This document is an invited chapter covering the specificities of ABC model choice, intended for the incoming Handbook of ABC by Sisson, Fan, and Beaumont (2017). Beyond exposing the potential pitfalls of ABC based posterior probabilities, the review emphasizes mostly the solution proposed by Pudlo et al. (2016) on the use of random forests for aggregating summary statistics and and for estimating the posterior probability of the most likely model via a secondary random fores.
This is the collection of solutions for all the exercises proposed in Bayesian Essentials with R (2014).
We perform three-dimensional numerical simulations of stellar winds of early-M dwarf stars. Our simulations incorporate observationally reconstructed large-scale surface magnetic maps, suggesting that the complexity of the magnetic field can play an important role in the angular momentum evolution of the star, possibly explaining the large distribution of periods in field dM stars, as reported in recent works. In spite of the diversity of the magnetic field topologies among the stars in our sample, we find that stellar wind flowing near the (rotational) equatorial plane carries most of the stellar angular momentum, but there is no preferred colatitude contributing to mass loss, as the mass flux is maximum at different colatitudes for different stars. We find that more non-axisymmetric magnetic fields result in more asymmetric mass fluxes and wind total pressures $p_{rm tot}$ (defined as the sum of thermal, magnetic and ram pressures). Because planetary magnetospheric sizes are set by pressure equilibrium between the planets magnetic field and $p_{rm tot}$, variations of up to a factor of $3$ in $p_{rm tot}$ (as found in the case of a planet orbiting at several stellar radii away from the star) lead to variations in magnetospheric radii of about 20 percent along the planetary orbital path. In analogy to the flux of cosmic rays that impact the Earth, which is inversely modulated with the non-axisymmetric component of the total open solar magnetic flux, we conclude that planets orbiting M dwarf stars like DT~Vir, DS~Leo and GJ~182, which have significant non-axisymmetric field components, should be the more efficiently shielded from galactic cosmic rays, even if the planets lack a protective thick atmosphere/large magnetosphere of their own.
We consider a class of toy models where a spatially flat universe is filled with a perfect fluid. The dynamics is found exactly for all these models. In one family, the perfect fluid is of the phantom type and we find that the universe is first contracting and then expanding while the dynamics is always accelerated. In a second family, the universe is first in an accelerated expansion stage, then in a decelerated expansion stage until it reaches a turning point after which it contracts in a decelerated way (increasing contraction rate) followed by another accelerated stage (decreasing contraction rate). We also consider the possibility to embed this perfect fluid in a realistic cosmology. The first family cannot be viable in a conventional big bang universe and requires a rebound in the very early universe. The second family is viable in the range $0<1+w_{DE,0}lesssim 0.09$ for a spatially closed universe with a curvature satisfying current bounds. Though many of the models in this family cannot be distinguished today from a universe dominated by a cosmological constant, the present accelerated expansion is transient and these universes will reach a turning point in the future before entering a contraction phase.
Approximate Bayesian computation (ABC) has become an essential tool for the analysis of complex stochastic models when the likelihood function is numerically unavailable. However, the well-established statistical method of empirical likelihood provides another route to such settings that bypasses simulations from the model and the choices of the ABC parameters (summary statistics, distance, tolerance), while being convergent in the number of observations. Furthermore, bypassing model simulations may lead to significant time savings in complex models, for instance those found in population genetics. The BCel algorithm we develop in this paper also provides an evaluation of its own performance through an associated effective sample size. The method is illustrated using several examples, including estimation of standard distributions, time series, and population genetics models.
The choice of the summary statistics used in Bayesian inference and in particular in ABC algorithms has bearings on the validation of the resulting inference. Those statistics are nonetheless customarily used in ABC algorithms without consistency checks. We derive necessary and sufficient conditions on summary statistics for the corresponding Bayes factor to be convergent, namely to asymptotically select the true model. Those conditions, which amount to the expectations of the summary statistics to asymptotically differ under both models, are quite natural and can be exploited in ABC settings to infer whether or not a choice of summary statistics is appropriate, via a Monte Carlo validation.
We investigate grand unified theories (GUTs) in scenarios where electroweak (EW) symmetry breaking is triggered by a light composite Higgs, arising as a Nambu-Goldstone boson from a strongly interacting sector. The evolution of the standard model (SM) gauge couplings can be predicted at leading order, if the global symmetry of the composite sector is a simple group G that contains the SM gauge group. It was noticed that, if the right-handed top quark is also composite, precision gauge unification can be achieved. We build minimal consistent models for a composite sector with these properties, thus demonstrating how composite GUTs may represent an alternative to supersymmetric GUTs. Taking into account the new contributions to the EW precision parameters, we compute the Higgs effective potential and prove that it realizes consistently EW symmetry breaking with little fine-tuning. The G group structure and the requirement of proton stability determine the nature of the light composite states accompanying the Higgs and the top quark: a coloured triplet scalar and several vector-like fermions with exotic quantum numbers. We analyse the signatures of these composite partners at hadron colliders: distinctive final states contain multiple top and bottom quarks, either alone or accompanied by a heavy stable charged particle, or by missing transverse energy.
We extract the pole positions, hadronic and gamma-gamma widths of sigma and f_0(980, from pi-pi and gamma-gamma scattering data using an improved analytic K-matrix model. Our results favour a large gluon component for the sigma and a bar ss or/and gluon component for the f_0(980) but neither a large four-quark nor a molecule component. Gluonium sigma_B production from J/psi, phi radiative and D_s semi-leptonic decays are also discussed.
The Adaptive Multiple Importance Sampling (AMIS) algorithm is aimed at an optimal recycling of past simulations in an iterated importance sampling scheme. The difference with earlier adaptive importance sampling implementations like Population Monte Carlo is that the importance weights of all simulated values, past as well as present, are recomputed at each iteration, following the technique of the deterministic multiple mixture estimator of Owen and Zhou (2000). Although the convergence properties of the algorithm cannot be fully investigated, we demonstrate through a challenging banana shape target distribution and a population genetics example that the improvement brought by this technique is substantial.
We extract directly (for the first time) the charmed (C=1) and bottom (B=-1) heavy-baryons (spin 1/2 and 3/2) mass-splittings due to SU(3) breaking using double ratios of QCD spectral sum rules (QSSR) in full QCD, which are less sensitive to the exact value and definition of the heavy quark mass, to the perturbative radiative corrections and to the QCD continuum contributions than the simple ratios commonly used for determining the heavy baryon masses. Noticing that most of the mass-splittings are mainly controlled by the ratio kappa= <bar ss>/<bar dd> of the condensate, we extract this ratio, by allowing 1 sigma deviation from the observed masses of the Xi_{c,b} and of the Omega_c. We obtain: kappa=0.74(3), which improves the existing estimates: kappa=0.70(10) from light hadrons. Using this value, we deduce M_{Omega_b}=6078.5(27.4) MeV which agrees with the recent CDF data but disagrees by 2.4 sigma with the one from D0. Predictions of the Xi_Q and of the spectra of spin 3/2 baryons containing one or two strange quark are given in Table 2. Predictions of the hyperfine splittings Omega*_Q- Omega_Q and Xi*_Q-Xi_Q are also given in Table 3. Starting for a general choice of the interpolating currents for the spin 1/2 baryons, our analysis favours the optimal value of the mixing angle b= (-1/5 -- 0) found from light and non-strange heavy baryons.
