@article{12706, abstract = {Allometric settings of population dynamics models are appealing due to their parsimonious nature and broad utility when studying system level effects. Here, we parameterise the size-scaled Rosenzweig-MacArthur differential equations to eliminate prey-mass dependency, facilitating an in depth analytic study of the equations which incorporates scaling parameters’ contributions to coexistence. We define the functional response term to match empirical findings, and examine situations where metabolic theory derivations and observation diverge. The dynamical properties of the Rosenzweig-MacArthur system, encompassing the distribution of size-abundance equilibria, the scaling of period and amplitude of population cycling, and relationships between predator and prey abundances, are consistent with empirical observation. Our parameterisation is an accurate minimal model across 15+ orders of mass magnitude.}, author = {Mckerral, Jody C. and Kleshnina, Maria and Ejov, Vladimir and Bartle, Louise and Mitchell, James G. and Filar, Jerzy A.}, issn = {1932-6203}, journal = {PLoS One}, number = {2}, pages = {e0279838}, publisher = {Public Library of Science}, title = {{Empirical parameterisation and dynamical analysis of the allometric Rosenzweig-MacArthur equations}}, doi = {10.1371/journal.pone.0279838}, volume = {18}, year = {2023}, } @article{13202, abstract = {Phosphatidylinositol-4,5-bisphosphate (PI(4,5)P2) plays an essential role in neuronal activities through interaction with various proteins involved in signaling at membranes. However, the distribution pattern of PI(4,5)P2 and the association with these proteins on the neuronal cell membranes remain elusive. In this study, we established a method for visualizing PI(4,5)P2 by SDS-digested freeze-fracture replica labeling (SDS-FRL) to investigate the quantitative nanoscale distribution of PI(4,5)P2 in cryo-fixed brain. We demonstrate that PI(4,5)P2 forms tiny clusters with a mean size of ∼1000 nm2 rather than randomly distributed in cerebellar neuronal membranes in male C57BL/6J mice. These clusters show preferential accumulation in specific membrane compartments of different cell types, in particular, in Purkinje cell (PC) spines and granule cell (GC) presynaptic active zones. Furthermore, we revealed extensive association of PI(4,5)P2 with CaV2.1 and GIRK3 across different membrane compartments, whereas its association with mGluR1α was compartment specific. These results suggest that our SDS-FRL method provides valuable insights into the physiological functions of PI(4,5)P2 in neurons.}, author = {Eguchi, Kohgaku and Le Monnier, Elodie and Shigemoto, Ryuichi}, issn = {1529-2401}, journal = {The Journal of Neuroscience}, number = {23}, pages = {4197--4216}, publisher = {Society for Neuroscience}, title = {{Nanoscale phosphoinositide distribution on cell membranes of mouse cerebellar neurons}}, doi = {10.1523/JNEUROSCI.1514-22.2023}, volume = {43}, year = {2023}, } @article{12916, abstract = {We apply a variant of the square-sieve to produce an upper bound for the number of rational points of bounded height on a family of surfaces that admit a fibration over P1 whose general fibre is a hyperelliptic curve. The implied constant does not depend on the coefficients of the polynomial defining the surface. }, author = {Bonolis, Dante and Browning, Timothy D}, issn = {2036-2145}, journal = {Annali della Scuola Normale Superiore di Pisa - Classe di Scienze}, number = {1}, pages = {173--204}, publisher = {Scuola Normale Superiore - Edizioni della Normale}, title = {{Uniform bounds for rational points on hyperelliptic fibrations}}, doi = {10.2422/2036-2145.202010_018}, volume = {24}, year = {2023}, } @phdthesis{14422, abstract = {Animals exhibit a remarkable ability to learn and remember new behaviors, skills, and associations throughout their lifetime. These capabilities are made possible thanks to a variety of changes in the brain throughout adulthood, regrouped under the term "plasticity". Some cells in the brain —neurons— and specifically changes in the connections between neurons, the synapses, were shown to be crucial for the formation, selection, and consolidation of memories from past experiences. These ongoing changes of synapses across time are called synaptic plasticity. Understanding how a myriad of biochemical processes operating at individual synapses can somehow work in concert to give rise to meaningful changes in behavior is a fascinating problem and an active area of research. However, the experimental search for the precise plasticity mechanisms at play in the brain is daunting, as it is difficult to control and observe synapses during learning. Theoretical approaches have thus been the default method to probe the plasticity-behavior connection. Such studies attempt to extract unifying principles across synapses and model all observed synaptic changes using plasticity rules: equations that govern the evolution of synaptic strengths across time in neuronal network models. These rules can use many relevant quantities to determine the magnitude of synaptic changes, such as the precise timings of pre- and postsynaptic action potentials, the recent neuronal activity levels, the state of neighboring synapses, etc. However, analytical studies rely heavily on human intuition and are forced to make simplifying assumptions about plasticity rules. In this thesis, we aim to assist and augment human intuition in this search for plasticity rules. We explore whether a numerical approach could automatically discover the plasticity rules that elicit desired behaviors in large networks of interconnected neurons. This approach is dubbed meta-learning synaptic plasticity: learning plasticity rules which themselves will make neuronal networks learn how to solve a desired task. We first write all the potential plasticity mechanisms to consider using a single expression with adjustable parameters. We then optimize these plasticity parameters using evolutionary strategies or Bayesian inference on tasks known to involve synaptic plasticity, such as familiarity detection and network stabilization. We show that these automated approaches are powerful tools, able to complement established analytical methods. By comprehensively screening plasticity rules at all synapse types in realistic, spiking neuronal network models, we discover entire sets of degenerate plausible plasticity rules that reliably elicit memory-related behaviors. Our approaches allow for more robust experimental predictions, by abstracting out the idiosyncrasies of individual plasticity rules, and provide fresh insights on synaptic plasticity in spiking network models. }, author = {Confavreux, Basile J}, issn = {2663 - 337X}, pages = {148}, publisher = {Institute of Science and Technology Austria}, title = {{Synapseek: Meta-learning synaptic plasticity rules}}, doi = {10.15479/at:ista:14422}, year = {2023}, } @phdthesis{14374, abstract = {Superconductivity has many important applications ranging from levitating trains over qubits to MRI scanners. The phenomenon is successfully modeled by Bardeen-Cooper-Schrieffer (BCS) theory. From a mathematical perspective, BCS theory has been studied extensively for systems without boundary. However, little is known in the presence of boundaries. With the help of numerical methods physicists observed that the critical temperature may increase in the presence of a boundary. The goal of this thesis is to understand the influence of boundaries on the critical temperature in BCS theory and to give a first rigorous justification of these observations. On the way, we also study two-body Schrödinger operators on domains with boundaries and prove additional results for superconductors without boundary. BCS theory is based on a non-linear functional, where the minimizer indicates whether the system is superconducting or in the normal, non-superconducting state. By considering the Hessian of the BCS functional at the normal state, one can analyze whether the normal state is possibly a minimum of the BCS functional and estimate the critical temperature. The Hessian turns out to be a linear operator resembling a Schrödinger operator for two interacting particles, but with more complicated kinetic energy. As a first step, we study the two-body Schrödinger operator in the presence of boundaries. For Neumann boundary conditions, we prove that the addition of a boundary can create new eigenvalues, which correspond to the two particles forming a bound state close to the boundary. Second, we need to understand superconductivity in the translation invariant setting. While in three dimensions this has been extensively studied, there is no mathematical literature for the one and two dimensional cases. In dimensions one and two, we compute the weak coupling asymptotics of the critical temperature and the energy gap in the translation invariant setting. We also prove that their ratio is independent of the microscopic details of the model in the weak coupling limit; this property is referred to as universality. In the third part, we study the critical temperature of superconductors in the presence of boundaries. We start by considering the one-dimensional case of a half-line with contact interaction. Then, we generalize the results to generic interactions and half-spaces in one, two and three dimensions. Finally, we compare the critical temperature of a quarter space in two dimensions to the critical temperatures of a half-space and of the full space.}, author = {Roos, Barbara}, issn = {2663 - 337X}, pages = {206}, publisher = {Institute of Science and Technology Austria}, title = {{Boundary superconductivity in BCS theory}}, doi = {10.15479/at:ista:14374}, year = {2023}, } @article{13207, abstract = {We consider the linear BCS equation, determining the BCS critical temperature, in the presence of a boundary, where Dirichlet boundary conditions are imposed. In the one-dimensional case with point interactions, we prove that the critical temperature is strictly larger than the bulk value, at least at weak coupling. In particular, the Cooper-pair wave function localizes near the boundary, an effect that cannot be modeled by effective Neumann boundary conditions on the order parameter as often imposed in Ginzburg–Landau theory. We also show that the relative shift in critical temperature vanishes if the coupling constant either goes to zero or to infinity.}, author = {Hainzl, Christian and Roos, Barbara and Seiringer, Robert}, issn = {1664-0403}, journal = {Journal of Spectral Theory}, number = {4}, pages = {1507–1540}, publisher = {EMS Press}, title = {{Boundary superconductivity in the BCS model}}, doi = {10.4171/JST/439}, volume = {12}, year = {2023}, } @article{14452, abstract = {The classical infinitesimal model is a simple and robust model for the inheritance of quantitative traits. In this model, a quantitative trait is expressed as the sum of a genetic and an environmental component, and the genetic component of offspring traits within a family follows a normal distribution around the average of the parents’ trait values, and has a variance that is independent of the parental traits. In previous work, we showed that when trait values are determined by the sum of a large number of additive Mendelian factors, each of small effect, one can justify the infinitesimal model as a limit of Mendelian inheritance. In this paper, we show that this result extends to include dominance. We define the model in terms of classical quantities of quantitative genetics, before justifying it as a limit of Mendelian inheritance as the number, M, of underlying loci tends to infinity. As in the additive case, the multivariate normal distribution of trait values across the pedigree can be expressed in terms of variance components in an ancestral population and probabilities of identity by descent determined by the pedigree. Now, with just first-order dominance effects, we require two-, three-, and four-way identities. We also show that, even if we condition on parental trait values, the “shared” and “residual” components of trait values within each family will be asymptotically normally distributed as the number of loci tends to infinity, with an error of order 1/M−−√⁠. We illustrate our results with some numerical examples.}, author = {Barton, Nicholas H and Etheridge, Alison M. and Véber, Amandine}, issn = {1943-2631}, journal = {Genetics}, number = {2}, publisher = {Oxford Academic}, title = {{The infinitesimal model with dominance}}, doi = {10.1093/genetics/iyad133}, volume = {225}, year = {2023}, } @misc{12949, abstract = {The classical infinitesimal model is a simple and robust model for the inheritance of quantitative traits. In this model, a quantitative trait is expressed as the sum of a genetic and a non-genetic (environmental) component and the genetic component of offspring traits within a family follows a normal distribution around the average of the parents’ trait values, and has a variance that is independent of the trait values of the parents. Although the trait distribution across the whole population can be far from normal, the trait distributions within families are normally distributed with a variance-covariance matrix that is determined entirely by that in the ancestral population and the probabilities of identity determined by the pedigree. Moreover, conditioning on some of the trait values within the pedigree has predictable effects on the mean and variance within and between families. In previous work, Barton et al. (2017), we showed that when trait values are determined by the sum of a large number of Mendelian factors, each of small effect, one can justify the infinitesimal model as limit of Mendelian inheritance. It was also shown that under some forms of epistasis, trait values within a family are still normally distributed.}, author = {Barton, Nicholas H}, keywords = {Quantitative genetics, infinitesimal model}, publisher = {Institute of Science and Technology Austria}, title = {{The infinitesimal model with dominance}}, doi = {10.15479/AT:ISTA:12949}, year = {2023}, } @inproceedings{14461, abstract = {Communication-reduction techniques are a popular way to improve scalability in data-parallel training of deep neural networks (DNNs). The recent emergence of large language models such as GPT has created the need for new approaches to exploit data-parallelism. Among these, fully-sharded data parallel (FSDP) training is highly popular, yet it still encounters scalability bottlenecks. One reason is that applying compression techniques to FSDP is challenging: as the vast majority of the communication involves the model’s weights, direct compression alters convergence and leads to accuracy loss. We present QSDP, a variant of FSDP which supports both gradient and weight quantization with theoretical guarantees, is simple to implement and has essentially no overheads. To derive QSDP we prove that a natural modification of SGD achieves convergence even when we only maintain quantized weights, and thus the domain over which we train consists of quantized points and is, therefore, highly non-convex. We validate this approach by training GPT-family models with up to 1.3 billion parameters on a multi-node cluster. Experiments show that QSDP preserves model accuracy, while completely removing the communication bottlenecks of FSDP, providing end-to-end speedups of up to 2.2x.}, author = {Markov, Ilia and Vladu, Adrian and Guo, Qi and Alistarh, Dan-Adrian}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, issn = {2640-3498}, location = {Honolulu, Hawaii, HI, United States}, pages = {24020--24044}, publisher = {ML Research Press}, title = {{Quantized distributed training of large models with convergence guarantees}}, volume = {202}, year = {2023}, } @inproceedings{14462, abstract = {We study fine-grained error bounds for differentially private algorithms for counting under continual observation. Our main insight is that the matrix mechanism when using lower-triangular matrices can be used in the continual observation model. More specifically, we give an explicit factorization for the counting matrix Mcount and upper bound the error explicitly. We also give a fine-grained analysis, specifying the exact constant in the upper bound. Our analysis is based on upper and lower bounds of the completely bounded norm (cb-norm) of Mcount . Along the way, we improve the best-known bound of 28 years by Mathias (SIAM Journal on Matrix Analysis and Applications, 1993) on the cb-norm of Mcount for a large range of the dimension of Mcount. Furthermore, we are the first to give concrete error bounds for various problems under continual observation such as binary counting, maintaining a histogram, releasing an approximately cut-preserving synthetic graph, many graph-based statistics, and substring and episode counting. Finally, we note that our result can be used to get a fine-grained error bound for non-interactive local learning and the first lower bounds on the additive error for (ϵ,δ)-differentially-private counting under continual observation. Subsequent to this work, Henzinger et al. (SODA, 2023) showed that our factorization also achieves fine-grained mean-squared error.}, author = {Fichtenberger, Hendrik and Henzinger, Monika H and Upadhyay, Jalaj}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, issn = {2640-3498}, location = {Honolulu, Hawaii, HI, United States}, pages = {10072--10092}, publisher = {ML Research Press}, title = {{Constant matters: Fine-grained error bound on differentially private continual observation}}, volume = {202}, year = {2023}, }