VDSP Science Day 2025 - Highlight Talks

Experimental quantum-enhanced kernel-based machine learning on a photonic processor

Zhenghao Yin (Supervisor: Philip Walther)

Learning from data and training specific models have been among the most challenging tasks for conventional computing machines, ranging from image recognition to natural language processing. Quantum mechanics offers a solution by utilizing superposition and entanglement to accelerate the solving of these tasks on quantum machines, a field known as quantum machine learning.

Here, we introduce quantum machine learning based on kernel methods, which is deployed on a photonic quantum processor. We show that our protocol outperforms state-of-the-art kernel methods such as gaussian and neural tangent kernels by exploiting quantum interference, and provides further improvements in accuracy by offering classical coherence.

Corrugation-Dominated Mechanical Softening of Defect-Engineered Graphene

Wael Joudi (Supervisor: Jani Kotakoski)

Graphene is a two-dimensional material that exhibits exotic properties compared to its three-dimensional counterpart graphite. For example, it is attributed with highest intrinsic stiffness ever measured [1], as well as an electric and thermal conductivity exceeding those of copper [2]. These properties arise from the material’s chemical structure, and therefore numerous studies have focused on tailoring them by modifying that structure. For example, the impact of atomic vacancies on the mechanical stiffness has been experimentally studied, but with contradictory outcomes with one case leading to a drastic boost while in the other a clear decrease is observed [3,4].

Here, we address this discrepancy by conducting the experiment in a clean and controlled ultra-high vacuum environment and find a drastic decrease in stiffness with increasing vacancy density. The stiffness reduction is explained by a semi-empirical model, in which vacancy-induced corrugation is the dominant factor for the softening of the material. Atomistic simulations reveal that this corrugation is only observed for vacancies with two or more missing carbon atoms, which is also reflected by the model. Lastly, we show that the root of this discrepancy is linked to omnipresent carbon-based surface contamination in non-cleaned samples, in which we observe roughly a doubling of the stiffness after introducing vacancies. The model presented in this work opens a way for tailoring the softness of the material for applications that require a certain amount of stretchability, such as wearable electronics.

 

[1] C. Lee et al., Science 321, 385 (2008)
[2] Novoselov et al., Nature 438, 197-200 (2005)
[3] López-Polín et al., Nature Physics 11, 26 (2015)
[4] Zandiatashbar et al., Nature Communications 5, 3186 (2014)

Original Publication: Joudi et al., Physical Review Letters 134, 166102 (2025)

Thermophoresis of polymers by mesoscale simulations

Lisa Sappl (Supervisor: Christos Likos)

From experimental research, it is known that particles immersed in a liquid with a temperature gradient exhibit thermophoretic behavior, i.e. they move along the temperature gradient in either direction. Even today, the underlying physics of this phenomenon are not completely clear. To gain a deeper understanding into this matter, in this work the thermophoretic motion of polymers is simulated using multiparticle collision dynamics, a well-established mesoscale simulation technique. Besides the intermolecular interaction between its monomers, the polymer also interacts with the solvent via a Lennard-Jones-like potential. The polymer-solvent interaction is modified by tuning an interaction parameter λ in the Lennard-Jones like potential, and its influence on the thermophoretic mobility DT of the polymer is investigated. We found, that with a purely repulsive polymer-solvent interaction, the polymer exhibits thermophilic behavior. To display thermophobic behavior, the polymer-solvent potential is required to have attractive areas.

Long-living magnons at the quantum limit

Rostyslav Serha (Supervisor: Andrii Chumak)

Magnons are the quanta of spin waves, the eigenexcitations of ordered magnetic media. These charge-neutral quasiparticles carry energy and momentum and are central to magnonics, which studies them as data carriers for computation and information technologies. A long-standing bottleneck for quantum experiments with magnons has been their short lifetimes, with maximum values of about one microsecond. In our work, we demonstrate a breakthrough, showing that in our best samples magnon lifetimes reach up to 18 microseconds at millikelvin temperatures. We generate these magnons by exploiting the readily accessible intrinsic nonlinearity of the system in a three-magnon splitting experiment. The lifetime of magnons excited in this way is determined by the power threshold required to reach the nonlinear regime. As the temperature is lowered, the threshold power decreases, and the accessible lifetime increases accordingly. This result opens the door to novel experiments and applications in quantum magnonics, allowing magnons to serve as effective carriers of quantum information.

Quantum Vacua of Chern-Simons Matter Theories

Sinan Moura Soysüren (Supervisor: Marcus Sperling)

Supersymmetric quantum field theories with eight supercharges exhibit moduli spaces of vacua with rich mathematical structure. Amongst the tools to study such moduli spaces, the magnetic quiver program offers a systematic approach which has proven very useful. In this talk, I will present a first approach to a magnetic quiver framework that captures the moduli space of N=3 and N=4 orthosymplectic Chern-Simons matter theories in three dimensions based on work with F. Marino and M. Sperling.