requested a review from ss2165 June 25, 2024 09:57

Merge branch 'main' into ab/bits-to-pytket

917fb65

ss2165 reviewed Jun 25, 2024

View reviewed changes

aborgna-q added 2 commits June 25, 2024 12:20

Fix last register search

f5e4590

Add tests with different/unordered registers

4649e17

aborgna-q requested a review from ss2165 June 25, 2024 11:29

ss2165 approved these changes Jun 25, 2024

View reviewed changes

aborgna-q added this pull request to the merge queue Jun 25, 2024

Merged via the queue into main with commit a6e9e13 Jun 25, 2024
16 checks passed

aborgna-q deleted the ab/bits-to-pytket branch June 25, 2024 11:52

hugrbot mentioned this pull request Jun 25, 2024

chore: release #409

Closed

aborgna-q mentioned this pull request Jun 25, 2024

Compute pytket qubit register reordering during encoding #429

Open

hugrbot mentioned this pull request Aug 1, 2024

chore: release #519

Merged

hugrbot mentioned this pull request Aug 5, 2024

chore: release #521

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: Drop linear bits, improve pytket encoding/decoding #420

feat: Drop linear bits, improve pytket encoding/decoding #420

aborgna-q commented Jun 21, 2024 •

edited

Loading

codecov bot commented Jun 21, 2024 •

edited

Loading

aborgna-q commented Jun 21, 2024

ss2165 left a comment •

edited

Loading

ss2165 Jun 24, 2024

aborgna-q Jun 24, 2024

ss2165 Jun 24, 2024

aborgna-q Jun 24, 2024

ss2165 Jun 24, 2024

aborgna-q Jun 24, 2024

ss2165 Jun 24, 2024

aborgna-q Jun 24, 2024

ss2165 Jun 24, 2024

ss2165 Jun 24, 2024

aborgna-q Jun 24, 2024

ss2165 Jun 24, 2024

ss2165 Jun 24, 2024

ss2165 left a comment

ss2165 Jun 25, 2024

ss2165 Jun 25, 2024

aborgna-q Jun 25, 2024

ss2165 left a comment

		None => RegisterUnit("c".to_string(), vec![self.bit_to_reg.len() as i64]),
		};

feat: Drop linear bits, improve pytket encoding/decoding #420

feat: Drop linear bits, improve pytket encoding/decoding #420

Conversation

aborgna-q commented Jun 21, 2024 • edited Loading

codecov bot commented Jun 21, 2024 • edited Loading

Codecov Report

aborgna-q commented Jun 21, 2024

ss2165 left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ss2165 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ss2165 left a comment

Choose a reason for hiding this comment

aborgna-q commented Jun 21, 2024 •

edited

Loading

codecov bot commented Jun 21, 2024 •

edited

Loading

ss2165 left a comment •

edited

Loading