arXiv subjects

Mafalda Dias

Publications and source records attributed to Mafalda Dias.

18 recordsLinked to original sources

arXiv2024

Guidelines for releasing a variant effect predictor

Computational methods for assessing the likely impacts of mutations, known as variant effect predictors (VEPs), are widely used in the assessment and interpretation of human genetic variation, as well as in other applications like protein engineering. Many different VEPs have been released to date, and there is tremendous variability in their underlying algorithms and outputs, and in the ways in which the methodologies and predictions are shared. This leads to considerable challenges for end users in knowing which VEPs to use and how to use them. Here, to address these issues, we provide guidelines and recommendations for the release of novel VEPs. Emphasising open-source availability, transparent methodologies, clear variant effect score interpretations, standardised scales, accessible predictions, and rigorous training data disclosure, we aim to improve the usability and interpretability of VEPs, and promote their integration into analysis and evaluation pipelines. We also provide a large, categorised list of currently available VEPs, aiming to facilitate the discovery and encourage the usage of novel methods within the scientific community.

q-bio.OT↗

arXiv2022

Tranception: protein fitness prediction with autoregressive transformers and inference-time retrieval

The ability to accurately model the fitness landscape of protein sequences is critical to a wide range of applications, from quantifying the effects of human variants on disease likelihood, to predicting immune-escape mutations in viruses and designing novel biotherapeutic proteins. Deep generative models of protein sequences trained on multiple sequence alignments have been the most successful approaches so far to address these tasks. The performance of these methods is however contingent on the availability of sufficiently deep and diverse alignments for reliable training. Their potential scope is thus limited by the fact many protein families are hard, if not impossible, to align. Large language models trained on massive quantities of non-aligned protein sequences from diverse families address these problems and show potential to eventually bridge the performance gap. We introduce Tranception, a novel transformer architecture leveraging autoregressive predictions and retrieval of homologous sequences at inference to achieve state-of-the-art fitness prediction performance. Given its markedly higher performance on multiple mutants, robustness to shallow alignments and ability to score indels, our approach offers significant gain of scope over existing approaches. To enable more rigorous model testing across a broader range of protein families, we develop ProteinGym -- an extensive set of multiplexed assays of variant effects, substantially increasing both the number and diversity of assays compared to existing benchmarks.

cs.LG↗

arXiv2019

Primordial Gravitational Waves and the Swampland

The swampland conjectures seek to distinguish effective field theories which can be consistently embedded in a theory of quantum gravity from those which can not (and are hence referred to as being in the swampland). We consider two such conjectures, known as the Swampland Distance and de Sitter Conjectures, showing that taken together they place bounds on the amplitude of primordial gravitational waves generated during single field slow-roll inflation. The bounds depend on two parameters which for reasonable estimates restrict the tensor-to-scalar ratio to be within reach of future surveys.

hep-th↗

arXiv2019

Pole N-flation

A second order pole in the scalar kinetic term can lead to a class of inflation models with universal predictions referred to as pole inflation or $α$-attractors. While this kinetic structure is ubiquitous in supergravity effective field theories, realising a consistent UV complete model in e.g. string theory is a non-trivial task. For one, one expects quantum corrections arising in the vicinity of the pole which may spoil the typical attractor dynamics. As a conservative estimate of the range of validity of supergravity models of pole inflation we employ the weak gravity conjecture (WGC). We find that this constrains the accessible part of the inflationary plateau by limiting the decay constant of the axion partner. For the original single complex field models, the WGC does not even allow the inflaton to reach the inflationary plateau region. We analyze if evoking the assistance of $N$ scalar fields from the open string moduli helps addressing these problems. Pole $N$-flation could improve radiative control by reducing the required range of each individual field. However, the WGC bound prohibiting pole inflation for a single such field persists even for a collective motion of $N$ such scalars if we impose the sublattice WGC. Finally, we outline steps towards an embedding of pole N-flation in type IIB string theory on fibred Calabi-Yau manifolds.

hep-th↗

arXiv2019

Inflation as an Information Bottleneck - A strategy for identifying universality classes and making robust predictions

In this work we propose a statistical approach to handling sources of theoretical uncertainty in string theory models of inflation. By viewing a model of inflation as a probabilistic graph, we show that there is an inevitable information bottleneck between the ultraviolet input of the theory and observables, as a simple consequence of the data processing theorem. This information bottleneck can result in strong hierarchies in the sensitivity of observables to the parameters of the underlying model and hence universal predictions with respect to at least some microphysical considerations. We also find other intriguing behaviour, such as sharp transitions in the predictions when certain hyperparameters cross a critical value. We develop a robust numerical approach to studying these behaviours by adapting methods often seen in the context of machine learning. We first test our approach by applying it to well known examples of universality, sharp transitions, and concentration phenomena in random matrix theory. We then apply the method to inflation with axion monodromy. We find universality with respect to a number of model parameters and that consistency with observational constraints implies that with very high probability certain perturbative corrections are non-negligible.

hep-th↗

arXiv2018

Seven Lessons from Manyfield Inflation in Random Potentials

We study inflation in models with many interacting fields subject to randomly generated scalar potentials. We use methods from non-equilibrium random matrix theory to construct the potentials and an adaption of the 'transport method' to evolve the two-point correlators during inflation. This construction allows, for the first time, for an explicit study of models with up to 100 interacting fields supporting a period of 'approximately saddle-point' inflation. We determine the statistical predictions for observables by generating over 30,000 models with 2-100 fields supporting at least 60 efolds of inflation. These studies lead us to seven lessons: i) Manyfield inflation is not single-field inflation, ii) The larger the number of fields, the simpler and sharper the predictions, iii) Planck compatibility is not rare, but future experiments may rule out this class of models, iv) The smoother the potentials, the sharper the predictions, v) Hyperparameters can transition from stiff to sloppy, vi) Despite tachyons, isocurvature can decay, vii) Eigenvalue repulsion drives the predictions. We conclude that many of the 'generic predictions' of single-field inflation can be emergent features of complex inflation models.

astro-ph.CO↗

arXiv2016

Numerical evaluation of the bispectrum in multiple field inflation

We present a complete framework for numerical calculation of the power spectrum and bispectrum in canonical inflation with an arbitrary number of light or heavy fields. Our method includes all relevant effects at tree-level in the loop expansion, including (i) interference between growing and decaying modes near horizon exit; (ii) correlation and coupling between species near horizon exit and on superhorizon scales; (iii) contributions from mass terms; and (iv) all contributions from coupling to gravity. We track the evolution of each correlation function from the vacuum state through horizon exit and the superhorizon regime, with no need to match quantum and classical parts of the calculation; when integrated, our approach corresponds exactly with the tree-level Schwinger or 'in-in' formulation of quantum field theory. In this paper we give the equations necessary to evolve all two- and three-point correlation functions together with suitable initial conditions. The final formalism is suitable to compute the amplitude, shape, and scale dependence of the bispectrum in models with |fNL| of order unity or less, which are a target for future galaxy surveys such as Euclid, DESI and LSST. As an illustration we apply our framework to a number of examples, obtaining quantitatively accurate predictions for their bispectra for the first time. Two accompanying reports describe publicly-available software packages that implement the method.

astro-ph.CO↗

arXiv2016

Simple emergent power spectra from complex inflationary physics

We construct ensembles of random scalar potentials for $N_f$ interacting scalar fields using non-equilibrium random matrix theory, and use these to study the generation of observables during small-field inflation. For $N_f={\cal O}({\rm few})$, these heavily featured scalar potentials give rise to power spectra that are highly non-linear, at odds with observations. For $N_f\gg 1$, the superhorizon evolution of the perturbations is generically substantial, yet the power spectra simplify considerably and become more predictive, with most realisations being well approximated by a linear power spectrum. This provides proof of principle that complex inflationary physics can give rise to simple emergent power spectra. We explain how these results can be understood in terms of large $N_f$ universality of random matrix theory.

astro-ph.CO↗

arXiv2016

The curvature perturbation at second order

We give an explicit relation, up to second-order terms, between scalar-field fluctuations defined on spatially-flat slices and the curvature perturbation on uniform-density slices. This expression is a necessary ingredient for calculating observable quantities at second-order and beyond in multiple-field inflation. We show that traditional cosmological perturbation theory and the `separate universe' approach yield equivalent expressions for superhorizon wavenumbers, and in particular that all nonlocal terms can be eliminated from the perturbation-theory expressions.

gr-qc↗

arXiv2016

Computing observables in curved multifield models of inflation - A guide (with code) to the transport method

We describe how to apply the transport method to compute inflationary observables in a broad range of multiple-field models. The method is efficient and encompasses scenarios with curved field-space metrics, violations of slow-roll conditions and turns of the trajectory in field space. It can be used for an arbitrary mass spectrum, including massive modes and models with quasi-single-field dynamics. In this note we focus on practical issues. It is accompanied by a Mathematica code which can be used to explore suitable models, or as a basis for further development.

astro-ph.CO↗

arXiv2015

Large Scale Power Suppression in a Multifield Landscape

Power suppression of the cosmic microwave background on the largest observable scales could provide valuable clues about the particle physics underlying inflation. Here we consider the prospect of power suppression in the context of the multifield landscape. Based on the assumption that our observable universe emerges from a tunnelling event and that the relevant features originate purely from inflationary dynamics, we find that the power spectrum not only contains information on single-field dynamics, but also places strong con- straints on all scalar fields present in the theory. We find that the simplest single-field models giving rise to power suppression do not generalise to multifield models in a straightforward way, as the resulting superhorizon evolution of the curvature perturbation tends to erase any power suppression present at horizon crossing. On the other hand, multifield effects do present a means of generating power suppression which to our knowledge has so far not been considered. We propose a mechanism to illustrate this, which we dub flume inflation.

astro-ph.CO↗

arXiv2014

A New Angle on Chaotic Inflation

N-flation is a radiatively stable scenario for chaotic inflation in which the displacements of $N \gg 1$ axions with decay constants $f_1 \le \ldots \le f_N < M_P$ lead to a super-Planckian effective displacement equal to the Pythagorean sum $f_{Py}$ of the $f_i$. We show that mixing in the axion kinetic term generically leads to the phenomenon of kinetic alignment, allowing for effective displacements as large as $\sqrt{N} f_{N} \ge f_{Py}$, even if $f_1, \ldots, f_{N-1}$ are arbitrarily small. At the level of kinematics, the necessary alignment occurs with very high probability, because of eigenvector delocalization. We present conditions under which inflation can take place along an aligned direction. Our construction sharply reduces the challenge of realizing N-flation in string theory.

hep-th↗

arXiv2013

The δN formula is the dynamical renormalization group

We derive the 'separate universe' method for the inflationary bispectrum, beginning directly from a field-theory calculation. We work to tree-level in quantum effects but to all orders in the slow-roll expansion, with masses accommodated perturbatively. Our method provides a systematic basis to account for novel sources of time-dependence in inflationary correlation functions, and has immediate applications. First, we use our result to obtain the correct matching prescription between the 'quantum' and 'classical' parts of the separate universe computation. Second, we elaborate on the application of this method in situations where its validity is not clear. As a by-product of our calculation we give the leading slow-roll corrections to the three-point function of field fluctuations on spatially flat hypersurfaces in a canonical, multiple-field model.

astro-ph.CO↗

arXiv2013

Scale-dependent bias from multiple-field inflation

We provide a formula for the scaling behaviour of the inflationary bispectrum in the 'squeezed' limit where one momentum becomes much smaller than the other two. This determines the scaling of the halo bias at low wavenumber and will be an important observable for the next generation of galaxy surveys. Our formula allows it to be predicted for the first time for a generic inflationary model with multiple light, canonically-normalized scalar fields.

astro-ph.CO↗

arXiv2013

Multifield consequences for D-brane inflation

We analyse the multifield behaviour in D-brane inflation when contributions from the bulk are taken into account. For this purpose, we study a large number of realisations of the potential; we find the nature of the inflationary trajectory to be very consistent despite the complex construction. Inflation is always canonical and occurs in the vicinity of an inflection point. Extending the transport method to non-slow-roll and to calculate the running, we obtain distributions for observables. The spectral index is typically blue and the running positive, putting the model under moderate pressure from WMAP7 constraints. The local f_NL and tensor-to-scalar ratio are typically unobservably small, though we find approximately 0.5% of realisations to give observably large local f_NL. Approximating the potential as sum-separable, we are able to give fully analytic explanations for the trends in observed behaviour. Finally we find the model suffers from the persistence of isocurvature perturbations, which can be expected to cause further evolution of adiabatic perturbations after inflation. We argue this is a typical problem for models of multifield inflation involving inflection points and renders models of this type technically unpredictive without a description of reheating.

astro-ph.CO↗

arXiv2012

Transport equations for the inflationary spectral index

We present a simple and efficient method to compute the superhorizon evolution of the spectral index in multi-field inflationary models, using transport equation techniques. We illustrate the evolution of n_s with time for various interesting potentials.

astro-ph.CO↗

arXiv2011

Cosmology at the boundary of de Sitter using the dS/QFT correspondence

Using the dS/QFT correspondence in the context of inflation allows for the study of interesting, otherwise inaccessible physics. In particular, by studying inflation via its dual field theory at the boundary of the de Sitter space, it may be possible to study a regime of strongly coupled gravity at early times. The purpose of this work is to completely express cosmological observables in terms of the free parameters of a dual field theory and to compare them with CMB data. In this way, constraints on the observational parameters constrains the validity of the strongly coupled inflation picture by imposing limits on the parameters of the field theory. The fit with data defines a limit for the consistency and validity of the approach taken and shows that, within this limit, the model is almost unconstrained, but quite predictive, producing power spectra of density perturbations extremely near scale invariance.

astro-ph.CO↗

arXiv2010

On the possibility of braneworld quintessential inflation

We examine the possibility of achieving quintessential inflation, where the same field serves as both inflaton and quintessence, in the context of a five-dimensional braneworld. Braneworld cosmology provides an appropriate environment as it permits inflation with much steeper potentials than the conventional scenario, which is favourable to a late-time quintessence. We explore a wide space of models, together with contemporary observational data, to determine in which contexts such a picture is possible. We find that such a scenario, although attractive, is in fact impossible to achieve for the potentials studied due to the restrictiveness of current data.

astro-ph.CO↗