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.

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

Sequence sampler #345

Merged
merged 48 commits into from
Mar 28, 2022
Merged
Show file tree
Hide file tree
Changes from 40 commits
Commits
Show all changes
48 commits
Select commit Hold shift + click to select a range
e122aea
A first gist of sequence sampler
lvignoli Feb 24, 2022
94fbdb2
Refactor sampler in independent module
lvignoli Mar 14, 2022
f027757
Consistent support for noises, SLM and modulation of Global and Local
lvignoli Mar 14, 2022
5b5f95b
Refactor the noises module
lvignoli Mar 14, 2022
7459af5
Improve docstrings and small refactoring
lvignoli Mar 14, 2022
c0ff5d7
Small typos in comments
lvignoli Mar 15, 2022
79fcd26
Fix the decay of global channels to local ones
lvignoli Mar 16, 2022
d983031
Fix the modulation feature
lvignoli Mar 16, 2022
922ffd3
Refactor
lvignoli Mar 16, 2022
e7f1798
Simplify the flow of samples.samples()
lvignoli Mar 21, 2022
9269cdd
Improve test and get total coverage
lvignoli Mar 21, 2022
1220cbd
Simplify the sampling by removing the Samples dataclass
lvignoli Mar 22, 2022
949cfd0
Fix a docstring
lvignoli Mar 22, 2022
b2f0e3e
Fix docstrings formatting and content
lvignoli Mar 23, 2022
8d09f12
Remove unnecessary nonzero check in amplitude noise
lvignoli Mar 23, 2022
9f35042
Fix the noise seed default value: defaults to None
lvignoli Mar 23, 2022
03e8558
Improve docstrings of helper functions in noise module
lvignoli Mar 23, 2022
19af68d
Apply noises before the SLM masking
lvignoli Mar 23, 2022
14f9c8b
Rename misnamed variables
lvignoli Mar 23, 2022
41d175d
Change the scope of _key_func() and remove the _GroupType class
lvignoli Mar 23, 2022
4c8b160
Fix dictionary construction with defaultdict
lvignoli Mar 23, 2022
1af88d0
Fix misnamed keys in tests
lvignoli Mar 23, 2022
8084b6e
Add a match to pytest.raises
lvignoli Mar 23, 2022
ff38f10
Move the noises to the simulation module
lvignoli Mar 23, 2022
96db04c
Improve the simulation.noises docstring
lvignoli Mar 23, 2022
7129c95
Remove noisy helper functions
lvignoli Mar 23, 2022
578c78d
Write test case for _write_dict exception
lvignoli Mar 23, 2022
f1b2f23
Reorder tests
lvignoli Mar 23, 2022
802e86e
Add a more fundamental test
lvignoli Mar 23, 2022
4b36792
Reword and add docstrings
lvignoli Mar 23, 2022
d3f207b
Fix other misnamed keys in tests
lvignoli Mar 23, 2022
85cbdbf
Fix SLM masking: consider only the seq._mask_times
lvignoli Mar 25, 2022
a88c31d
Change for defaultdict in new_qdict
lvignoli Mar 25, 2022
97126fc
Add a note about performance for sampling of global channels
lvignoli Mar 25, 2022
0d752fd
Simply testing of sequence
lvignoli Mar 25, 2022
77d97e0
Refactor the testing of sequence with a helper function
lvignoli Mar 25, 2022
fe65aa4
Fix the testing of blackman modulation
lvignoli Mar 25, 2022
1f05e03
Move noise-related tests to an independent testing files
lvignoli Mar 25, 2022
2c8c07a
Fix the unneeded # pragma: no cover in _sample_slots
lvignoli Mar 25, 2022
9697a2d
Add a Failing test for the SLM, for discussion
lvignoli Mar 25, 2022
198c094
Move the comment on performance
lvignoli Mar 28, 2022
373e10f
Make the SLM detectino more idiomatic
lvignoli Mar 28, 2022
0720c50
Simplify the assertions for sequence in XY
lvignoli Mar 28, 2022
eeeba30
Remove superfluous print statements
lvignoli Mar 28, 2022
0201e14
Test SLM sampling alongside the check against simulation
lvignoli Mar 28, 2022
aefbc9e
Refactor using a helper for nested dicts
lvignoli Mar 28, 2022
35d4b75
Rearrange the test file
lvignoli Mar 28, 2022
12c9344
Merge branch 'develop' into WIP/sampling
HGSilveri Mar 28, 2022
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
23 changes: 23 additions & 0 deletions pulser/sampler/__init__.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
# Copyright 2020 Pulser Development Team
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""Module sampler enables the sampling of pulser sequences.

Samples of a sequence are needed for plotting and simulation.

Typical usage:

sampler.sample(sequence)
"""
from pulser.sampler.sampler import sample
283 changes: 283 additions & 0 deletions pulser/sampler/sampler.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,283 @@
"""Exposes the sample() functions.

It contains many helpers.

NOTE(perf): it not very efficient to hold the same data for all qubit in global
channels, but i remains manageable for register # with less than 100 qubits.
lvignoli marked this conversation as resolved.
Show resolved Hide resolved
"""
from __future__ import annotations

import itertools
from collections import defaultdict
from typing import Callable, List, Optional, cast

import numpy as np

import pulser.simulation.noises as noises
from pulser.channels import Channel
from pulser.pulse import Pulse
from pulser.sampler.samples import QubitSamples
from pulser.sequence import Sequence, _TimeSlot


def sample(
seq: Sequence,
modulation: bool = False,
common_noises: Optional[list[noises.NoiseModel]] = None,
global_noises: Optional[list[noises.NoiseModel]] = None,
) -> dict:
"""Samples the given Sequence and returns a nested dictionary.

It is intended to be used like the json.dumps() function.

Args:
seq (Sequence): A pulser.Sequence instance.
modulation (bool): Flag to account for the modulation of AOM/EOM
before sampling.
common_noises (Optional[list[LocalNoise]]): A list of the noise sources
for all channels.
global_noises (Optional[list[LocalNoise]]): A list of the noise sources
for global channels.

Returns:
A nested dictionnary of the samples of the amplitude, detuning and
phase at every nanoseconds for all channels.
"""
if common_noises is None:
common_noises = []
if global_noises is None:
global_noises = []

# 1. determine if the global channel decay to a local one
# 2. extract samples
# 3. modulate
# 4. apply noises/SLM
# 5. write samples

samples: dict[str, list[QubitSamples]] = {}
addrs: dict[str, str] = {}

for ch_name, ch in seq.declared_channels.items():
s: list[QubitSamples]

addr = seq.declared_channels[ch_name].addressing

ch_noises = list(common_noises)

slm_on = len(seq._slm_mask_targets) > 0 and len(seq._slm_mask_time) > 0
lvignoli marked this conversation as resolved.
Show resolved Hide resolved

if addr == "Global":
decay = slm_on or len(global_noises) > 0 or len(common_noises) > 0
if decay:
addr = "Decayed"
ch_noises.extend(global_noises)

addrs[ch_name] = addr

strategy = _group_between_retargets if modulation else _regular
s = _sample_channel(seq, ch_name, strategy)
if modulation:
s = _modulate(ch, s)

s = noises.apply(s, ch_noises)

if slm_on: # Update the samples of masked qubits during SLM on times
for i, _ in enumerate(s):
if s[i].qubit in seq._slm_mask_targets:
ti, tf = seq._slm_mask_time[0], seq._slm_mask_time[1]
s[i].amp[ti:tf] = 0.0
# apply only on amp since it's just a shutter

samples[ch_name] = s

# format the samples in the simulation dict form
d = _write_dict(seq, samples, addrs)

return d


def _prepare_dict(seq: Sequence, N: int) -> dict:
"""Constructs empty dict of size N.

Usually N is the duration of seq.
"""

def new_qty_dict() -> dict:
return {
"amp": np.zeros(N),
"det": np.zeros(N),
"phase": np.zeros(N),
}

def new_qdict() -> dict:
return defaultdict(new_qty_dict)

if seq._in_xy:
return {
"Global": {"XY": new_qty_dict()},
"Local": {"XY": new_qdict()},
}
else:
return {
"Global": defaultdict(new_qty_dict),
"Local": defaultdict(new_qdict),
}


def _write_dict(
seq: Sequence,
samples: dict[str, list[QubitSamples]],
addrs: dict[str, str],
) -> dict:
"""Export the given samples to a nested dictionary."""
# Get the duration
if not _same_duration(samples):
raise ValueError("All the samples do not share the same duration.")
N = list(samples.values())[0][0].amp.size

d = _prepare_dict(seq, N)

for ch_name, some_samples in samples.items():
basis = seq.declared_channels[ch_name].basis
addr = addrs[ch_name]
if addr == "Global":
# Take samples on only one qubit and write them
a_qubit = next(iter(seq._qids))
to_write = [x for x in some_samples if x.qubit == a_qubit]
lvignoli marked this conversation as resolved.
Show resolved Hide resolved
for s in to_write:
d["Global"][basis]["amp"] += s.amp
d["Global"][basis]["det"] += s.det
d["Global"][basis]["phase"] += s.phase
else:
for s in some_samples:
d["Local"][basis][s.qubit]["amp"] += s.amp
d["Local"][basis][s.qubit]["det"] += s.det
d["Local"][basis][s.qubit]["phase"] += s.phase
return d


def _same_duration(samples: dict[str, list[QubitSamples]]) -> bool:
lvignoli marked this conversation as resolved.
Show resolved Hide resolved
durations: list[int] = []
flatten_samples: list[QubitSamples] = []
for some_samples in samples.values():
flatten_samples.extend(some_samples)
for s in flatten_samples:
durations.extend((s.amp.size, s.det.size, s.phase.size))
return durations.count(durations[0]) == len(durations)


def _sample_channel(
seq: Sequence, ch_name: str, strategy: TimeSlotExtractionStrategy
) -> list[QubitSamples]:
"""Compute a list of QubitSamples for a channel."""
qs: list[QubitSamples] = []
grouped_slots = strategy(seq._schedule[ch_name])

for group in grouped_slots:
ss = _sample_slots(seq.get_duration(), *group)
qs.extend(ss)
return qs


def _sample_slots(N: int, *slots: _TimeSlot) -> list[QubitSamples]:
"""Gather samples of a list of _TimeSlot in a single Samples instance."""
# Same target in one group, guaranteed by the strategy (this seems
# weird, it's not enforced by the structure,bad design?)
lvignoli marked this conversation as resolved.
Show resolved Hide resolved
qubits = slots[0].targets
amp, det, phase = np.zeros(N), np.zeros(N), np.zeros(N)
pulse_slots = [s for s in slots if isinstance(s.type, Pulse)]
for s in pulse_slots:
pulse = cast(Pulse, s.type)
amp[s.ti : s.tf] += pulse.amplitude.samples
det[s.ti : s.tf] += pulse.detuning.samples
phase[s.ti : s.tf] += pulse.phase
qs = [
QubitSamples(
amp=amp.copy(), det=det.copy(), phase=phase.copy(), qubit=q
)
for q in qubits
]
return qs


TimeSlotExtractionStrategy = Callable[[List[_TimeSlot]], List[List[_TimeSlot]]]
"""Extraction strategy of _TimeSlot's of a Channel.

It's an alias for functions that returns a list of lists of _TimeSlots.
_TimeSlots in the same group MUST share the same targets.

NOTE:
This strategy type is used mostly for the necessity to extract samples
differently when taking into account the modulation of AOM/EOM. Despite
there are only two cases, whether it's necessary to modulate a local
channel or not, this pattern can accomodate for future needs.
"""


def _regular(ts: list[_TimeSlot]) -> list[list[_TimeSlot]]:
"""No grouping performed, return only the pulses."""
return [[x] for x in ts if isinstance(x.type, Pulse)]


def _group_between_retargets(
ts: list[_TimeSlot],
) -> list[list[_TimeSlot]]:
"""Filter and group _TimeSlots together.

Group the input slots by groups of successive Pulses and delays between
two target operations. Consider the following sequence consisting of pulses
A B C D E F, targeting different qubits:

.---A---B------.---C--D--E---.----F--
^ ^ ^
| | |
target q0 target q1 target q0

It will group the pulses' _TimeSlot's in batches (A B), (C D E) and (F),
returning the following list of list of _TimeSlot instances:

[[A, B], [C, D, E], [F]]

Args:
ts (list[_TimeSlot]): A list of TimeSlot from a Sequence schedule.

Returns:
A list of list of _TimeSlot. _TimeSlot instances are successive and
share the same targets. They are of type either Pulse or "delay", all
"target" ones are discarded.
"""
TO_KEEP = "pulses_and_delays"

def key_func(x: _TimeSlot) -> str:
if isinstance(x.type, Pulse) or x.type == "delay":
return TO_KEEP
else:
return "other"

grouped_slots: list[list[_TimeSlot]] = []

for key, group in itertools.groupby(ts, key_func):
g = list(group)
if key != TO_KEEP:
continue
grouped_slots.append(g)

return grouped_slots


def _modulate(ch: Channel, samples: list[QubitSamples]) -> list[QubitSamples]:
"""Modulate local samples according to the hardware specs.

Additional parameters will probably be needed (keep_end, etc).
"""
modulated_samples: list[QubitSamples] = []
for s in samples:
modulated_samples.append(
QubitSamples(
amp=ch.modulate(s.amp),
det=ch.modulate(s.det),
phase=ch.modulate(s.phase),
qubit=s.qubit,
)
)
return modulated_samples
24 changes: 24 additions & 0 deletions pulser/sampler/samples.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,24 @@
"""Defines samples dataclasses."""
from __future__ import annotations

from dataclasses import dataclass

import numpy as np

from pulser.sequence import QubitId


@dataclass
class QubitSamples:
"""Gathers samples concerning a single qubit."""

amp: np.ndarray
det: np.ndarray
phase: np.ndarray
qubit: QubitId

def __post_init__(self) -> None:
if not len(self.amp) == len(self.det) == len(self.phase):
raise ValueError(
"ndarrays amp, det and phase must have the same length."
)
Loading