@article{23095,
  abstract     = {The PXP chain was recently shown to exhibit superdiffusive energy transport with Kardar-Parisi-Zhang-like
scaling, z ≈ 3/2, joining a growing number of spin chains with this exponent. An understanding of how this
anomalous hydrodynamics emerges from microscopics is, however, still lacking. In this work, we show that
finite-temperature energy transport in this model provides a window into the emergence of superdiffusion. At
finite temperature, the energy autocorrelation function exhibits a crossover from short-time coherent dynamics to
long-time hydrodynamics. The short-time behavior is dominated by a single magnon band and can be understood
analytically. In momentum space, this regime is characterized by spectral weight near q = π. The damping time
τ , which separates the short-time magnon-dominated behavior from the late-time hydrodynamics, grows rapidly
upon cooling, consistent with an activated form τ (β) ∼ βeβ with a gap scale set by the magnon band. At longer
times, the spectral weight transfers to q = 0 and the running decay exponent drifts toward the superdiffusive
value z = 3/2. Finite-temperature energy transport therefore provides a bridge between microscopic magnon
physics and late-time superdiffusion in the PXP model.},
  author       = {Shengtao, Jiang and Desaules, Jean-Yves Marc and Ljubotina, Marko and Scaffidi, Thomas},
  issn         = {2469-9969},
  journal      = {Physical Review B},
  number       = {8},
  publisher    = {American Physical Society},
  title        = {{Finite-temperature crossover from coherent magnons to energy superdiffusion in the PXP model}},
  doi          = {10.1103/mx9k-1mc6},
  volume       = {114},
  year         = {2026},
}

@article{23115,
  abstract     = {We investigate the association between galaxies and neutral O I absorption systems at z ∼ 6, which trace metalenriched gas during the epoch of reionization. We identify 40 galaxies across six quasar fields, residing in 15
overdensities within 300 kpc of the background sight lines. Five O I absorption systems are associated with five of
these overdensities, yielding a covering fraction of
0.27 0.10
+0.13
within 300 kpc. The absorption occurs beyond typical
virial radii, indicating that the gas traces extended overdensity environments rather than individual galaxy halos,
unlike the z ∼ 0 circumgalactic medium (CGM), which is largely bound to halos. These galaxy-associated
absorbers account for ∼35% of all O I systems seen in blind quasar surveys, implying the remainder arise in
lower-mass galaxies below our detection threshold or in dense neutral intergalactic medium pockets. The CGM
around these galaxies contains ≳ 2 × 106 M⊙ of oxygen, comparable to the interstellar medium oxygen mass of
the galaxies themselves, suggesting that the surrounding environment holds as much metal mass as the galaxies.
All five galaxy-associated systems show significantly higher
log NCII NOI ( / )
ratios than absorbers lacking galaxy
associations. Furthermore, relative abundance ratios ([Si/O], [C/O]) reveal that four of the five exhibit
enrichment patterns consistent with Population III nucleosynthesis at the outskirts of galaxy overdensities. These
rare systems offer a unique window into the role of first-generation stars in shaping the early metal enrichment of
galaxies and their environments.},
  author       = {Higginson, Jack and Bordoloi, Rongmon and Simcoe, Robert A. and Matthee, Jorryt J and Kashino, Daichi and Mackenzie, Ruari and Kramarenko, Ivan and Lilly, Simon J. and Eilers, Anna-Christina and Naidu, Rohan P. and Yue, Minghao},
  issn         = {1538-4357},
  journal      = {The Astrophysical Journal},
  number       = {1},
  publisher    = {IOP Publishing},
  title        = {{EIGER. VIII. First stars signatures in the connection between O I absorption and galaxies in the epoch of reionization}},
  doi          = {10.3847/1538-4357/ae3e81},
  volume       = {999},
  year         = {2026},
}

@article{23117,
  abstract     = {We present a comprehensive analysis of the relationship between galaxies and the intergalactic medium (IGM) during the late stages of cosmic reionization, based on the complete JWST EIGER dataset. Using deep NIRCam 3.5 μm slitless spectroscopy, we construct a sample of 948 [O iii]λ5008-emitting galaxies with −21.4 ≲ MUV ≲ −17.2 spanning 5.33 < z < 6.97 along six quasar sight lines. We correlate these galaxies with Lyα and Lyβ transmission measured from high-resolution quasar spectra across multiple redshift intervals. We find clear redshift evolution in the correlation between galaxy density and transmission: it is suppressed in overdense regions at z < 5.50, while enhanced at 5.70 < z < 6.15. The intermediate range exhibits a transitional behavior. Cross-correlation measurements further reveal excess absorption within ∼8 cMpc of galaxies at low redshifts, and enhanced transmission at intermediate scales (∼5–20 cMpc) at z > 5.70. Statistical tests using mock catalogs with realistic galaxy clustering but no correlation with the transmission field confirm that the observed correlations are unlikely to arise by chance. The evolving signals can be explained by stronger absorption in overdense regions, combined with the competing influences of local radiation fields and the rising background radiation. While local radiation dominates ionization of the surrounding IGM at earlier times, the background becomes increasingly important, eventually surpassing the impact of nearby galaxies. These results support an inside-out progression of reionization, with ionized regions originating around clustered, star-forming galaxies and gradually extending into underdense regions.},
  author       = {Kashino, Daichi and Lilly, Simon J. and Matthee, Jorryt J and Mackenzie, Ruari and Eilers, Anna-Christina and Bordoloi, Rongmon and Simcoe, Robert A. and Naidu, Rohan P. and Yue, Minghao and Liu, Bin},
  issn         = {1538-4357},
  journal      = {The Astrophysical Journal},
  number       = {2},
  publisher    = {IOP Publishing},
  title        = {{EIGER. VII. The evolving relationship between galaxies and the intergalactic medium in the final stages of reionization}},
  doi          = {10.3847/1538-4357/ae2799},
  volume       = {997},
  year         = {2026},
}

@article{23111,
  abstract     = {In single-molecule electronics, 4,4′-bipyridine is a well-characterized molecule that serves both as a fundamental testbed for developing experimental methodologies and showcases the complex behaviors expected from molecular electronic components. In break junction experiments, it is generally understood to exhibit a high- and a low-conductance state that are attributed to a tilted and stretched single-molecule configuration. Despite this established view, we suggest a supramolecular origin for these conductance states that has not previously been considered. Our findings indicate that the high- and low-conductance states may arise from the presence of two molecules and one molecule, respectively, challenging the conventional interpretation. Using a state-of-the-art machine learning force field called the neuroevolution potential, we find that the conventional interpretation of a tilted and stretched configuration is inconsistent with our simulations. Instead, our results suggest that the presence of two molecules promotes junction geometries and gold rearrangements that stabilize a high-conductance state whose transmission exceeds twice that of the low-conductance state. Furthermore, we compare this interpretation to the existing literature of diverse experimental data on 4,4′-bipyridine in break junctions and find that it is consistent with the broader body of experimental evidence. Our results highlight the structural complexity of molecular behavior within break junctions, suggesting even more complex dynamics than originally anticipated. In this light, a broader examination of the dynamics of molecules in single-molecule break junctions with machine learning-assisted molecular dynamics is warranted.},
  author       = {Bro-Jørgensen, William and Lin, Junfeng and Hamill, Joseph M. and Solomon, Gemma C.},
  issn         = {2694-2496},
  journal      = {ACS Nanoscience Au},
  number       = {4},
  pages        = {662--673},
  publisher    = {American Chemical Society},
  title        = {{Machine learning-enhanced molecular dynamics of 4,4′-bipyridinein a break junction: A supramolecular origin of the high conductance}},
  doi          = {10.1021/acsnanoscienceau.6c00052},
  volume       = {6},
  year         = {2026},
}

@article{23116,
  abstract     = {The Legacy Survey of Space and Time (LSST) is expected to observe up to ∼100 million quasars in the next decade. In this work, we show that it is possible to use such data to measure the characteristic frequency evolution of a “chirp” induced by gravitational waves, which can serve as robust evidence for the presence of a compact supermassive black hole binary. Following the LSST specifications, we generate mock lightcurves consisting of (i) a post-Newtonian chirp produced by orbital motion through, e.g., relativistic Doppler boosting, (ii) a damped random walk representing intrinsic quasar variability, and (iii) Gaussian photometric errors, while assuming nonuniform observations with extended gaps over a period of 10 yr. Through a fully Bayesian analysis, we show that we can simultaneously measure the chirp and noise parameters with little degeneracy between the two. For chirp signals with an amplitude of A = 0.5 mag and a range of times to merger (tm = 15–104 yr), we can typically measure a nonzero amplitude and positive frequency derivative with over 5σ credibility. For binaries with tm = 50 yr, we achieve 3σ (5σ) confidence that the signal is chirping for A ≳ 0.1 (A  >  0.2). The median runtime of our analysis is 5.6 minutes, with a minimum as low as 35 s, making it scalable to a large number of lightcurves. This implies that LSST could, on its own, establish the presence of a compact supermassive black hole binary, and thus discover gravitational wave sources detectable by LISA and by Pulsar Timing Arrays.},
  author       = {Xin, Chengcheng and Isi, Maximiliano and Farr, Will M. and Haiman, Zoltán},
  issn         = {1538-4357},
  journal      = {The Astrophysical Journal},
  number       = {1},
  publisher    = {IOP Publishing},
  title        = {{Identifying compact chirping SMBHBs in LSST using Bayesian analysis}},
  doi          = {10.3847/1538-4357/ae40b3},
  volume       = {999},
  year         = {2026},
}

@article{23096,
  abstract     = {The first generation of stars, known as population III (Pop III), formed from primordial gas consisting solely of hydrogen and helium and is believed to have emerged only a few hundred million years after the Big Bang. Detecting the chemical enrichment of metal-poor circumgalactic gas offers a promising way to trace the enrichment signature of Pop III stars. Along the sight line to the quasar SDSS J0100+2802, a metal absorber at z = 5.945, showing over-abundant carbon and silicon compared with solar, has been reported to be consistent with the enrichment pattern of Pop III stars. Here, using data recorded by the James Webb Space Telescope, we report the discovery of an unusually metal-poor galaxy overdensity of 17 members (mean metallicity ~3% of solar) near this metal absorber, which is ~0.4 dex more metal-poor than coeval galaxies in similarly overdense environments. This less chemically evolved system may have provided favourable conditions for preserving the absorption signatures of Pop III enrichment. We infer a minimum dark matter halo mass of log(Mh,min/M⊙) = 10.68+0.93 −1.72, supporting late-time Pop III formation at the outskirts of atomic hydrogen cooling haloes. Our findings open a promising observational pathway to identifying the chemical imprints of the first stars and constrain the conditions for their formation.},
  author       = {Li, Zihao and Kakiichi, Koki and Christensen, Lise and Cai, Zheng and D’Odorico, Valentina and Matthee, Jorryt J and Kashino, Daichi and Bordoloi, Rongmon and Mackenzie, Ruari and Berg, Trystyn A.M. and Vanni, Irene and Salvadori, Stefania and Venditti, Alessandra and Zhang, Shiwu and Bosman, Sarah E.I. and Bañados, Eduardo and Davies, Frederick B. and Fan, Xiaohui and Jun, Hyunsung D. and Jin, Xiangyu and Li, Mingyu and Rojas-Ruiz, Sofía and Wang, Feige and Yang, Jinyi and Zou, Siwei and Zhang, Huanian and Zhu, Yongda},
  issn         = {2397-3366},
  journal      = {Nature Astronomy},
  publisher    = {Springer Nature},
  title        = {{First-star imprints in a metal-poor galaxy overdensity near the end of reionization}},
  doi          = {10.1038/s41550-026-02993-5},
  year         = {2026},
}

@article{23112,
  abstract     = {We study the joint spectral properties of two coupled random matrices H(1) and H(2), which are either real symmetric or complex Hermitian. The entries of these matrices exhibit polynomially decaying correlations, both within each matrix and between them. Surprisingly, we find that under extremely weak decorrelation condition, permitting H(1) and H(2) to be almost fully correlated, the fluctuations of their individual eigenvalues in the bulk of the spectrum are still asymptotically independent. Furthermore, we demonstrate that this decorrelation condition is optimal.},
  author       = {Kolupaiev, Oleksii},
  issn         = {1350-7265},
  journal      = {Bernoulli},
  keywords     = {correlated random matrix, eigenvalue perturbation theory, Local law, Wigner-type matrix},
  number       = {4},
  pages        = {2999--3023},
  publisher    = {Bernoulli Society for Mathematical Statistics and Probability},
  title        = {{Spectral independence of almost fully correlated random matrices}},
  doi          = {10.3150/26-bej1976},
  volume       = {32},
  year         = {2026},
}

@article{23060,
  author       = {Ziletti, Chiara},
  isbn         = {9783702512187},
  issn         = {1012-0327},
  journal      = {Scrinium},
  location     = {Barcelona, Spain},
  pages        = {181--186},
  publisher    = {Anton Pustet Verlag},
  title        = {{International Council on Archives (ICA) Barcelona Congress 2025, 27-30 October 2025}},
  volume       = {80},
  year         = {2026},
}

@article{23114,
  author       = {Pieber, Bartholomäus and Nijs, Anne},
  issn         = {1099-0690},
  journal      = {European Journal of Organic Chemistry},
  publisher    = {Wiley},
  title        = {{The EuChemS division of organic chemistry and EurJOC - partners in advancing organic chemistry}},
  doi          = {10.1002/ejoc.70847},
  year         = {2026},
}

@article{23123,
  abstract     = {How many points can be placed in an 𝑛 × 𝑛 grid sothat every (affine) line contains at most 𝑘 points? Weprove that for 𝑛 ⩾ 𝑘 ⩾ 1037 the maximum number ofpoints is exactly 𝑘𝑛. Our proof builds on the recent workof Kovács, Nagy and Szabó (who proved an analogousresult when 𝑘 is at least about √𝑛 log 𝑛), incorporat-ing ideas of Jain and Pham. Using the same approach,we also obtain new bounds for higher-dimensionalextensions of this problem},
  author       = {Grebennikov, Aleksandr and Kwan, Matthew Alan},
  issn         = {1469-2120},
  journal      = {Bulletin of the London Mathematical Society},
  number       = {10},
  publisher    = {Wiley},
  title        = {{No‐(k+1)‐in‐line problem for large constant k}},
  doi          = {10.1112/blms.70510},
  volume       = {58},
  year         = {2026},
}

@article{23121,
  abstract     = {A widespread observation in molecular evolution is that X-linked genes evolve faster—and often adapt faster—than autosomal genes. Three main ideas have been put forward to explain “faster-X” adaptation. The first is that new beneficial mutations are typically partially recessive with respect to fitness, leading to more efficient selection in hemizygous males. The second is that sex differences in recombination, or sex differences in variance in reproductive success, enhance the effective population size on the X chromosome relative to autosomes. Both ideas have limitations: the first conflicts with theories of dominance, in which beneficial mutations are proposed to be partially dominant, while the second only applies to some taxa. Here we tested a third, newly-proposed theory, in which mutations with large “scaled phenotypic effects” (i.e., large effects relative to the distance to a phenotypic optimum) experience more positive selection on the X. Specifically, we used three proxies for scaled phenotypic effects—amino-acid dissimilarity, amino-acid conservation, and gene age—and estimated X and autosomal rates of adaptation of nonsynonymous mutations across effect-size classes. We did this in three lineages with a well-documented faster-X: Drosophila melanogaster, Mus musculus, and Homo sapiens. As expected, we found that large-effect mutations were more likely to be under purifying selection and less likely to be under positive selection than small-effect mutations. However, contrary to the new theory, faster-X adaptation was not enriched among large-effect mutations. Overall, our results highlight an unresolved gap between patterns of sex chromosome evolution and theories of dominance for fitness.},
  author       = {Ruzicka, Filip and Vicoso, Beatriz},
  issn         = {1759-6653},
  journal      = {Genome Biology and Evolution},
  keywords     = {evolutionary biology, population genetics, sex chromosomes, adaptation, genetic divergence, McDonald-Kreitman test, Fisher’s geometric model},
  publisher    = {Oxford University Press},
  title        = {{Is faster-X adaptation due to large-effect mutations? An empirical test of a new theory}},
  doi          = {10.1093/gbe/evag259},
  year         = {2026},
}

@article{23124,
  abstract     = {Among the set of hypersurfaces of degree d and dimension  defined by the vanishing
of a homogeneous polynomial with coefficients in the set {−1, +1}, we investigate the
probability that a hypersurface contains a rational point as d and  tend to ∞.},
  author       = {Browning, Timothy D and Sawin, Will},
  issn         = {1464-3847},
  journal      = {The Quarterly Journal of Mathematics},
  publisher    = {Oxford University Press},
  title        = {{Random diophantine equations of large degree}},
  doi          = {10.1093/qmath/haag037},
  year         = {2026},
}

@phdthesis{22984,
  author       = {Teplova, Anastasiia},
  isbn         = {978-3-99078-093-0},
  issn         = {2663-337X },
  pages        = {190},
  publisher    = {Institute of Science and Technology Austria},
  title        = {{Chasing the elusive: Investigating nucleotide cyclase activity in gibberellin and strigolactone signalling; Cellular mechanisms of gravity sensing and response in roots}},
  doi          = {10.15479/AT-ISTA-22984},
  year         = {2026},
}

@phdthesis{23079,
  abstract     = {We study the computational aspects of the two central solution concepts of stochastic games
played on finite graphs for an infinite number of steps: the value in two-player zero-sum
games, and Nash equilibria in multi-player games. In both settings, exact solutions are in
tractable. Fortwo-playerzero-sumconcurrentstochasticgames,valuescanbeirrational,and
the associated decision problem is SQRT-SUM hard. For multi-player turn-based stochastic
games,theequilibriumprobabilitiescanbeirrational,anddecidingtheconstrainedexistence
of a Nash equilibrium is undecidable in the presence of memory, and ∃R-complete when the
players are restricted to stationary strategies. Hence, in both settings, the natural computational problem is the approximation of the solution within an arbitrary additive error. In
this thesis, we present improved complexity results and algorithms for such approximation
problems.
The first part of the thesis considers two-player zero-sum concurrent stochastic games
with two classic objectives: (a) stateful-discounted objectives, which are similar to the classic
discounted-sum objectives, but states are associated with different discount factors rather
than a single discount factor; and (b) parity objectives, which are a canonical representation
for ω-regular objectives. For stateful-discounted objectives, given an ordering of the discount
factors, the limit value is the limit of the value of the stateful-discounted objectives, as
the discount factors approach zero according to the given order. The value-approximation
problem is known to be in EXPSPACE for the limit value of stateful-discounted objectives and
in PSPACEforparityobjectives,andthebest-knownalgorithmsforbothproblemsareatleast
exponential time, with an exponential dependence on the number of states and actions. Our
main results for the value-approximation problem for the limit value of stateful-discounted
objectives and parity objectives areas follows: (a)we establish TFNPNP complexity; and (b)we
present algorithms that improve the dependency on the number of actions in the exponent
from linear to logarithmic. In particular, if the number of states is constant, our algorithms
run in polynomial time.
The second part of the thesis considers multi-player turn-based stochastic games with
terminal payoffs, where the players are restricted to stationary strategies. We study the
approximate constrained existence problem of Nash equilibria: given a game that has a Nash
equilibrium in which the payoff of every player lies in a given interval, compute a strategy
profile that is an ε-Nash equilibrium and that satisfies the same constraints up to an additive
error ε. Our main results are as follows: (a) we present an FNPNP algorithm for the problem;
and(b)we establish that its decision version is NP-hard, already with two players. For the
upperbound,weshowthatifaconstrainedNashequilibriumexists,thenthereexists one
in which the non-zero probabilities are at least an inverse of a double-exponential in the
i
input size, and that such a strategy profile can be represented, up to an additive error ε, with
floating-point numbers of polynomial size. We complement the upper bound with a lower
bound on this approach: we present a game with only five players in which every constrained
Nash equilibrium requires double-exponentially small probabilities.
The two parts share a common technical toolkit: bounds on solutions of systems of
polynomial inequalities that arise from real algebraic geometry, the floating-point representation of doubly-exponentially small numbers, and guess-and-verify procedures that resolve
single-player games with an NP oracle.},
  author       = {Asadi, Ali},
  issn         = {2791-4585},
  pages        = {83},
  publisher    = {Institute of Science and Technology Austria},
  title        = {{Approximation of values and equilibria in stochastic games: Complexity and algorithms}},
  doi          = {10.15479/AT-ISTA-23079},
  year         = {2026},
}

@phdthesis{22626,
  author       = {Fiedler, Christine},
  isbn         = {978-3-99078-086-2},
  issn         = {2663-337X},
  pages        = {141},
  publisher    = {Institute of Science and Technology Austria},
  title        = {{Mechanistic insight into solution-processed p-type tin chalcogenides as a basis for designing their n-type analogs}},
  doi          = {10.15479/AT-ISTA-22626},
  year         = {2026},
}

@phdthesis{22017,
  author       = {Kleinhanns, Tobias},
  isbn         = {978-3-99078-081-7},
  issn         = {2663-337X},
  pages        = {59},
  publisher    = {Institute of Science and Technology Austria},
  title        = {{Unraveling the origin and evolution of defects to enable advanced thermoelectric performance}},
  doi          = {10.15479/AT-ISTA-22017},
  year         = {2026},
}

@phdthesis{22664,
  abstract     = {The widespread adoption of apps like Whatsapp and Signal has translated into billions of people all around the world communicating on a regular basis by making use of services that offer end-to-end encryption and even provide security guarantees when a user's device is compromised.

This was made possible by the introduction of the Double Ratchet Algorithm~\cite{double_ratchet}, which was designed for a setting where communication takes place between two parties.
However, in practice, many apps offer the possibility of creating groups.
The protocols they use to secure communication are inefficient for large group which has the undesireable consequence that the aforementioned apps have established limits on the group size of roughly 1000 users.
This has motivated the introduction of the Messaging Layer Security (MLS) standard~\cite{rfc9420} by the IETF which is based on a primitive called Continuous Group Key Agreement (CGKA)~\cite{C:ACDT20}.

This primitive allows a group of users to maintain a shared secret key that is frequently rotated by the group members in order to change group membership, achieve forward secrecy (FS) and post compromise security (PCS).
Most protocols are based on binary trees where the nodes are associated to a pair formed by public key and a secret key.
Each leaf corresponds to one of the group members and a user knows the secret keys associated to nodes along the path from their leaf to the root.
When a user wants to update their key material they have to change $ \log(N) $ many keys.
This requires uploading $ \log(N) $ many ciphertexts to communicate the new keys to the rest of the group members in a way that respects the tree structure.

In this thesis we study how much communication between group members is required in order to add and remove users from a group as well as in order to provide PCS when we consider CGKAs built using standard cryptographic primitives like pseudo-random functions and public-key encryption. Furthermore, we also consider the case of MLS and provide the first lower bound showing that its communication complexity is much worse than previously believed, i.e., it is very far from $ \log(N) $.
Finally, we also propose a variant of MLS which provably achieves the same security properties with a much lower communication cost.},
  author       = {Cueto Noval, Miguel},
  isbn         = { 978-3-99078-087-9},
  issn         = {2663-337X},
  pages        = {187},
  publisher    = {Institute of Science and Technology Austria},
  title        = {{Towards efficient secure group messaging}},
  doi          = {10.15479/AT-ISTA-22664},
  year         = {2026},
}

@phdthesis{22857,
  abstract     = {Artificial intelligence and machine learning have undergone an unprecedented evolution in the past decade, motivating a research effort toward a theory able to capture the qualitative behavior of large-scale neural systems. A central puzzle has been the clear benefit of scaling architecture size and overfitting the training set in supervised learning tasks. This evidence, in apparent contradiction with classical statistical learning theory, pushed researchers to develop a new theory capturing the interplay between the algorithmic and architectural bias of training and the specific target function, differently from previous methods rooted in uniform stability.
This approach has enabled a grounded understanding of novel learning regimes, typically through formal limits where the number of training samples $n$, data dimensions $d$, and model parameters $p$ grow to infinity at different rates. \\
In this thesis, we follow this approach, focusing on the trustworthiness of high-dimensional models: properties that are difficult to control during training or deployment and often emerge under unpredictable or adversarial conditions. In such settings, it is crucial to formally ensure a priori the reliability of machine learning systems.
First, we study data memorization, both as label fitting and as the storage of private information about training samples in trained parameters. We prove that $p = \Omega(n)$ parameters are sufficient for a deep neural network to memorize a generic set of labels, and for a model to memorize spurious features across training data. We then give evidence that $p = \Omega(dn)$ parameters are instead necessary for an adversary to reconstruct the full training set from the trained parameters.
Second, we study robustness, both to adversarial perturbations and to distribution shift. We first prove that $p = \Omega(dn)$ parameters can be sufficient for a class of neural networks to overfit the training data while guaranteeing robustness to adversarial perturbations. Then, we focus on spurious correlations learning in high-dimensional regression, studying the effect of the ridge regularization parameter in the proportional regime $n = \Theta(d)$, and connecting it via an equivalence argument to the role of over-parameterization $p = \Omega(n)$ in neural networks. We also investigate the architectural bias of attention-based networks, showing that they are sensitive to the replacement of individual words in an embedded sentence, allowing them to generalize on sentences where the contextual meaning depends on one or few words.
Finally, we study differentially private optimization in high-dimensional regimes. We prove that standard private gradient methods do not suffer in the over-parameterized regime $p = \Omega(n)$, challenging the current wisdom based on stability-derived generalization bounds. We then consider linear regression in the proportional regime $n = \Theta(d)$, showing that standard private gradient descent can achieve optimal rates under appropriate hyper-parameter scaling, such as sufficiently small gradient clipping constants, whose role is still debated in practice.},
  author       = {Bombari, Simone},
  isbn         = {978-3-99078-091-6},
  issn         = {2663-337X},
  keywords     = {machine learning, high-dimensional statistics, deep learning theory, privacy, memorization, robustness},
  pages        = {446},
  publisher    = {Institute of Science and Technology Austria},
  title        = {{Trustworthy machine learning in high dimensions}},
  doi          = {10.15479/AT-ISTA-22857},
  year         = {2026},
}

@article{23043,
  abstract     = {Mendelian randomization, the standard instrument variable causal-inference approach for genomics, can be severely biased by inadequate selection of appropriate instrument variables and horizontal pleiotropy. We introduce Causal Inference GWAS, a graphical approach that selects genetic instruments from summary statistics controlling for linkage and pleiotropy, accommodates rare variants and binary outcomes, distinguishes direct from indirect risk factors in high dimensions, and flags latent confounding. Applied to 9 risk factors, 4 metabolic disease outcomes, and 8.4M variants across 458,747 UK Biobank individuals, our method runs in 20 minutes, identifying only 696 genome-wide valid instruments. We replicate nearly all previously reported risk factor-outcome paths, but find that most are indistinguishable from unmeasured confounding. This suggests robust biobank causal inference will require longitudinal and family data to resolve temporal precedence from reverse causation. Our approach offers a principled first step toward screening genuine causal signal from phenotypic correlation in biobank data.},
  author       = {Machnik, Nick N and Mahmoudi, Seyed Mahdi and Borczyk, Malgorzata and Krätschmer, Ilse and Bauer, Markus J. and Robinson, Matthew Richard},
  issn         = {2041-1723},
  journal      = {Nature Communications},
  publisher    = {Springer Nature},
  title        = {{Causal inference for multiple risk factors and diseases from genomics data}},
  doi          = {10.1038/s41467-026-76877-7},
  volume       = {17},
  year         = {2026},
}

@article{23152,
  abstract     = {We present a panchromatic optical-millimeter characterization of GNz7q, a recently identified X-ray-weak,
rapidly growing red quasar embedded within a dusty starburst galaxy at z = 7.1899, using the full suite of JWST/
NIRCam, NIRSpec, Mid-Infrared Instrument (MIRI), and archival NOEMA observations. Our deep NIRSpec/
G395M spectroscopy reveals unambiguous broad Balmer emission (FWHM = 2221 ± 20 km s
−1
), confirming a
super-Eddington accreting black hole (BH) (λEdd = 2.7 ± 0.4) with a mass of
log(MBH/M ) = 7.55 ± 0.34,
using accretion-rate-corrected BH mass estimators. After subtracting the point source, we robustly detect stellar
emission from the host galaxy across multiple NIRCam and MIRI filters. Our joint morphological-spectral
analysis yields a stellar mass of = ± * log(M /M ) 10.5 0.4
and an intense star formation rate of
SFR = 330 ± 97 M⊙ yr−1
, confirming the host as a massive, dusty starburst galaxy. We find that GNz7q lies
on the local MBH–M* relation (MBH/M* ≃ 0.001) and is well positioned to evolve into the locus of massive Sloan
Digital Sky Survey (SDSS) quasars with
log(MBH/M ) 9
and M* ≈ 1011 M⊙ at z ∼ 6, owing to its remarkably
rapid growth in both the BH and its host galaxy. This stands in stark contrast to many recently reported JWST
active galactic nucleus populations at similar redshifts, including the little red dots, whose weak or undetected star
formation makes it difficult for them to grow into the massive galaxies hosting SDSS-like quasars. These results
suggest that GNz7q marks a rare, pivotal phase of early BH–galaxy coevolution, plausibly providing a crucial
direct pathway to the supermassive BH systems within the first billion years of the Universe.},
  author       = {Fei, Qinyue and Fujimoto, Seiji and Brammer, Gabriel and Li, Ruancun and Ho, Luis C. and Bromm, Volker and Álvarez-Márquez, Javier and Asada, Yoshihisa and Barro, Guillermo and Colina, Luis and Dayal, Pratika and Finkelstein, Steven L. and Fynbo, Johan P.U. and Ginolfi, Michele and Inayoshi, Kohei and Kokorev, Vasily and Leung, Gene C.K. and Matthee, Jorryt J and Meyer, Romain A. and Naidu, Rohan P. and Onoue, Masafusa and Pérez-González, Pablo G. and Steinhardt, Charles L. and Valentino, Francesco and Walter, Fabian and Xiao, Mengyuan and Zhang, Haowen},
  issn         = {1538-4357},
  journal      = {The Astrophysical Journal},
  number       = {1},
  publisher    = {IOP Publishing},
  title        = {{Direct pathway to the early supermassive Black Holes: A red super-Eddington quasar in a massive starburst host at z = 7.2}},
  doi          = {10.3847/1538-4357/aea07b},
  volume       = {1010},
  year         = {2026},
}