Quantum Doctoral (QDOC) projects at Aalto University

SPM not only enables visualizing atomic and elec- tronic structures at the nanoscale, but also provides an opportunity to manipulate individual atoms or molecules on surfaces. If this is paired with the recent success in on-surface synthesis of molec- ular systems, in principle we have a practical route to construct molecular nanostructures from the bottom-up in experiments. However, even achieving reproducible bond-breaking remains challeng- ing and prone to error. Scaling this to fabricate large molecular assemblies and engineer complex electronic states requires fine control of hundreds of basic manipulations like dissociation, move- ment, and association. Deep reinforcement learning (DRL) has recently emerged as a promising method to automate various issues in SPM through learning manipulation parameters from a DRL agent’s own experience and behaviour in an interactive environment. We earlier designed a deep re- inforcement learning algorithm to select continuous action parameters to steer the motion of atoms, including SPM tip-start and -end positions, bias voltage and tunnelling conductance. In ACUMEN, we will extend this to the full set of actions needed to autonomously dissociate, manipulate and associate molecules to form novel molecular structures. The project’s ultimate objective will be to autonomously fabricate a user designed spin state from nanographene precursors.

Scanning Probe Microscopy (SPM) techniques have been routinely used to characterize atomic and electronic structure with atomic resolution, but they lack reliable chemical information of the structures. Moreover, classical spectroscopy techniques such as infrared and Raman, which can be used to identify the molecular composition of complex materials without damaging the sample, are strongly limited by the Abbe diffraction limit. The sensitivity of Raman spectroscopy can be increased immensely in TERS, by using a localized plas- monic field produced at a tip apex when a scanning probe microscope is integrated into the Raman spectrometer. TERS can then achieve ultrahigh spatial resolution in optical imaging, even resolv- ing the vibrational modes of molecular structures. However, data interpretation and quantitative analysis for complex molecular structures are challenging due to the complex nature of contrast in the TERS experiment. For instance, systems that may contain radical moieties, metal centers, or molecules that present spin crossover, have not even yet been considered theoretically.
MASQUE will establish a comprehensive database of simulated TERS images and then use this data to train a deep learning infrastructure to recognize the fingerprint characteristics of TERS images. By doing so, a rapid route to an atomically and chemically resolved map of the systems being studied can be achieved by matching descriptors predicted from experimental data with those in the dataset of simulated TERS. This can be linked to a detailed electronic characterization via SPM images and spectroscopy. Ultimately, the MASQUE approach can be applied during experiments via an automation strategy.

Practical applications of quantum computers include cryptanalysis, which is based on arithmetic circuits. Quantum computing architectures with layouts of more than two dimensions, such as neutral-atom or 3D integrated superconducting chips, have the potential to be more resource-efficient with respect to the circuit depth of practical quantum algorithms. In the short term, quantum circuit optimization is motivated by the high noise levels of the NISQ devices: shallower circuits are more noise-resilient. In the long term, optimization methods will lower the number of qubits required to achieve fault-tolerance: depth necessitates stronger error-correction which in turn increases the number of qubits. We call a 2D+ architecture, one where the qubits are either arranged in 3D, or where qubit shuttling is possible in 2D.
We quantify the feasibility of large scale quantum computing for military and industrial applications. To the best of our knowledge this is the first academic project which investigates the co-design of quantum circuits for cryptanalysis together with the architecture of the quantum computers. The project will generate the technology, and the necessary teaching materials related to the feasibility of practical large scale quantum applications: a) highlight the architectural advantages of 2D+ qubit architectures; b) advance state of the art resource estimation of quantum cryptanalysis circuits.

Mesoscopic physics concerns the phase-coherent transport of electrons in low-dimensional
conductors such as quantum point contacts and nanowires. Experiments have shown that time-dependent electric currents can be generated by modulating the voltage difference across a conductor. By contrast, much less is known about the generation of heat currents by time-dependent temperature biases. This project pushes vigorously into this mostly uncharted territory of dynamic heat transport in quantum-electronic conductors, which is an important topic for many quantum technologies. We will develop a dynamic scattering theory of time- dependent heat currents, which is a central task for the emerging field of quantum thermodynamics. Based on this framework, we will investigate the quantum properties of heat pulses, including their quantum statistics, their interferometry, and their interactions, with the goal of devising efficient schemes for controlling and managing waste heat in quantum circuits.

Many aspects of modern life depend on quick and secure telecommunication networks and information processing. To achieve even better, securer, and faster data sharing than currently available, experts are developing future devices using the principles of quantum physics. These designs depend on individual photons to encode and distribute information across quantum networks and among quantum chips. Yet, the current devices to generate single photons lack the accuracy and stability necessary for quantum information technology. The find of quantum defects in van der Waals (2D vdW) materials presents possibilities to overcome the problems with the NV defects in diamond. These materials exhibit unique properties like interlayer coupling, unsaturated surfaces and planar structure that are advantageous for photon emitter manufacturing. It has been shown that single, identical photons can be generated on demand by positioning a metallic probe over a designated point in a common 2D semiconductor material.

This project researches and develops novel techniques for characterizing and optimizing non-functional parameters, including performance and interaction couplings, for emerging complex software in hybrid quantum computing environments. We study interaction models, service models, composition models, and non-functional characteristics for hybrid quantum computing applications to develop techniques to analyze the software performance and to optimize the execution of the applications in hybrid quantum computing environments.

Variational quantum algorithms (VQAs) are hybrid classical-quantum algorithms, which have been developed for applications in quantum simulation and machine learning. VQAs leverage simultaneously the power of both classical high-performance computing and quantum computing. It is essential to design the algorithms around the physical limitations of the current and near-term hardware platforms. In this project we develop methods to generate hardware-optimized variational quantum circuits for near-term quantum computers.

Silicon photonics is seen as the most promising future photonic alternative to electronic integrated circuits. It is further known that for quantum information, high-dimensional quantum states are superior due to their increased information capacity and simplified processing. In this project, we study how the existing silicon technology can incorporate such high-dimensional quantum states in the form of the multimode light. We seek to confirm the non-classical properties of multimode light in a silicon chip, with the ultimate goal of creating light-based integrated quantum devices.

Van der Waals materials provide a highly flexible platform to engineer exotic quantum states. Heavy-fermion quantum materials harvest effects from quantum magnetism, leading to exotic correlated states controlled by macroscopic entanglement. In this project, we will develop strategies to control heavy-fermion van der Waals materials using twisting, gating, and substrate engineering, with the objective of controlling correlated states that can ultimately enable a future generation of heavy-fermion quantum devices.

Quantum materials provide a platform to create new exotic quantum excitations. In this project, we will develop machine learning methodologies to infer the nature of quantum excitations in tip-enhanced optical scanning tunneling spectroscopy experiments. This project will combine the theory of quantum materials, machine learning, and scanning tunneling spectroscopy to detect exotic quantum excitations that ultimately enable new quasiparticle-based quantum technologies.

Semiconducting materials with induced superconductivity can demonstrate Josephson physics that can be electrostatically tuned in-situ, but have been plagued with challenges in scalable fabrication. Recent demonstrations of wafer-scale production of these materials now make it possible to develope Josephson field effect transistor (JoFET) components that may be useful to solve the scalability bottleneck in developing quantum technologies. In this doctoral project, we will develop one elusive yet essential component for quantum computing – a Josephson microwave isolator.

We study the theoretical foundations of distributed quantum computing. While many major questions related to the foundations of classical distributed algorithms have been solved recently, very little is currently known about the potential for quantum advantage in the distributed setting. We focus on distributed graph algorithms, and we study the quantum analogue of the classical LOCAL model of distributed computing. We aim at establishing fundamental limits on distributed quantum advantage, and identifying problems that can be solved more efficiently with distributed quantum algorithms.

Nano-mechanical oscillators coupled to microwave resonators have been used by the host group to engineer quantum states of motion. One type of oscillator has been particularly successful: drum membranes, because they are light, and their motion is easy to detect. Interestingly, owing to the small separation between the vibrating membrane and the substrate surface facing it, these resonators are in principle largely affected by the Casimir force. The latter is a force between reflective (metallic) surfaces that can be understood as a manifestation of quantum fluctuations of the electromagnetic field. In this project, we will use this fact to attempt the first measurement of the Casimir force between superconducting surfaces. Superconductors essentially behave like other metallic mirrors, except that, contrary to regular metals, they are perfect, non-lossy mirrors for small electromagnetic frequencies (below the so-called superconducting gap). Therefore, the measurement of the Casimir effect in superconductors has been proposed to isolate the impact of low frequency quantum fluctuations and shed light on their currently debated contribution to the Casimir force.

The tiny electric currents flowing in active neurons in our brains generate magnetic fields that are detectable just outside the head. Magnetoencephalography (MEG) refers to measuring these fields to obtain information about the timing and location of active brain regions. These femtotesla-level fields have traditionally been measured using superconducting quantum interference devices (SQUIDs). However, these sensors require cryogenic temperatures to operate, which makes MEG a clumsy and expensive technique. Fortunately, recent advances in quantum optics have improved the sensitivity of optically pumped magnetometers (OPM) such that they can pick up these weak field and enable sensor arrays that adapt to the head size, thus at least theoretically providing higher sensitivity and spatial resolution. This project focuses on improving OPM sensors further by extending their bandwidth and improving their sensitivity and tolerance against interfering fields. The aim is to make OPMs more suitable for MEG and thereby, in the longer run, to increase the adoption of MEG in basic neuroscience and in clinical diagnostics.

In the past, there has been a trade-off in making faster qubit gates with the expense of shorter
qubit lifetimes or more power dissipation. However, together with Hat Lab [arXiv:2306.1062],
we have been the first [arXiv:2402.08906] to demonstrate that qubits can be simultaneously
protected from decay to the drive line and quickly driven, thanks to a filter at the qubit frequency
and driving at 1/3 frequency. This project studies the limits of driving superconducting qubits
with frequencies below or above the qubit frequency, extending up to optical frequencies, with
the aim of introducing a new standard of driving long-lived qubits.

In collaboration with VTT and Bluefors, Aalto university has built the world-most-sensitive bolometers, i.e., thermal radiation sensors [Kokkoniemi et al., Nature 586, 47 (2020)]. We will combine this unique device with cryogenic filters and tailored superconducting qubits to demonstrate a breakthrough in the readout of superconducting qubits and a spectrometer to sense the lowest amount of noise subject to the cryogenic environment.

Arrays of SQUID loops and their asymmetric counterparts, SNAILS, provide schemes for quantum limited amplification and sensing, as well as promising settings for quantum simulations of cosmological effects (Hawking radiation, entropy of black holes, etc.). This thesis project will concentrate on microwave experiments on such arrays and investigate, besides quantum simulations, possibilities for developing new kinds of magnetometer and amplification schemes (for example directional TWPAs) using these systems.

The project will develop new superconducting quantum materials needed to realize so-called topological qubits, where the quantum information is protected by the system topology. In particular, we will focus on boosting the topological gap of the system that should allow device operation at higher temperatures. The project builds on recent breakthrough results from the host group and utilizes an effective collaboration network within the Quantum Materials programme of the Finnish Quantum Flagship.

The project will develop new superconducting quan- tum materials needed as building blocks for topological qubits, where the quantum information is protected by the system topology. We will focus on systems where the system topology can be controlled with external electric fields, which would be extremely desirable for device applications. The project builds on recent breakthrough results from the host group and utilizes an effective col- laboration network within the Quantum Materials programme of the Finnish Quantum Flagship.

The PhD projects will focus in understanding how light stimulation controlled at the level of single light quanta can be directly linked to the neural outputs of the retina and human and mouse perception. Building upon our current knowledge we will attack the following specific questions: 1) What retinal circuits and retinal output neurons contribute the human perception of the dimmest light decrements, 2) What constitutes the key sources of variance in visual perception in conditions, where light inputs are precise and where variance is removed from input photon counts, 3) What is the perceptual memory component engaged in the detection of the dimmest light increments and decrements.

We will propose practical solutions for the quantum- security problem of backscatter communication Internet-of-Things (IoT) systems. Specifically, we will provide quantum solutions for the identity authentication problem, studying the possibility of using RFID passive quantum tags, leveraging their no-cloning feature for unconditional authentication. We will explore quantum security solutions for quantum communication based on covertness and quantum computational hybrid models. The implementation of these ideas will be supported by an interdisciplinarity and international team of renowned scientists. This project contributes toquantum science and technology, as backscatter IoT quantum security is currently an open problem.

This project addresses critical limitations in quantum state transfer between microwave photons and optical photons, which is essential for realizing the full potential of quantum networks. The state-of-the-art microwave-to-optics transduction efficiency of on-chip systems is five orders of magnitude smaller than the required quantum limit. This doctoral project investigates a new hybrid quantum system that couples the spin of photoluminescent Er ions to magnons in an Er-doped YIG optomagnonic cavity. Calculations based on recently developed models indicate that the external microwave-to-optics transduction efficiency of this system could reach unity. If successful, this would enable the readout of a superconducting quantum qubit via an optomagnonic interface.

Sensing is a critical technology, permeating technologies with societal impact ranging from medicine to monitoring of climate change. This project aims at reaching unprecedented detection sensitivities by employing superconducting devices operated in the microwave frequency range.

While universal quantum computing is the stated Holy Grail of Quantum Technologies 2.0, its implementation on a scale to solve high-impact problems remains elusive. Quantum Reservoir Computing (QRC) is a new class of (noisy) reservoir computing (RC), where the input signal is mapped on the state of the reservoir (digital or analogue structure with a large space of states and nonlinear response). The reservoir state provides the input for a conventional neural network and nonlinear evolution allows the use of simple training methods for the readout, which significantly speeds up training and makes the approach feasible for real-time tasks. The reservoir can be chosen to optimise its performance with respect to the expected type of inputs. QRC systems are small, their realisation and running inexpensive, and training can be simple and fast.

AI techniques, predominantly machine learning and deep learning algorithms, have emerged as
powerful tools for encoding structural, relational, and spatial data; and graph neural networks (GNNs) in particular have enabled state-of-the-art performance across drug discovery tasks (Garg
COSB, 2024), including protein design (Ingraham et al. NeurIPS, 2019), molecule generation and property optimization (Verma et al. NeurIPS, 2022), and antibody design (Verma et al. ICML, 2023). However, methods based on message-passing GNNs are known to have representational limitations (Garg et al. ICML, 2020), and struggle particularly with out-of-distribution generalization (OOD), namely, ability to perform well on new molecular data that might vary significantly, distributionally, from the training data. This severely limits their benefits from a practical perspective.
We will develop principled Quantum GNN algorithms to address these shortcomings of the classical
GNNs, to advance drug discovery. Recently, methods based on persistent homology - a key tool from topological data analysis - have been proposed to augment the capabilities of GNNs (Immonen et al.
NeurIPS, 2023); we will leverage some of the insights from that work to design novel quantum
topological descriptors and investigate their efficacy in different drug discovery settings.

This PhD project delves into the uncharted territories of artificially synthesized quantum materials, leveraging the unique properties of moiré heterostructures to engineer emergent and quantum behaviours critical to condensed-matter physics. Through precise manipulation of matter's (quasi-)particles and various interactions, we aim to unveil new unconventional quantum states, particularly focusing on the creation of new ferroelectrics and magnetics via innovative van der Waals moiré structures. The sensitivity of layered van der Waals structures to the stacking arrangement of their constituent layers opens up a realm of possibilities for engineering emergent phenomena, such as superconductivity, correlated insulation, and magnetism (Science, 379, eadg0014 (2023)). By exploiting this sensitivity, this project intends to artificially engineer novel unconventional quantum states, with an initial emphasis on engineering vector electric dipoles and spins for tunable topological states, vector textures and valley properties, which do not exist in their bulk counterparts. This groundbreaking research aims to push the boundaries of our understanding and utilization of quantum materials, setting the stage for significant advances in condensed matter physics.