@phdthesis{7514, abstract = {We study the interacting homogeneous Bose gas in two spatial dimensions in the thermodynamic limit at fixed density. We shall be concerned with some mathematical aspects of this complicated problem in many-body quantum mechanics. More specifically, we consider the dilute limit where the scattering length of the interaction potential, which is a measure for the effective range of the potential, is small compared to the average distance between the particles. We are interested in a setting with positive (i.e., non-zero) temperature. After giving a survey of the relevant literature in the field, we provide some facts and examples to set expectations for the two-dimensional system. The crucial difference to the three-dimensional system is that there is no Bose–Einstein condensate at positive temperature due to the Hohenberg–Mermin–Wagner theorem. However, it turns out that an asymptotic formula for the free energy holds similarly to the three-dimensional case. We motivate this formula by considering a toy model with δ interaction potential. By restricting this model Hamiltonian to certain trial states with a quasi-condensate we obtain an upper bound for the free energy that still has the quasi-condensate fraction as a free parameter. When minimizing over the quasi-condensate fraction, we obtain the Berezinskii–Kosterlitz–Thouless critical temperature for superfluidity, which plays an important role in our rigorous contribution. The mathematically rigorous result that we prove concerns the specific free energy in the dilute limit. We give upper and lower bounds on the free energy in terms of the free energy of the non-interacting system and a correction term coming from the interaction. Both bounds match and thus we obtain the leading term of an asymptotic approximation in the dilute limit, provided the thermal wavelength of the particles is of the same order (or larger) than the average distance between the particles. The remarkable feature of this result is its generality: the correction term depends on the interaction potential only through its scattering length and it holds for all nonnegative interaction potentials with finite scattering length that are measurable. In particular, this allows to model an interaction of hard disks.}, author = {Mayer, Simon}, issn = {2663-337X}, pages = {148}, publisher = {Institute of Science and Technology Austria}, title = {{The free energy of a dilute two-dimensional Bose gas}}, doi = {10.15479/AT:ISTA:7514}, year = {2020}, } @inproceedings{10672, abstract = {The family of feedback alignment (FA) algorithms aims to provide a more biologically motivated alternative to backpropagation (BP), by substituting the computations that are unrealistic to be implemented in physical brains. While FA algorithms have been shown to work well in practice, there is a lack of rigorous theory proofing their learning capabilities. Here we introduce the first feedback alignment algorithm with provable learning guarantees. In contrast to existing work, we do not require any assumption about the size or depth of the network except that it has a single output neuron, i.e., such as for binary classification tasks. We show that our FA algorithm can deliver its theoretical promises in practice, surpassing the learning performance of existing FA methods and matching backpropagation in binary classification tasks. Finally, we demonstrate the limits of our FA variant when the number of output neurons grows beyond a certain quantity.}, author = {Lechner, Mathias}, booktitle = {8th International Conference on Learning Representations}, location = {Virtual ; Addis Ababa, Ethiopia}, publisher = {ICLR}, title = {{Learning representations for binary-classification without backpropagation}}, year = {2020}, } @inproceedings{10673, abstract = {We propose a neural information processing system obtained by re-purposing the function of a biological neural circuit model to govern simulated and real-world control tasks. Inspired by the structure of the nervous system of the soil-worm, C. elegans, we introduce ordinary neural circuits (ONCs), defined as the model of biological neural circuits reparameterized for the control of alternative tasks. We first demonstrate that ONCs realize networks with higher maximum flow compared to arbitrary wired networks. We then learn instances of ONCs to control a series of robotic tasks, including the autonomous parking of a real-world rover robot. For reconfiguration of the purpose of the neural circuit, we adopt a search-based optimization algorithm. Ordinary neural circuits perform on par and, in some cases, significantly surpass the performance of contemporary deep learning models. ONC networks are compact, 77% sparser than their counterpart neural controllers, and their neural dynamics are fully interpretable at the cell-level.}, author = {Hasani, Ramin and Lechner, Mathias and Amini, Alexander and Rus, Daniela and Grosu, Radu}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, issn = {2640-3498}, location = {Virtual}, pages = {4082--4093}, title = {{A natural lottery ticket winner: Reinforcement learning with ordinary neural circuits}}, year = {2020}, } @inproceedings{10190, abstract = {The verification of concurrent programs remains an open challenge, as thread interaction has to be accounted for, which leads to state-space explosion. Stateless model checking battles this problem by exploring traces rather than states of the program. As there are exponentially many traces, dynamic partial-order reduction (DPOR) techniques are used to partition the trace space into equivalence classes, and explore a few representatives from each class. The standard equivalence that underlies most DPOR techniques is the happens-before equivalence, however recent works have spawned a vivid interest towards coarser equivalences. The efficiency of such approaches is a product of two parameters: (i) the size of the partitioning induced by the equivalence, and (ii) the time spent by the exploration algorithm in each class of the partitioning. In this work, we present a new equivalence, called value-happens-before and show that it has two appealing features. First, value-happens-before is always at least as coarse as the happens-before equivalence, and can be even exponentially coarser. Second, the value-happens-before partitioning is efficiently explorable when the number of threads is bounded. We present an algorithm called value-centric DPOR (VCDPOR), which explores the underlying partitioning using polynomial time per class. Finally, we perform an experimental evaluation of VCDPOR on various benchmarks, and compare it against other state-of-the-art approaches. Our results show that value-happens-before typically induces a significant reduction in the size of the underlying partitioning, which leads to a considerable reduction in the running time for exploring the whole partitioning.}, author = {Chatterjee, Krishnendu and Pavlogiannis, Andreas and Toman, Viktor}, booktitle = {Proceedings of the 34th ACM International Conference on Object-Oriented Programming, Systems, Languages, and Applications}, issn = {2475-1421}, keywords = {safety, risk, reliability and quality, software}, location = {Athens, Greece}, publisher = {ACM}, title = {{Value-centric dynamic partial order reduction}}, doi = {10.1145/3360550}, volume = {3}, year = {2019}, } @phdthesis{69, abstract = {A qubit, a unit of quantum information, is essentially any quantum mechanical two-level system which can be coherently controlled. Still, to be used for computation, it has to fulfill criteria. Qubits, regardless of the system in which they are realized, suffer from decoherence. This leads to loss of the information stored in the qubit. The upper bound of the time scale on which decoherence happens is set by the spin relaxation time. In this thesis I studied a two-level system consisting of a Zeeman-split hole spin confined in a quantum dot formed in a Ge hut wire. Such Ge hut wires have emerged as a promising material system for the realization of spin qubits, due to the combination of two significant properties: long spin coherence time as expected for group IV semiconductors due to the low hyperfine interaction and a strong valence band spin-orbit coupling. Here, I present how to fabricate quantum dot devices suitable for electrical transport measurements. Coupled quantum dot devices allowed the realization of a charge sensor, which is electrostatically and tunnel coupled to a quantum dot. By integrating the charge sensor into a radio-frequency reflectometry setup, I performed for the first time single-shot readout measurements of hole spins and extracted the hole spin relaxation times in Ge hut wires.}, author = {Vukušić, Lada}, issn = {2663-337X}, pages = {103}, publisher = {Institute of Science and Technology Austria}, title = {{Charge sensing and spin relaxation times of holes in Ge hut wires}}, doi = {10.15479/AT:ISTA:TH_1047}, year = {2018}, } @article{104, abstract = {The biotrophic pathogen Ustilago maydis, the causative agent of corn smut disease, infects one of the most important crops worldwide – Zea mays. To successfully colonize its host, U. maydis secretes proteins, known as effectors, that suppress plant defense responses and facilitate the establishment of biotrophy. In this work, we describe the U. maydis effector protein Cce1. Cce1 is essential for virulence and is upregulated during infection. Through microscopic analysis and in vitro assays, we show that Cce1 is secreted from hyphae during filamentous growth of the fungus. Strikingly, Δcce1 mutants are blocked at early stages of infection and induce callose deposition as a plant defense response. Cce1 is highly conserved among smut fungi and the Ustilago bromivora ortholog complemented the virulence defect of the SG200Δcce1 deletion strain. These data indicate that Cce1 is a core effector with apoplastic localization that is essential for U. maydis to infect its host.}, author = {Seitner, Denise and Uhse, Simon and Gallei, Michelle C and Djamei, Armin}, journal = {Molecular Plant Pathology}, number = {10}, pages = {2277 -- 2287}, publisher = {Wiley}, title = {{The core effector Cce1 is required for early infection of maize by Ustilago maydis}}, doi = {10.1111/mpp.12698}, volume = {19}, year = {2018}, } @phdthesis{1130, abstract = {In this thesis we present a computer-aided programming approach to concurrency. Our approach helps the programmer by automatically fixing concurrency-related bugs, i.e. bugs that occur when the program is executed using an aggressive preemptive scheduler, but not when using a non-preemptive (cooperative) scheduler. Bugs are program behaviours that are incorrect w.r.t. a specification. We consider both user-provided explicit specifications in the form of assertion statements in the code as well as an implicit specification. The implicit specification is inferred from the non-preemptive behaviour. Let us consider sequences of calls that the program makes to an external interface. The implicit specification requires that any such sequence produced under a preemptive scheduler should be included in the set of sequences produced under a non-preemptive scheduler. We consider several semantics-preserving fixes that go beyond atomic sections typically explored in the synchronisation synthesis literature. Our synthesis is able to place locks, barriers and wait-signal statements and last, but not least reorder independent statements. The latter may be useful if a thread is released to early, e.g., before some initialisation is completed. We guarantee that our synthesis does not introduce deadlocks and that the synchronisation inserted is optimal w.r.t. a given objective function. We dub our solution trace-based synchronisation synthesis and it is loosely based on counterexample-guided inductive synthesis (CEGIS). The synthesis works by discovering a trace that is incorrect w.r.t. the specification and identifying ordering constraints crucial to trigger the specification violation. Synchronisation may be placed immediately (greedy approach) or delayed until all incorrect traces are found (non-greedy approach). For the non-greedy approach we construct a set of global constraints over synchronisation placements. Each model of the global constraints set corresponds to a correctness-ensuring synchronisation placement. The placement that is optimal w.r.t. the given objective function is chosen as the synchronisation solution. We evaluate our approach on a number of realistic (albeit simplified) Linux device-driver benchmarks. The benchmarks are versions of the drivers with known concurrency-related bugs. For the experiments with an explicit specification we added assertions that would detect the bugs in the experiments. Device drivers lend themselves to implicit specification, where the device and the operating system are the external interfaces. Our experiments demonstrate that our synthesis method is precise and efficient. We implemented objective functions for coarse-grained and fine-grained locking and observed that different synchronisation placements are produced for our experiments, favouring e.g. a minimal number of synchronisation operations or maximum concurrency.}, author = {Tarrach, Thorsten}, issn = {2663-337X}, pages = {151}, publisher = {Institute of Science and Technology Austria}, title = {{Automatic synthesis of synchronisation primitives for concurrent programs}}, doi = {10.15479/at:ista:1130}, year = {2016}, } @phdthesis{1401, abstract = {The human ability to recognize objects in complex scenes has driven research in the computer vision field over couple of decades. This thesis focuses on the object recognition task in images. That is, given the image, we want the computer system to be able to predict the class of the object that appears in the image. A recent successful attempt to bridge semantic understanding of the image perceived by humans and by computers uses attribute-based models. Attributes are semantic properties of the objects shared across different categories, which humans and computers can decide on. To explore the attribute-based models we take a statistical machine learning approach, and address two key learning challenges in view of object recognition task: learning augmented attributes as mid-level discriminative feature representation, and learning with attributes as privileged information. Our main contributions are parametric and non-parametric models and algorithms to solve these frameworks. In the parametric approach, we explore an autoencoder model combined with the large margin nearest neighbor principle for mid-level feature learning, and linear support vector machines for learning with privileged information. In the non-parametric approach, we propose a supervised Indian Buffet Process for automatic augmentation of semantic attributes, and explore the Gaussian Processes classification framework for learning with privileged information. A thorough experimental analysis shows the effectiveness of the proposed models in both parametric and non-parametric views.}, author = {Sharmanska, Viktoriia}, issn = {2663-337X}, pages = {144}, publisher = {Institute of Science and Technology Austria}, title = {{Learning with attributes for object recognition: Parametric and non-parametrics views}}, doi = {10.15479/at:ista:1401}, year = {2015}, } @phdthesis{1405, abstract = {Motivated by the analysis of highly dynamic message-passing systems, i.e. unbounded thread creation, mobility, etc. we present a framework for the analysis of depth-bounded systems. Depth-bounded systems are one of the most expressive known fragment of the π-calculus for which interesting verification problems are still decidable. Even though they are infinite state systems depth-bounded systems are well-structured, thus can be analyzed algorithmically. We give an interpretation of depth-bounded systems as graph-rewriting systems. This gives more flexibility and ease of use to apply depth-bounded systems to other type of systems like shared memory concurrency. First, we develop an adequate domain of limits for depth-bounded systems, a prerequisite for the effective representation of downward-closed sets. Downward-closed sets are needed by forward saturation-based algorithms to represent potentially infinite sets of states. Then, we present an abstract interpretation framework to compute the covering set of well-structured transition systems. Because, in general, the covering set is not computable, our abstraction over-approximates the actual covering set. Our abstraction captures the essence of acceleration based-algorithms while giving up enough precision to ensure convergence. We have implemented the analysis in the PICASSO tool and show that it is accurate in practice. Finally, we build some further analyses like termination using the covering set as starting point.}, author = {Zufferey, Damien}, issn = {2663-337X}, pages = {134}, publisher = {Institute of Science and Technology Austria}, title = {{Analysis of dynamic message passing programs}}, doi = {10.15479/at:ista:1405}, year = {2013}, } @phdthesis{2964, abstract = {CA3 pyramidal neurons are important for memory formation and pattern completion in the hippocampal network. These neurons receive multiple excitatory inputs from numerous sources. Therefore, the rules of spatiotemporal integration of multiple synaptic inputs and propagation of action potentials are important to understand how CA3 neurons contribute to higher brain functions at cellular level. By using confocally targeted patch-clamp recording techniques, we investigated the biophysical properties of rat CA3 pyramidal neuron dendrites. We found two distinct dendritic domains critical for action potential initiation and propagation: In the proximal domain, action potentials initiated in the axon backpropagate actively with large amplitude and fast time course. In the distal domain, Na+-channel mediated dendritic spikes are efficiently evoked by local dendritic depolarization or waveforms mimicking synaptic events. These findings can be explained by a high Na+-to-K+ conductance density ratio of CA3 pyramidal neuron dendrites. The results challenge the prevailing view that proximal mossy fiber inputs activate CA3 pyramidal neurons more efficiently than distal perforant inputs by showing that the distal synapses trigger a different form of activity represented by dendritic spikes. The high probability of dendritic spike initiation in the distal area may enhance the computational power of CA3 pyramidal neurons in the hippocampal network. }, author = {Kim, Sooyun}, issn = {2663-337X}, pages = {65}, publisher = {Institute of Science and Technology Austria}, title = {{Active properties of hippocampal CA3 pyramidal neuron dendrites}}, year = {2012}, } @phdthesis{3962, author = {Pflicke, Holger}, issn = {2663-337X}, publisher = {Institute of Science and Technology Austria}, title = {{Dendritic cell migration across basement membranes in the skin}}, year = {2010}, }