Hi all,
Tomorrow Roman Wixinger will present his semester project “Uncomputation
and Entanglement in High-level Quantum Programming Languages”, conducted
under the supervision of Prof. Christian Mendl at TU München. See below for
the abstract. We start as usual at 2pm on zoom:
https://ethz.zoom.us/j/362994444.
Best,
Joe
Quantum programming languages traditionally focus on the hardware level
and are therefore not really good at representing the intentions of the
programmer. Explicit formulation of uncomputation, which is essential
for the safe and efficient use of the qubits, makes the code
unnecessarily complex. In recent work, Vechev et al. (2020) introduced
Silq, a high-level language that allows for safe, automatic
uncomputation just using its type system. This feature makes the code
significantly shorter and more intuitive. The type system can also
ensure that any program that compiles is physical. In this project, we
compared Silq’s solution of handling uncomputation with other approaches
and give an overview of the features of quantum languages. We have also
tried to understand whether a qubit can be safely discarded by directly
looking at the entanglement.
Hi all,
Apologies for the late email. This afternoon Diane Saint Aubin will tell us
about her semester project, entitled "Lower and upper bounds on quantum key
distribution protocols". See below for the abstract. We'll start as usual
at 2pm in/at https://ethz.zoom.us/j/362994444.
Best,
Joe
Abstract:
The 3-state quantum key distribution protocol has two entanglement-based
descriptions, raising the question of which one is best used to compute the
key rate. We show that the descriptions are equivalent, and that
symmetrising the state used in the protocol can give a lower bound on the
secret key rate. In addition, we provide an upper bound for the BB84
protocol using the intrinsic information which just outperforms a recent
bound by Xing Wang. For the device independent protocol based on the CHSH
game, thresholds on the key rate are found depending on the observed
parameters. We find an upper bound for protocols using one way-classical
communication, which is much tighter than the previous two-way classical
communication bound for low values of the Bell violation.
Hi all,
Tomorrow our new postdoc Christopher Chubb will tell us about his research,
in particular his latest paper "General tensor network decoding of 2D Pauli
codes". See below for the abstract, or the full paper at
https://arxiv.org/abs/2101.04125. We start at 2pm in zoom:
https://ethz.zoom.us/j/362994444.
Best,
Joe
Abstract:
In this work we develop a general tensor network decoder for 2D codes.
Specifically, we propose a decoder which approximates maximally likelihood
decoding for 2D stabiliser and subsystem codes subject to Pauli noise. For
a code consisting of $n$ qubits our decoder has a runtime of $O(n\log
n+n\chi^3)$, where $\chi$ is an approximation parameter. We numerically
demonstrate the power of this decoder by studying four classes of codes
under three noise models, namely regular surface codes, irregular surface
codes, subsystem surface codes and colour codes, under bit-flip, phase-flip
and depolarising noise. We show that the thresholds yielded by our decoder
are state-of-the-art, and numerically consistent with optimal thresholds
where available, suggesting that the tensor network decoder well
approximates optimal decoding in all these cases. Novel to our decoder is
an efficient and effective approximate contraction scheme for arbitrary 2D
tensor networks, which may be of independent interest.
Hi all,
Tomorrow David Ittah will tell us about his master thesis on "Multi-level
IR for Quantum Program Optimization", which was supervised by Torsten
Hoefler from the CS department and Thomas Häner from Microsoft (formerly
Troyer group). See below for the abstract. The zoom link is
https://ethz.zoom.us/j/362994444.
Best,
Joe
Abstract:
Intermediate representations (IR) have traditionally provided numerous
benefits to compilation systems, in particular in the domain of static
program analysis and optimization. However, the quantum programming
landscape has yet to produce similarly powerful IRs, instead focusing on
the development of embedded domain-specific languages (eDSL). As these only
feature rudimentary IRs in the form of data structures for quantum
circuits, or simple quantum assembly (QASM) languages, they lack meaningful
integration with their classical host compilation infrastructure. As a
remedy, we propose a novel quantum IR design based on exposing SSA-like
quantum dataflow in the IR alongside classical dataflow. Our language- and
hardware-agnostic IR is designed to enable classical-quantum
co-optimization, and with the intent to maximize the reuse of existing
compilation infrastructure. An implementation in the MLIR compiler
framework demonstrates that ∼99.8% of savings identified by ProjectQ’s
run-time optimizations on Shor’s algorithm can be exploited by static
optimizations in our IR. A resource estimation routine to count the number
of rotation gates in Shor’s algorithm is shown to run 5-6 orders of
magnitudes faster on application-scale input sizes than on existing systems
in ProjectQ and Qiskit.
Hi all,
Tomorrow at 2pm Severin Meng will tell us about his master thesis, entitled
"Thermodynamic Properties of Passive States". See below for the abstract.
The meeting link is https://ethz.zoom.us/j/362994444.
Best,
Joe
Abstract:
A passive state is defined such that no unitary transformation can lower
its average energy. Within the set of passive states one finds the
well-known thermal states, which are described by a quantum analogue of
classical temperature. Interesting quantities like the efficiency, cooling
performance or even possibility of a state transition within settings that
use a thermal state resource are characterised by the state's temperature
and Hamiltonian.
Non-thermal passive states appear naturally, for example at the end of a
thermodynamic process like cooling, and they may thus be used as an initial
state in a different thermodynamic scenario. There is no unique temperature
of a non-thermal passive state that characterises its thermodynamic
performance. We investigate which properties of passive states determine
their thermodynamic behaviour. We approach this task by exploring different
thermodynamic settings that involve passive states. The considered settings
include an autonomous heat engine and refrigerator as well as state
transformations and more abstract work extraction protocols.
We find different parameters for the various scenarios that describe the
thermodynamics of the passive state. The set of relevant parameters
includes the temperatures of the thermal state of the same energy as well
as the thermal state of the same entropy, the asymptotic activation energy,
the free energy and entropy as well as specific combinations of energy gaps
and virtual temperatures within the passive state.
Hi all,
Tomorrow Henrik will kick off the QIT Seminar in the new year. He will tell
us about the connection between entropy and reversible catalysis, from
arXiv:2012.05573 <https://arxiv.org/abs/2012.05573> . The talk starts at
2pm on zoom: https://ethz.zoom.us/j/362994444.
Best,
Joe
Hi all,
Tomorrow at 2pm Jan Li will tell us about his master thesis
on "Device-independent Quantum Key distribution and Nonlocality
Distillation", which he did with Dr. Koon Tong Goh and Prof. Dr. Charles Ci
Wen Lim at CQT and NUS in Singapore. See below for the abstract. The zoom
link is https://ethz.zoom.us/j/362994444.
Best,
Joe
Abstract:
In this project, we have investigated the use of local wirings as described
in Short et al.
[table 1 https://arxiv.org/abs/quant-ph/0508120] to revive zero keyrates of
DIQKD set-ups to positive keyrates. In this project, we first considered
the correlations generated by a two qubit set-up. Due to the high
computational cost of considering all possible wirings, we checked for a
subclass of extremal wirings if there is revival. For this model, we found
that some revival was possible for some wirings and that revival could be
pushed further by wiring multiple times. Next, we moved to a more realistic
model, the all-photonics system as is considered in Tsujimoto et al. [
https://journals.aps.org/pra/abstract/10.1103/PhysRevA.98.063842]. For this
model however, we were not able to find any improvements and further
research is needed.
Hi all,
Tomorrow Amira Abbas from IBM Zürich and the University of Durban will tell
us about "The power of quantum neural networks". The full abstract is
below. Join us at 2pm on zoom, at https://ethz.zoom.us/j/362994444.
Best,
Joe
Abstract:
Fault-tolerant quantum computers offer the promise of dramatically
improving machine learning through speed-ups in computation or improved
model scalability. In the near-term, however, the benefits of quantum
machine learning are not so clear. Understanding expressibility and
trainability of quantum models–-and quantum neural networks in
particular–-requires further investigation. In this work, we use tools from
information geometry to define a notion of expressibility for quantum and
classical models. The effective dimension, which depends on the Fisher
information, is used to prove a novel generalisation bound and establish a
robust measure of expressibility. We show that quantum neural networks are
able to achieve a significantly better effective dimension than comparable
classical neural networks. To then assess the trainability of quantum
models, we connect the Fisher information spectrum to barren plateaus, the
problem of vanishing gradients. Importantly, certain quantum neural
networks can show resilience to this phenomenon and train faster than
classical models due to their favourable optimisation landscapes, captured
by a more evenly spread Fisher information spectrum. Our work is the first
to demonstrate that well-designed quantum neural networks offer an
advantage over classical neural networks through a higher effective
dimension and faster training ability, which we verify on real quantum
hardware.