arXiv subjects

Sebastian Berndt

Publications and source records attributed to Sebastian Berndt.

At least 19 recordsLinked to original sources

arXiv2026

Laser Spectroscopy of Thulium Isotopes Near the (N=82) Shell Closure: Nuclear Moment and Charge Radius of ${}^{152\mathrm{m}}\mathrm{Tm}$

We report on resonance ionization laser spectroscopy measurements performed on both neutron-deficient and neutron-rich thulium ($\mathrm{Tm}, Z=69$) isotopes. Isotope shifts were determined for three atomic ground-state transitions at wavelengths of $389.8\,\mathrm{nm}$, $388.4\,\mathrm{nm}$, and $388.8\,\mathrm{nm}$ in the isotopes ${}^{152\mathrm{m}}\mathrm{Tm}$, ${}^{153}\mathrm{Tm}$, ${}^{154\mathrm{m}}\mathrm{Tm}$, and ${}^{169}\mathrm{Tm}$. In addition, for the $389.8\,\mathrm{nm}$ transition, measurements were extended to the isotope ${}^{170}\mathrm{Tm}$, and the hyperfine structure was partially resolved for all five isotopes. For this transition, the isotope shift could be determined for one more isotope, ${}^{154\mathrm{m}}\mathrm{Tm}$. From the extracted hyperfine coupling constants, the nuclear magnetic dipole moment for ${}^{152\mathrm{m}}\mathrm{Tm}$ was determined for the first time, resulting in $\mu\left({}^{152\mathrm{m}}\mathrm{Tm}\right) = 5.8(3) \mu_\mathrm{N}$. Furthermore, the mean-square nuclear charge radius $\delta\langle r^2\rangle^{152\mathrm{m},169} = -1.86(25)\,\mathrm{fm}^2$ for ${}^{152\mathrm{m}}\mathrm{Tm}$ was extracted from the measured isotope shifts.

nucl-ex↗

arXiv2026

Steganography Without Modification: Hidden Communication via LLM Seeds

We demonstrate that widely deployed Large Language Model (LLM) inference stacks harbor a steganographic channel that requires no modification to model weights, sampling code, or output distributions. The channel exploits a structural property of deterministic decoding: pseudo-random number generators (PRNGs) used in inverse-transform sampling produce a seed-dependent sequence of token-level probability intervals that can be reconstructed from the generated text alone. A sender encodes a secret message in the PRNG seed before generation; a receiver reconstructs the intervals and recovers the seed, and thus the hidden payload, by exhaustive search over the seed space. We formalize two operational modes. In the known-prompt setting, sender and receiver share the prompt, enabling exact interval reconstruction and perfect seed recovery via forced alignment. In the unknown-prompt setting, only the generated text is available; approximate interval reconstruction combined with a maximum-hit-count scoring strategy still permits reliable recovery from sufficiently long outputs. Extensive experiments across six model families and five heterogeneous text domains show that, in the known-prompt setting, full 32-bit seed recovery from the complete 2^32 candidate space achieves up to 100% accuracy, depending on model and text domain, within 300 tokens and under 35 seconds on a single GPU. In the unknown-prompt setting, recovery reaches near-perfect accuracy at 600-800 tokens in about 12 seconds. We further analyze the influence of prompting strategies, tokenization ambiguities, and sampling hyperparameters on channel reliability. Moreover, we discuss several applications of our results: First, it allows for the steganographic transmission of 32 bits, but also shows that ignorance of the prompt is not a valid security assumption.

cs.CR↗

arXiv2025

Lightweight Unified Sha-3/Shake Architecture with a Fault-Resilient State

Hash functions have become a key part of standard Post-quantum cryptography (PQC) schemes, especially Sha-3 and Shake, calling arXiv:submit/7045552 [cs.AR] 3 Dec 2025 for lightweight implementation. A fault-resilient design is always desirable to make the whole PQC system reliable. We, therefore, propose a) a unified hash engine supporting Sha-3 and Shake that follows a byte-wise in-place partitioning mechanism of the so-called Keccak state, and b) an according fault detection for Keccak state protection exploiting its cube structure by deploying two-dimensional parity checks. It outperforms the state-of-the-art (SoA) regarding area requirements at competitive register-level fault detection by achieving 100% detection of three and still near 100% of higher numbers of Keccak state faults. Unlike SoA solutions, the proposed unified hash engine covers all standard hash configurations. Moreover, the introduced multidimensional cross-parity check mechanism achieves a 3.7x improvement in area overhead, with an overall 4.5x smaller fault-resilient engine design as demonstrated in ASIC and FPGA implementations. Integrated into a RISC-V environment, the unified hash engine with the integrated fault-resilient mechanism introduced less than 8% area overhead. Our approach thus provides a robust and lightweight fault-detection solution for protecting hash functions deployed in resource-constrained PQC applications.

cs.AR↗

arXiv2025

Improved limit on the effective electron neutrino mass with the ECHo-1k experiment

The effective electron neutrino mass can be determined by analyzing the endpoint region of the $^{163}$Ho electron capture spectrum, provided a measurement with high energy resolution and high statistics using calorimetric techniques. Here, the Electron Capture in $^{163}$Ho collaboration, ECHo, presents an analysis of the most precise $^{163}$Ho spectrum currently available, obtained with the ECHo-1k experiment and comprising about 200 million events. A very low background rate of $B=9.1(1.3)\times 10^{-6}$ /eV/pixel/day was achieved allowing for a reliable analysis of the endpoint region. The derived endpoint energy $Q = 2862(4)$ eV is in excellent agreement with the one independently determined via Penning-trap mass spectrometry of $Q=2863.2(6)$ eV [1]. The upper limit of the effective electron neutrino mass is improved by almost a factor 2 compared to the lowest current value [2], reaching $m_{\nu_\mathrm{e}} < 15 $ eV/c${^2}$ (90\% credible interval).

hep-ex↗

arXiv2025

Provable one-poison backdoor attacks on linear models and ReLU neural networks

Backdoor poisoning attacks are a threat to machine learning models that are trained on data collected from untrusted sources; these attacks enable attackers to inject malicious behavior into the model that can be triggered by specially crafted inputs. Prior work has established bounds on the success of backdoor attacks and their impact on the benign learning task, however, an open question is what amount of poison data is needed for a successful backdoor attack. Typical attacks either use few samples but need much information about the data points, or need to poison many data points. In this paper, we show that an adversary can mount a one-poison backdoor attack without knowledge of individual training data, requiring only coarse geometric bounds of the input space and training parameters. We identify provably sufficient conditions that allow an adversary with one poison sample with high probability to inject a backdoor into linear models and MLPs. We show that our backdoor has zero backdooring error and the injection does not significantly impact the benign learning task performance.

cs.LG↗

arXiv2023

New Support Size Bounds for Integer Programming, Applied to Makespan Minimization on Uniformly Related Machines

Mixed-integer linear programming (MILP) is at the core of many advanced algorithms for solving fundamental problems in combinatorial optimization. The complexity of solving MILPs directly correlates with their support size, which is the minimum number of non-zero integer variables in an optimal solution. A hallmark result by Eisenbrand and Shmonin (Oper. Res. Lett., 2006) shows that any feasible integer linear program (ILP) has a solution with support size $s\leq 2m\cdot\log(4mΔ)$, where $m$ is the number of constraints, and $Δ$ is the largest coefficient in any constraint. Our main combinatorial result are improved support size bounds for ILPs. To improve granularity, we analyze for the largest $1$-norm $A_{\max}$ of any column of the constraint matrix, instead of $Δ$. We show a support size upper bound of $s\leq m\cdot(\log(3A_{\max})+\sqrt{\log(A_{\max})})$, by deriving a new bound on the -1 branch of the Lambert $\mathcal{W}$ function. Additionally, we provide a lower bound of $m\log(A_{\max})$, proving our result asymptotically optimal. Furthermore, we give support bounds of the form $s\leq 2m\cdot\log(1.46A_{\max})$. These improve upon the previously best constants by Aliev. et. al. (SIAM J. Optim., 2018), because all our upper bounds hold equally with $A_{\max}$ replaced by $\sqrt{m}Δ$. Using our combinatorial result, we obtain the fastest known approximation schemes (EPTAS) for the fundamental scheduling problem of makespan minimization of uniformly related machines ($Q\mid\mid C_{\max}$).

cs.DS↗

arXiv2023

Dash: Accelerating Distributed Private Convolutional Neural Network Inference with Arithmetic Garbled Circuits

The adoption of machine learning solutions is rapidly increasing across all parts of society. As the models grow larger, both training and inference of machine learning models is increasingly outsourced, e.g. to cloud service providers. This means that potentially sensitive data is processed on untrusted platforms, which bears inherent data security and privacy risks. In this work, we investigate how to protect distributed machine learning systems, focusing on deep convolutional neural networks. The most common and best-performing mixed MPC approaches are based on HE, secret sharing, and garbled circuits. They commonly suffer from large performance overheads, big accuracy losses, and communication overheads that grow linearly in the depth of the neural network. To improve on these problems, we present Dash, a fast and distributed private convolutional neural network inference scheme secure against malicious attackers. Building on arithmetic garbling gadgets [BMR16] and fancy-garbling [BCM+19], Dash is based purely on arithmetic garbled circuits. We introduce LabelTensors that allow us to leverage the massive parallelity of modern GPUs. Combined with state-of-the-art garbling optimizations, Dash outperforms previous garbling approaches up to a factor of about 100. Furthermore, we introduce an efficient scaling operation over the residues of the Chinese remainder theorem representation to arithmetic garbled circuits, which allows us to garble larger networks and achieve much higher accuracy than previous approaches. Finally, Dash requires only a single communication round per inference step, regardless of the depth of the neural network, and a very small constant online communication volume.

cs.CR↗

arXiv2021

Util::Lookup: Exploiting key decoding in cryptographic libraries

Implementations of cryptographic libraries have been scrutinized for secret-dependent execution behavior exploitable by microarchitectural side-channel attacks. To prevent unintended leakages, most libraries moved to constant-time implementations of cryptographic primitives. There have also been efforts to certify libraries for use in sensitive areas, like Microsoft CNG and Botan, with specific attention to leakage behavior. In this work, we show that a common oversight in these libraries is the existence of \emph{utility functions}, which handle and thus possibly leak confidential information. We analyze the exploitability of base64 decoding functions across several widely used cryptographic libraries. Base64 decoding is used when loading keys stored in PEM format. We show that these functions by themselves leak sufficient information even if libraries are executed in trusted execution environments. In fact, we show that recent countermeasures to transient execution attacks such as LVI \emph{ease} the exploitability of the observed faint leakages, allowing us to robustly infer sufficient information about RSA private keys \emph{with a single trace}. We present a complete attack, including a broad library analysis, a high-resolution last level cache attack on SGX enclaves, and a fully parallelized implementation of the extend-and-prune approach that allows a complete key recovery at medium costs.

cs.CR↗

arXiv2021

Load Balancing: The Long Road from Theory to Practice

There is a long history of approximation schemes for the problem of scheduling jobs on identical machines to minimize the makespan. Such a scheme grants a $(1+ε)$-approximation solution for every $ε> 0$, but the running time grows exponentially in $1/ε$. For a long time, these schemes seemed like a purely theoretical concept. Even solving instances for moderate values of $ε$ seemed completely illusional. In an effort to bridge theory and practice, we refine recent ILP techniques to develop the fastest known approximation scheme for this problem. An implementation of this algorithm reaches values of $ε$ lower than $2/11\approx 18.2\%$ within a reasonable timespan. This is the approximation guarantee of MULTIFIT, which, to the best of our knowledge, has the best proven guarantee of any non-scheme algorithm.

cs.DS↗

arXiv2021

Robust Online Algorithms for Dynamic Choosing Problems

Semi-online algorithms that are allowed to perform a bounded amount of repacking achieve guaranteed good worst-case behaviour in a more realistic setting. Most of the previous works focused on minimization problems that aim to minimize some costs. In this work, we study maximization problems that aim to maximize their profit. We mostly focus on a class of problems that we call choosing problems, where a maximum profit subset of a set objects has to be maintained. Many known problems, such as Knapsack, MaximumIndependentSet and variations of these, are part of this class. We present a framework for choosing problems that allows us to transfer offline $α$-approximation algorithms into $(α-epsilon)$-competitive semi-online algorithms with amortized migration $O(1/ε)$. Moreover we complement these positive results with lower bounds that show that our results are tight in the sense that no amortized migration of $o(1/ε)$ is possible.

cs.DS↗

arXiv2020

Tightness of Sensitivity and Proximity Bounds for Integer Linear Programs

We consider ILPs, where each variable corresponds to an integral point within a polytope $\mathcal{P}$, i. e., ILPs of the form $\min\{c^{\top}x\mid \sum_{p\in\mathcal P\cap \mathbb Z^d} x_p p = b, x\in\mathbb Z^{|\mathcal P\cap \mathbb Z^d|}_{\ge 0}\}$. The distance between an optimal fractional solution and an optimal integral solution (called proximity) is an important measure. A classical result by Cook et al.~(Math. Program., 1986) shows that it is at most $Δ^{Θ(d)}$ where $Δ$ is the largest coefficient in the constraint matrix. Another important measure studies the change in an optimal solution if the right-hand side $b$ is replaced by another right-hand side $b'$. The distance between an optimal solution $x$ w.r.t.~$b$ and an optimal solution $x'$ w.r.t.~$b'$ (called sensitivity) is similarly bounded, i. e., $\lVert b-b' \rVert_{1}\cdot Δ^{Θ(d)}$, also shown by Cook et al. Even after more than thirty years, these bounds are essentially the best known bounds for these measures. While some lower bounds are known for these measures, they either only work for very small values of $Δ$, require negative entries in the constraint matrix, or have fractional right-hand sides. Hence, these lower bounds often do not correspond to instances from algorithmic problems. This work presents for each $Δ> 0$ and each $d > 0$ ILPs of the above type with non-negative constraint matrices such that their proximity and sensitivity is at least $Δ^{Θ(d)}$. Furthermore, these instances are closely related to instances of the Bin Packing problem as they form a subset of columns of the configuration ILP. We thereby show that the results of Cook et al. are indeed tight, even for instances arising naturally from problems in combinatorial optimization.

cs.CC↗

arXiv2020

Solving Packing Problems with Few Small Items Using Rainbow Matchings

An important area of combinatorial optimization is the study of packing and covering problems, such as Bin Packing, Multiple Knapsack, and Bin Covering. Those problems have been studied extensively from the viewpoint of approximation algorithms, but their parameterized complexity has only been investigated barely. For problem instances containing no "small" items, classical matching algorithms yield optimal solutions in polynomial time. In this paper we approach them by their distance from triviality, measuring the problem complexity by the number $k$ of small items. Our main results are fixed-parameter algorithms for vector versions of Bin Packing, Multiple Knapsack, and Bin Covering parameterized by $k$. The algorithms are randomized with one-sided error and run in time $4^{k} \cdot k! \cdot n^{O(1)}$. To achieve this, we introduce a colored matching problem to which we reduce all these packing problems. The colored matching problem is natural in itself and we expect it to be useful for other applications. We also present a deterministic fixed-parameter for Bin Packing with run time $(k!)^{2}\cdot k \cdot 2^{k}\cdot n\cdot \log(n)$.

cs.DS↗

arXiv2020

New Bounds for the Vertices of the Integer Hull

The vertices of the integer hull are the integral equivalent to the well-studied basic feasible solutions of linear programs. In this paper we give new bounds on the number of non-zero components -- their support -- of these vertices matching either the best known bounds or improving upon them. While the best known bounds make use of deep techniques, we only use basic results from probability theory to make use of the concentration of measure effect. To show the versatility of our techniques, we use our results to give the best known bounds on the number of such vertices and an algorithm to enumerate them. We also improve upon the known lower bounds to show that our results are nearly optimal. One of the main ingredients of our work is a generalization of the famous Hoeffding bound to vector-valued random variables that might be of general interest.

cs.DS↗

arXiv2019

Robust Online Algorithms for Dynamic Problems

Online algorithms that allow a small amount of migration or recourse have been intensively studied in the last years. They are essential in the design of competitive algorithms for dynamic problems, where objects can also depart from the instance. In this work, we give a general framework to obtain so called robust online algorithms for these dynamic problems: these online algorithm achieve an asymptotic competitive ratio of $γ+ε$ with migration $O(1/ε)$, where $γ$ is the best known offline asymptotic approximation ratio. In order to use our framework, one only needs to construct a suitable online algorithm for the static online case, where items never depart. To show the usefulness of our approach, we improve upon the best known robust algorithms for the dynamic versions of generalizations of Strip Packing and Bin Packing, including the first robust algorithm for general $d$-dimensional Bin Packing and Vector Packing.

cs.DS↗

arXiv2019

Positive-Instance Driven Dynamic Programming for Graph Searching

Research on the similarity of a graph to being a tree - called the treewidth of the graph - has seen an enormous rise within the last decade, but a practically fast algorithm for this task has been discovered only recently by Tamaki (ESA 2017). It is based on dynamic programming and makes use of the fact that the number of positive subinstances is typically substantially smaller than the number of all subinstances. Algorithms producing only such subinstances are called positive-instance driven (PID). We give an alternative and intuitive view on this algorithm from the perspective of the corresponding configuration graphs in certain two-player games. This allows us to develop PID-algorithms for a wide range of important graph parameters such as treewidth, pathwidth, and treedepth. We analyse the worst case behaviour of the approach on some well-known graph classes and perform an experimental evaluation on real world and random graphs.

cs.DS↗

arXiv2019

Online Bin Covering with Limited Migration

Semi-online models where decisions may be revoked in a limited way have been studied extensively in the last years. This is motivated by the fact that the pure online model is often too restrictive to model real-world applications, where some changes might be allowed. A well-studied measure of the amount of decisions that can be revoked is the migration factor $β$: When an object $o$ of size $s(o)$ arrives, the decisions for objects of total size at most $β\cdot s(o)$ may be revoked. Usually $β$ should be a constant. This means that a small object only leads to small changes. This measure has been successfully investigated for different, classic problems such as bin packing or makespan minimization. The dual of makespan minimization - the Santa Claus or machine covering problem - has also been studied, whereas the dual of bin packing - the bin covering problem - has not been looked at from such a perspective. In this work, we extensively study the bin covering problem with migration in different scenarios. We develop algorithms both for the static case - where only insertions are allowed - and for the dynamic case, where items may also depart. We also develop lower bounds for these scenarios both for amortized migration and for worst-case migration showing that our algorithms have nearly optimal migration factor and asymptotic competitive ratio (up to an arbitrary small $\eps$). We therefore resolve the competitiveness of the bin covering problem with migration.

cs.DS↗

arXiv2018

Practical Access to Dynamic Programming on Tree Decompositions

Parameterized complexity theory has lead to a wide range of algorithmic breakthroughs within the last decades, but the practicability of these methods for real-world problems is still not well understood. We investigate the practicability of one of the fundamental approaches of this field: dynamic programming on tree decompositions. Indisputably, this is a key technique in parameterized algorithms and modern algorithm design. Despite the enormous impact of this approach in theory, it still has very little influence on practical implementations. The reasons for this phenomenon are manifold. One of them is the simple fact that such an implementation requires a long chain of non-trivial tasks (as computing the decomposition, preparing it,...). We provide an easy way to implement such dynamic programs that only requires the definition of the update rules. With this interface, dynamic programs for various problems, such as 3-coloring, can be implemented easily in about 100 lines of structured Java code. The theoretical foundation of the success of dynamic programming on tree decompositions is well understood due to Courcelle's celebrated theorem, which states that every MSO-definable problem can be efficiently solved if a tree decomposition of small width is given. We seek to provide practical access to this theorem as well, by presenting a lightweight model-checker for a small fragment of MSO. This fragment is powerful enough to describe many natural problems, and our model-checker turns out to be very competitive against similar state-of-the-art tools.

cs.DS↗

arXiv2018

On the Gold Standard for Security of Universal Steganography

While symmetric-key steganography is quite well understood both in the information-theoretic and in the computational setting, many fundamental questions about its public-key counterpart resist persistent attempts to solve them. The computational model for public-key steganography was proposed by von Ahn and Hopper in EUROCRYPT 2004. At TCC 2005, Backes and Cachin gave the first universal public-key stegosystem - i.e. one that works on all channels - achieving security against replayable chosen-covertext attacks (SS-RCCA) and asked whether security against non-replayable chosen-covertext attacks (SS-CCA) is achievable. Later, Hopper (ICALP 2005) provided such a stegosystem for every efficiently sampleable channel, but did not achieve universality. He posed the question whether universality and SS-CCA-security can be achieved simultaneously. No progress on this question has been achieved since more than a decade. In our work we solve Hopper's problem in a somehow complete manner: As our main positive result we design an SS-CCA-secure stegosystem that works for every memoryless channel. On the other hand, we prove that this result is the best possible in the context of universal steganography. We provide a family of 0-memoryless channels - where the already sent documents have only marginal influence on the current distribution - and prove that no SS-CCA-secure steganography for this family exists in the standard non-look-ahead model.

cs.CR↗