Professional Documents
Culture Documents
TMQS 2024 - Program
TMQS 2024 - Program
Monday, June 3
2
Tuesday, June 4
• measurements
• quantum circuits
• fundamental algorithms
3
Wednesday, June 5
• project work
• project presentations
4
Thursday, June 6
t.b.a.
5
Friday, June 7
t.b.a.
6
Abstracts (in chronological order)
A BSTRACT :
The development of quantum-computing hardware has currently reached the stage where specific
classes of simulations that are challenging for state-of-the-art classical algorithms can be executed
on quantum computers. Maximally exploiting the power of such quantum hardware – the so-called
utility-scale hardware – requires combining advanced quantum- and classical-computing simulation
methods. Tensor network methods are currently the core of the most efficient strategies for classi-
cal simulations of quantum computers. Within this talk, I will present strategies for leveraging tensor
networks for designing utility-scale quantum simulations. I will first describe how tensor networks can
inspire the design of new hybrid variational quantum algorithms. I will then present strategies to lever-
age tensor networks to quantify the complexity of quantum-computing simulations. For both methods,
I will describe their potential application to natural-science simulations.
A BSTRACT :
Tensor networks are a family of quantum-inspired algorithms that have proven immensely powerful for
simulating quantum systems. At the same time, there are cases where they fall short of providing
satisfying results due to their limitations in representing highly entangled quantum states. Quantum
computers on the other hand should soon be able to prepare a wide range of interesting states that go
beyond the capabilities of tensor networks, but quantum algorithms are held back by problems such
as hardware noise and exponentially concentrating variational loss landscapes. In this talk, we will
discuss the potential synergy between tensor networks on classical hardware and quantum circuits
on quantum hardware. We will first study a range of protocols for translating tensor network quantum
states to their quantum circuit representations, including one protocol that promises to have the best
trade-off between computational cost, accuracy and quantum circuit depth. We will then move to the
quantum computer and explore applications which are unlocked due to previously having done a sig-
nificant amount of work on classical hardware. In particular, we will see that warm-starting variational
optimization of quantum circuits with solutions found by tensor networks can turn previously fruitless
optimizations into ones that reliably converge to high-quality solutions. Hopefully this work inspires
algorithms that embrace the strengths of classical and quantum hardware and that synergize them
accordingly.
7
Exploring optimal tensor network architectures through tensor
network structure search (TN-SS)
A BSTRACT :
Tensor Networks (TN) have emerged as a compelling framework with extensive applications in quan-
tum computing, machine learning, and signal processing, yet selecting the most effective TN model—a
process known as Tensor Network Structure Search (TN-SS)—presents significant computational chal-
lenges. In my upcoming presentation, I will provide an in-depth overview of our novel approach to tack-
ling these challenges, focusing on the formulation of the problem and strategic solutions through a dis-
crete optimization lens. I will particularly highlight three groundbreaking algorithms: TNGA (ICML’20),
TNLS (ICML’22), and TnALE (ICML’23), along with their theoretical contributions to the field. Addition-
ally, time permitting, I will discuss our recent progress in leveraing Large Language Models (LLMs) to
boost the study of TN-SS. We propose a new "prompt pipeline", by which LLMs can explore unknown
and more efficient TN-SS algorithms automatically, going beyond the current limited understanding
from "human" researchers.
t.b.a.
A BSTRACT :
t.b.a.
A BSTRACT :
The potential in the electronic Schrödinger equation depends on the distances between the electrons.
The electronic wave functions therefore possess singularities at the points in R3N where two of the
considered N electrons meet. Orbital-based approaches are not well suited to resolve these singular-
ities. It is therefore an attractive idea to use ansatz functions that explicitly depend on the differences
of the particle positions. The talk will present an approach that allows to treat such ansatz functions in
a tensor framework. The idea is to represent the wave functions as traces of even higher-dimensional
separable functions. The main tool is a fast iterative method for solving Laplace-like equations with
right-hand sides that themselves are traces of such functions. The amazing properties of this iterative
method are based on a measure concentration effect.
8
Hamiltonian truncation tensor networks for quantum field theories
A BSTRACT :
Understanding the equilibrium properties and out of equilibrium dynamics of quantum field theories
are key aspects of fundamental problems in theoretical particle physics and cosmology. However, their
classical simulation is highly challenging. In this work, we introduce a tensor network method for the
classical simulation of continuous quantum field theories that is suitable for the study of low-energy
eigenstates and out-of-equilibrium time evolution. The method is built on Hamiltonian truncation and
tensor network techniques, bridging the gap between two successful approaches. One of the key
developments is the exact construction of matrix product state representations of global projectors,
crucial for the implementation of interacting theories. Despite featuring a relatively high computational
effort, our method dramatically improves predictive precision compared to exact diagonalisation-based
Hamiltonian truncation, allowing the study of so far unexplored parameter regimes and dynamical
effects. We corroborate trust in the accuracy of the method by comparing it with exact theoretical
results for ground state properties of the sine-Gordon model. We then proceed with discussing (1+1)-
dimensional quantum electrodynamics, the massive Schwinger model, for which we accurately locate
its critical point and study the growth and saturation of momentum-space entanglement in sudden
quenches.
A BSTRACT :
Tensor networks have evolved into competitive, state-of-the-art tools for the simulation of low-
dimensional quantum many-body systems. One-dimensional matrix product state (MPS) methods are
well-established, offering a variety of algorithms for ground state searches, thermal state calculations,
and time evolution. In two spatial dimensions, (infinite) projected entangled pair state have proven
particularly important for addressing contemporary challenges in condensed matter physics. However,
their optimization has remained challenging, necessitating the development of novel algorithms tai-
lored to this structure, especially for ground state searches. This talk provides an overview of common
iPEPS methods, highlighting advancements in ground state calculations such as the variational update
technique.
9
Quantum noise characterization with a tensor network quantum
jump method
A BSTRACT :
In this talk, we will discuss a novel approach to characterizing the noise in noisy quantum circuits
through the Tensorized Quantum Jump Method (TJM). The well-known Quantum Jump Method (Monte
Carlo wave function), which is used to approximate Lindbladian dynamics, can be transferred to a
tensor network algorithm via Strang splitting of the Lindbladian and the help of a dynamical low-rank
approximation through the Time-Dependent Variational Principle. Choosing the sparse Pauli strings of
the Sparse Pauli-Lindblad Model (SPLM) as Lindblad operators makes this method a new approach to
characterizing quantum noise in large systems by learning the corresponding noise parameters.
A BSTRACT :
We introduce a class of quantum non-Markovian processes – dubbed process trees – that exhibit
polynomially decaying temporal correlations and memory distributed across time scales. This class
of processes is described by a tensor network with tree-like geometry whose component tensors are
(1) causality-preserving maps (superprocesses) and (2) locality-preserving temporal change of scale
transformations. We show that the long-range correlations in this class of processes tends to originate
almost entirely from memory effects, and can accommodate genuinely quantum power-law correla-
tions in time. Importantly, this class allows efficient computation of multi-time correlation functions. To
showcase the potential utility of this model-agnostic class for numerical simulation of physical models,
we show how it can approximate the strong memory dynamics of the paradigmatic spin-boson model,
in terms of arbitrary multitime features. In contrast to an equivalently costly matrix product operator
representation, the ansatz produces a fiducial characterization of the relevant physics. Our work lays
the foundation for the development of more efficient numerical techniques in the field of strongly inter-
acting open quantum systems, as well as the theoretical development of a temporal renormalization
group scheme.
10
Quantum Computing: a future perspective for scientific comput-
ing
A BSTRACT :
Quantum computing is rapidly emerging as a new method of scientific computing. It has the potential
to solve problems much faster than it is possible with classical computers. Examples are applications
in logistics, drug design, medicine finances and many more. In addition, with quantum computers
problems can be tackled that are very hard or even impossible to address with classical computers. In
this talk, we will introduce the Center for Quantum Technology and Applications at DESY in Zeuthen
and give real world examples of applications which can already now be computed on existing quantum
computers.
A BSTRACT :
Collider experiments play a central role in understanding the subatomic structure of matter, as well as
developing and verifying the fundamental theory of elementary particle interactions. However, com-
prehending scattering processes at a fundamental level in theory remains a significant challenge. The
necessarily involved time evolution and the with time rapidly increasing bond dimension in Tensor Net-
works make simulating the scattering process with this classical method challenging. On the other
hand, quantum computers hold great promise to efficiently simulate real-time dynamics of lattice field
theories. In this work, we take the first step in this direction towards simulating fermionic scattering
using a digital quantum computing approach. Specifically, we propose a method based on Givens
rotation to prepare the initial state of the fermionic scattering process, which consists of two fermionic
wave packets with opposite momenta. With a time evolution operator based on the underlying Hamil-
tonian acting on the initial state, the two fermionic wave packets propagate and eventually interact with
each other. Using the lattice Thirring model as the test bed, monitoring the particle density and the
entropy produced during the scattering process, we observe an elastic scattering between fermions
and anti-fermions in the strong interaction region. In addition, we perform a small-scale demonstration
on IBM’s quantum hardware, showing that our method is suitable for current and near-term quantum
devices.
t.b.a.
A BSTRACT :
t.b.a.
11
Simulating the dynamics of open quantum-systems with tensor
networks
A BSTRACT :
Tensor-network methods have become the de-facto standard to describe low-dimensional quantum
systems where methodical advancements allow to efficiently simulate the out-of-equilibrium dynamics
of strongly correlated systems. In this talk, I will give an introduction to one of the standard tech-
niques to compute the time-evolution of quenched quantum systems, the time-dependent variational
principle (TDVP). I will then discuss how TDVP can be employed to simulate the dynamics of projected-
purified matrix-product states, a new representation which we developed to efficiently represent mixed
Fermi-Bose systems, and show how we apply this toolbox to investigate open quantum-systems out-
of-equilibrium.
A BSTRACT :
The strong interaction between excitonic and vibrational degrees of freedom plays a major role in the
quantum dynamics of a variety of complex systems. One prominent example in photophysics is so-
called singlet fission, wherein an incident photon’s absorption in an organic semiconductor prompts
the transformation of a singlet exciton into two distinct triplet excitons. This process offers substantial
potential for improving the efficacy of solar cells by facilitating the production of multiple charge carriers
from a single absorbed photon. In the context of matrix product states (MPS), singlet fission dynamics
have been investigated either via unitary dynamics or by means of master equations. The first approach
is limited in the reachable system sizes and simulation times by the rapid growth of entanglement,
while the second one tends to be inaccurate for realistic spectral densities. We partially overcome both
limitations by applying the numerically exact mesoscopic leads method, which is based on Markovian
embedding. This is combined with state-of-the-art time-evolution methods for MPS and a pure-state
unraveling method. We study singlet fission in a rubrene crystal, modeling five excitonic and up to 200
vibrational degrees of freedom. We find that we can consider large systems and reach long simulation
times at a relatively low computational cost. The combined methods that we describe and benchmark
can be applied to the study of many intriguing phenomena, ranging from photosynthesis to optically-
induced high-temperature superconductivity.
12
Compositional tensor networks
A BSTRACT :
We consider deep networks built by the composition of function tensor networks with tree structure
(HT/TT), but focusing function tensor trains. Compositional sparsity, coined by W. Dahmen, has been
inspired by Deep Neural Networks. The individual neural network layers are replaced by tree-based
tensor networks (HT/TT). The key is that we use the composition of non-linear functions as a numerical
tool for approximation building deep tensor networks. As a particular example we consider flow-driven
deep tensor networks and its relation to optimal control, optimal transport and variational mean filed
games. By a variational Monte Carlo approach we derive an optimization procedure for the computation
(training) of the tensors.
13
Participants (in alphabetical order)
14
Venue
Zuse Institute Berlin
The TMQS workshop will be held at the Zuse Institute Berlin in the lecture hall (room
2005) next to the lobby (main entrance of the circular building).
Public Transport
As for public transportation, ZIB is most easily reached via the Dahlem Dorf subway
station (U3, M11) or the bus stops Arnimallee (X83) and Königin-Luise-Platz/Botanischer
Garten (101), respectively. The site can be accessed from Arnimallee as well as from
Altensteinstraße.
15
Lunch, restaurants & cafes
Coffee breaks and (casual) lunch will be offered to the participants who have indicated
their participation in the registration form. If you desire some more variety, you can
also visit one of the nearby eateries. Here is a small overview of restaurants, cafes,
and supermarkets close to ZIB:
1 eßkultur: Daily changing dishes, incl. vegetarian 3 Alter Krug: German cuisine
2 Luise Dahlem: Steaks, burgers, pizzas, tartes 4 FU Mensa: wide selection of inexpensive food
Ristorante Piaggio: Italian cuisine
5 Pi Cafe: coffee shop with rooftop terrace
really good life: Burgers, incl. vegetarian/vegan
+ coffee shop, bakeries, supermarket 6 Englers: Italian cuisine
16