Adding `matmul` to `CircuitComponent`s #347

SamFerracin · 2024-02-14T16:07:05Z

Context:
Adds matmul to circuit components.
Also changes/adds a bunch of stuff that is needed for it, such as:

Adds equality on wires and circuit components (so that tests become easier)
Changes Wires.__rshift__ into Wires.__matmul__, so that all rshift (matmul) in lab_dev mean the same thing, i.e. that they do (do not) add missing adjoints

Tests:
For the tests, it applies @ to sequences of states and gates, and compares the representation of the resulting component to those that one gets by implementing the same sequence with mrmustard.lab. Small differences found, which can be discussed offline

**Description of the Change:** Using python built-in logger, this PR silences the annoying `ComplexWarning`s coming from `tf.cast`. Additionally, it gets rid of tensorflow's import messages. --------- Co-authored-by: Filippo Miatto <ziofil@users.noreply.github.com>

**Context:** In the latest MrM release, `scipy` was unpinned. Syncing the `develop` branch with this latest release causes pytest errors due to improper type handling on `scipy`'s side. **Description of the Change:** - Syncing `develop` with latest `main` - Fixing scipy error by adding a `dtype` parameter to `math.sqrtm` so that it never returns arrays of an unsupported type (specifically, `complex256`) - Updating dependencies - Modifying the CHANGELOG in preparation for release --------- Co-authored-by: zy <48225584+zeyueN@users.noreply.github.com>

**Context:** The previous solution based on `logging` does not appear to work on every OS

**Context:** ``` from mrmustard.lab import * Vacuum(1) ``` Executing the code above triggers a long list of annoying warnings, displayed between the recap of the state and the phase-space plot. **Description of the Change:** This PR removes the cause of the warnings, without changing the way that the plot looks

**Context:** ``` probs = np.array([1e-6 for _ in range(300)]) results = [math.Categorical(probs, "") for _ in range(100)] assert len(set(results)) > 1 ``` When using the numpy backend, the code above fails: `results` contains 100 identical values. This is because `math.Categorical` uses `np.random.multinomial`, which assumes that the probabilities sum up to `1` (if not, the last probability is increased such that the resulting probabilities sum up to `1`)

…merge-dev

**Context:** Add a new file triples.py in the physics module to include all triples in Bargmann representations of the states and transformations. This PR will set up the basic physics for Bargmann objects. **Description of the Change:** **Benefits:** **Possible Drawbacks:** **Related GitHub Issues:** --------- Co-authored-by: SamFerracin <samuele.ferracin@xanadu.ai> Co-authored-by: SamFerracin <samuele.ferracin@gmail.com>

…merge-dev

…o matmul-cc

**Context:** Bug fix of the triples for attenuator **Description of the Change:** c=1.0 for attenuator **Benefits:** **Possible Drawbacks:** **Related GitHub Issues:**

…matmul-cc

mrmustard/lab_dev/circuit_components.py

tests/test_lab_dev/test_circuit_components.py

codecov · 2024-02-14T19:47:34Z

Codecov Report

All modified and coverable lines are covered by tests ✅

❗ No coverage uploaded for pull request base (prototype-tn@07a114e). Click here to learn what that means.

Additional details and impacted files

@@               Coverage Diff               @@
##             prototype-tn     #347   +/-   ##
===============================================
  Coverage                ?   85.03%           
===============================================
  Files                   ?       74           
  Lines                   ?     5440           
  Branches                ?        0           
===============================================
  Hits                    ?     4626           
  Misses                  ?      814           
  Partials                ?        0

Files	Coverage Δ
mrmustard/lab/abstract/transformation.py	`89.53% <100.00%> (ø)`
mrmustard/lab_dev/__init__.py	`100.00% <100.00%> (ø)`
mrmustard/lab_dev/circuit_components.py	`94.39% <100.00%> (ø)`
mrmustard/lab_dev/simulator.py	`98.03% <100.00%> (ø)`
mrmustard/lab_dev/states.py	`100.00% <100.00%> (ø)`
mrmustard/lab_dev/transformations.py	`98.00% <100.00%> (ø)`
mrmustard/lab_dev/wires.py	`100.00% <100.00%> (ø)`
mrmustard/math/backend_manager.py	`98.11% <100.00%> (ø)`
mrmustard/math/backend_numpy.py	`100.00% <ø> (ø)`
mrmustard/math/backend_tensorflow.py	`100.00% <ø> (ø)`
... and 6 more

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 07a114e...66aca97. Read the comment docs.

sylviemonet

Nice PR!
I've checked the logic in side matmul which works for me!

Just leave an open question to discuss: if we do this matmul in simulator and with Fock in each step, would the idea to have the tensor network doesn't exist anymore?

mrmustard/lab_dev/states.py

mrmustard/lab_dev/transformations.py

mrmustard/physics/triples.py

tests/test_lab_dev/test_circuit_components.py

…epresentation docs to make them consistent (#341) **Context:** Fock and ArrayAnsatz, docs and tests --------- Co-authored-by: Filippo Miatto <miatto@gmail.com> Co-authored-by: Filippo Miatto <ziofil@users.noreply.github.com> Co-authored-by: Yuan <16817699+sylviemonet@users.noreply.github.com> Co-authored-by: Sebastián Duque Mesa <675763+sduquemesa@users.noreply.github.com> Co-authored-by: zy <48225584+zeyueN@users.noreply.github.com>

SamFerracin · 2024-02-15T21:34:10Z

Nice PR! I've checked the logic in side matmul which works for me!

Just leave an open question to discuss: if we do this matmul in simulator and with Fock in each step, would the idea to have the tensor network doesn't exist anymore?

@sylviemonet the tensor network idea still exists, but it is orthogonal to the matmul/rshift.
When we specify a path for a circuit, the setter will take care to check that the path is valid for the given circuit based on the connectivity of the components. This way, the check is done only once and outside of the simulation (remember that I may want to simulate the same circuit 1000 times)

sylviemonet

It looks good to me now!

mrmustard/lab_dev/__init__.py

**Context:** There are three bugs fixed in this PR. 1. the projection onto the Vacuum state will use the generaldyne where the projection function is not correct. For example, `from mrmustard.lab import *;Vacuum(3) << Vacuum(3)` The projection in the example above is not 1.0, which should be 1.0. This has been fixed by changing the functions inside generaldyne. 3. add more text to explain about complex_gaussian_integral and other function in bargmann.py. 4. fix the problems that the state will be reassigned during the running of the circuit. For example, `from mrmustard.lab import *; vac0 = Vacuum(1); d1 = Dgate(0.1, 0.1); vac0 >> d1; vac0.bargmann()` the state `vac0` state will be modified after going through the first Dgate, which is not the behaviour we want. **Description of the Change:** **Benefits:** **Possible Drawbacks:** **Related GitHub Issues:** --------- Co-authored-by: Filippo Miatto <ziofil@users.noreply.github.com> Co-authored-by: SamFerracin <samuele.ferracin@xanadu.ai>

…tests (#348) **Context:** Because of the tests were `np.allclose(b, 0)`, we haven't tested the length of the b vector just the values. Hence we have mistakes in the length of b Vector of rotation gate, thermal state and attenuator/amplifier channels. In this PR, we fix the b vectors of them and fix the tests with respect to the shape of the vector it should be. We also fix the type hint for the functions. **Description of the Change:** **Benefits:** **Possible Drawbacks:** **Related GitHub Issues:** --------- Co-authored-by: SamFerracin <samuele.ferracin@gmail.com>

…matmul2-cc

mrmustard/lab_dev/wires.py

mrmustard/lab_dev/circuit_components.py

…o matmul-cc

ziofil

LGTM!

SamFerracin and others added 21 commits February 1, 2024 14:27

Removing TF warnings on import with os instead of logging (#339)

429b16c

**Context:** The previous solution based on `logging` does not appear to work on every OS

Wires final (merge second) (#330)

312d7e2

Merge branch 'develop' of https://github.com/XanaduAI/MrMustard into …

87ee2dc

…merge-dev

fixed

3e7a91a

empty

bf86dfd

wires >> to @

6e97ca4

eq and neq for wires

c8dfa6a

Merge branch 'develop' of https://github.com/XanaduAI/MrMustard into …

f6292cf

…merge-dev

Merge branch 'merge-dev' of https://github.com/XanaduAI/MrMustard int…

edb9675

…o matmul-cc

fixed matmul

0050424

added tests

b69efc7

conflics

6f3492a

oops

790f90c

fix the c of the attenuator triples needs to be 1 (#346)

2dac56c

**Context:** Bug fix of the triples for attenuator **Description of the Change:** c=1.0 for attenuator **Benefits:** **Possible Drawbacks:** **Related GitHub Issues:**

clean up

aceffca

Merge branch 'develop' of https://github.com/XanaduAI/MrMustard into …

d4071ce

…matmul-cc

SamFerracin added the no changelog Pull request does not require a CHANGELOG entry label Feb 14, 2024

SamFerracin requested review from ziofil and sylviemonet February 14, 2024 16:07

SamFerracin commented Feb 14, 2024

View reviewed changes

mrmustard/lab_dev/circuit_components.py Show resolved Hide resolved

black

4c81fac

SamFerracin commented Feb 14, 2024

View reviewed changes

tests/test_lab_dev/test_circuit_components.py Show resolved Hide resolved

SamFerracin added 2 commits February 14, 2024 14:25

fixed math.gather

10d2390

fixed tests and docs of math.gather

4f112dd

error fixed elsewhere

b6c8ac4

sylviemonet reviewed Feb 15, 2024

View reviewed changes

mrmustard/lab_dev/states.py Show resolved Hide resolved

mrmustard/lab_dev/transformations.py Show resolved Hide resolved

mrmustard/physics/triples.py Outdated Show resolved Hide resolved

sylviemonet reviewed Feb 15, 2024

View reviewed changes

tests/test_lab_dev/test_circuit_components.py Show resolved Hide resolved

add global phase tests for two dgates

650ec58

sylviemonet self-requested a review February 20, 2024 19:40

sylviemonet approved these changes Feb 20, 2024

View reviewed changes

ziofil reviewed Feb 21, 2024

View reviewed changes

mrmustard/lab_dev/__init__.py Show resolved Hide resolved

sylviemonet and others added 3 commits February 21, 2024 12:17

Merge branch 'develop' of https://github.com/XanaduAI/MrMustard into …

d7ae0be

…matmul2-cc

ziofil reviewed Feb 21, 2024

View reviewed changes

mrmustard/lab_dev/wires.py Outdated Show resolved Hide resolved

ziofil reviewed Feb 21, 2024

View reviewed changes

mrmustard/lab_dev/circuit_components.py Show resolved Hide resolved

SamFerracin added 2 commits February 21, 2024 16:35

CR

851fa77

Merge branch 'matmul-cc' of https://github.com/XanaduAI/MrMustard int…

99b34b3

…o matmul-cc

ziofil approved these changes Feb 21, 2024

View reviewed changes

SamFerracin added 2 commits February 21, 2024 17:04

Merge branch 'matmul2-cc' into matmul-cc

e9571b5

black

66aca97

SamFerracin merged commit 4d53347 into prototype-tn Feb 21, 2024
7 checks passed

SamFerracin deleted the matmul-cc branch February 21, 2024 22:19

SamFerracin restored the matmul-cc branch February 27, 2024 17:03

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding `matmul` to `CircuitComponent`s #347

Adding `matmul` to `CircuitComponent`s #347

SamFerracin commented Feb 14, 2024

codecov bot commented Feb 14, 2024 •

edited

Loading

sylviemonet left a comment •

edited

Loading

SamFerracin commented Feb 15, 2024 •

edited

Loading

sylviemonet left a comment

ziofil left a comment

Adding __matmul__ to CircuitComponents #347

Adding __matmul__ to CircuitComponents #347

Conversation

SamFerracin commented Feb 14, 2024

codecov bot commented Feb 14, 2024 • edited Loading

Codecov Report

sylviemonet left a comment • edited Loading

Choose a reason for hiding this comment

SamFerracin commented Feb 15, 2024 • edited Loading

sylviemonet left a comment

Choose a reason for hiding this comment

ziofil left a comment

Choose a reason for hiding this comment

Adding `matmul` to `CircuitComponent`s #347

Adding `matmul` to `CircuitComponent`s #347

codecov bot commented Feb 14, 2024 •

edited

Loading

sylviemonet left a comment •

edited

Loading

SamFerracin commented Feb 15, 2024 •

edited

Loading