Search NASA⌕ Search

SEARCH · Search NASA

Results for “maximum likelihood”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 865 records · Page 48

Elliptically-Contoured Tensor-variate Distributions with Application to Image Learning

Statistical analysis of tensor-valued data has largely used the tensor-variate normal (TVN) distribution that may be inadequate for data arising from distributions with heavier or lighter tails. We study a general family of elliptically contoured (EC) TV distributions and derive its characterizations, moments, marginal, and conditional distributions. We describe procedures for maximum likelihood estimation from data that are (1) uncorrelated draws from an EC distribution, (2) from a scale mixture of the TVN distribution, and (3) from an underlying but unknown EC distribution, for which we extend Tyler’s robust estimator. A detailed simulation study highlights the benefits of choosing an EC distribution over the TVN for heavier-tailed data. We develop TV classification rules using discriminant analysis and EC errors and show that they better predict cats and dogs from images in the Animal Faces-HQ dataset than the TVN-based rules. A novel tensor-on-tensor regression and TV analysis of variance (TANOVA) framework under EC errors is also demonstrated to better characterize gender, age, and ethnic origin than the usual TVN-based TANOVA in the celebrated labeled faces of the wild dataset.

97 MATHEMATICS AND COMPUTING↗

Parameter Estimation for Electrode Degradation: Learning in the Face of Model-Experiment Discrepancies

Use of physics-based models to interpret battery degradation data over the course of cycling can provide deeper physical insight into the internal states of the system and how they evolve. We present a neural network trained on simulations generated by a previously published physics-based model for a lithium trivanadate (LVO) cathode to estimate parameters that evolve over the course of cycling. We focus on the robustness of the neural network through two case studies that probe different kinds of discrepancies between model and experiment: nonideal data and imperfect model. In the former, the experimental protocols do not meet the assumption made in the training data generated by the physics-based model, while in the latter, the physics-based model fails to describe all of the measured cathode behavior even under ideal conditions. When there is total model-experiment agreement, a neural network estimates parameters with improved accuracy compared to a maximum likelihood analysis using the same set of simulations. However, in both types of model-experiment discrepancy, the neural network returned biased parameter estimates. We introduce a data augmentation procedure into the neural network training to mitigate these effects and improve robustness, and employ it to estimate parameters for a cycling LVO cathode.

Mayilvahanan, Karthik S. (ORCID:0000000316282332)↗

Molecular phylogenetic analysis of the Paguristes tortugae Schmitt, 1933 complex and selected other Paguroidea (Crustacea: Decapoda: Anomura)

Morphological characters, as presently applied to describe members of the Paguristes tortugae Schmitt, 1933 species complex, appear to be of limited value in inferring phylogenetic relationships within the genus, and may have similarly misinformed understanding of relationships between members of this complex and those presently assigned to the related genera Areopaguristes Rahayu & McLaughlin, 2010 and Pseudopaguristes McLaughlin, 2002. Previously undocumented observations of similarities and differences in color patterns among populations additionally suggest genetic divergences within some species, or alternatively seem to support phylogenetic groupings of some species. In the present study, a Maximum Likelihood (ML) phylogenetic analysis was undertaken based on the H3, 12S mtDNA, and 16S mtDNA sequences of 148 individuals, primarily representatives of paguroid species from the western Atlantic. This molecular analysis supported a polyphyletic Diogenidae Ortmann, 1892, although incomplete taxonomic sampling among the genera of Diogenidae limits the utility of this finding for resolving family level relationships. Several hypotheses regarding the evolutionary relationships among hermit crab genera were refuted by the Kishino-Hasegawa (KH). ShimodairaHasegawa (SH) and Approximately Unbiased (AU) tree topology tests, among them the hypothesis that Areopaguristes is monophyletic. A lack of support for the monophyly of Areopaguristes calls into question the phylogenetic validity of gill number for the differentiation of Paguristes, Areopaguristes, and Pseudopaguristes. The study was inconclusive with regard to the relationships among these three genera, but previously unknown diversity within both Paguristes and Areopaguristes was demonstrated. Existence of an undescribed species confounded under the name Paguristes tortugae Schmitt, 1933 was supported by genetics, morphology, and coloration. A second undescribed species with remarkable similarity to Areopaguristes hummi Wass, 1955 was discovered based on genetics and coloration.

12S↗

A Bayesian Approach to Regional Decadal Predictability: Sparse Parameter Estimation in High-Dimensional Linear Inverse Models of High-Latitude Sea Surface Temperature Variability

Stochastic reduced models are an important tool in climate systems whose many spatial and temporal scales cannot be fully discretized or underlying physics may not be fully accounted for. One form of reduced model, the linear inverse model (LIM), has been widely used for regional climate predictability studies—typically focusing more on tropical or midlatitude studies. However, most LIM fitting techniques rely on point estimation techniques deriving from fluctuation–dissipation theory. In this methodological study we explore the use of Bayesian inference techniques for LIM parameter estimation of sea surface temperature (SST), to quantify the skillful decadal predictability of Bayesian LIM models at high latitudes. We show that Bayesian methods, when compared to traditional point estimation methods for LIM-type models, provide better calibrated probabilistic skill, while simultaneously providing better point estimates due to the regularization effect of the prior distribution in high-dimensional problems. We compare the effect of several priors, as well as maximum likelihood estimates, on 1) estimating parameter values on a perfect model experiment and 2) producing calibrated 1-yr SST anomaly forecast distributions using a preindustrial control run of the Community Earth System Model (CESM). Finally, we employ a host of probabilistic skill metrics to determine the extent to which an LIM can forecast SST anomalies at high latitudes. We find that the choice of prior distribution has an appreciable impact on estimation outcomes, and priors that emphasize physically relevant properties enhance the model’s ability to capture variability of SST anomalies.

54 ENVIRONMENTAL SCIENCES↗

Estimating Sparse Direct Effects in Multivariate Regression With the Spike-and-Slab LASSO

The multivariate regression interpretation of the Gaussian chain graph model simultaneously parametrizes (i) the direct effects of p predictors on q outcomes and (ii) the residual partial covariances between pairs of outcomes. We introduce a new method for fitting sparse versions of these models with spike-and-slab LASSO (SSL) priors. We develop an Expectation Conditional Maximization algorithm to obtain sparse estimates of the p × q matrix of direct effects and the q × q residual precision matrix. Our algorithm iteratively solves a sequence of penalized maximum likelihood problems with self-adaptive penalties that gradually filter out negligible regression coefficients and partial covariances. Because it adaptively penalizes individual model parameters, our method is seen to outperform fixed-penalty competitors on simulated data. We establish the posterior contraction rate for our model, buttressing our method’s excellent empirical performance with strong theoretical guarantees. Using our method, we estimated the direct effects of diet and residence type on the composition of the gut microbiome of elderly adults.

EM algorithm↗

Sempervirens: A Fast Reconstruction Algorithm for Noisy and Incomplete Binary Matrix Representations of Trees

Applications such as reconstructing cell lineage trees (represented as phylogenetic trees) from single-cell sequencing data require reconstructing a {0,1}-matrix that has many errors and missing entries. We introduce Sempervirens, a very fast matrix reconstruction algorithm for noisy and incomplete matrix representations of phylogenetic trees. Sempervirens uses an iterative maximum-likelihood approach to determine the topology tree represented by the corrupted data. We show that Sempervirens is at least three orders of magnitude faster than other methods on thousand by thousand matrices, with the speed gap widening with larger matrices. We also show that Sempervirens matches state-of-the-art methods in reconstruction accuracy. The speed of Sempervirens enables it to be tractably applied to reconstructing much larger matrices than those that other methods can reconstruct. In addition to experimental results, we justify the algorithm with a mathematical treatment of its subprocedures.

algorithms↗

Gamma-Ray Simulated Spectrum Deconvolution of a LaBr₃ 1-in. x 1-in. Scintillator for Nondestructive ATR Fuel Burnup On-Site Predictions

A deconvolution method for a LaBr₃ 1"x1" detector for nondestructive Advanced Test Reactor (ATR) fuel burnup applications was developed. The method consisted of obtaining the detector response function, applying a deconvolution algorithm to 1”x1” LaBr₃ simulated, data along with evaluating the effects that deconvolution have on nondestructively determining ATR fuel burnup. The simulated response function of the detector was obtained using MCNPX as well with experimental data. The Maximum-Likelihood Expectation Maximization (MLEM) deconvolution algorithm was selected to enhance one-isotope source-simulated and fuel- simulated spectra. The final evaluation of the study consisted of measuring the performance of the fuel burnup calibration curve for the convoluted and deconvoluted cases. The methodology was developed in order to help design a reliable, high resolution, rugged and robust detection system for the ATR fuel canal capable of collecting high performance data for model validation, along with a system that can calculate burnup and using experimental scintillator detector data.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

X-ray computed tomography using partially coherent Fresnel diffraction with application to an optical fiber

A reconstruction algorithm for partially coherent x-ray computed tomography (XCT) including Fresnel diffraction is developed and applied to an optical fiber. The algorithm is applicable to a high-resolution tube-based laboratory-scale x-ray tomography instrument. The computing time is only a few times longer than the projective counterpart. The algorithm is used to reconstruct, with projections and diffraction, a tilt series acquired at the micrometer scale of a graded-index optical fiber using maximum likelihood and a Bayesian method based on the work of Bouman and Sauer. The inclusion of Fresnel diffraction removes some reconstruction artifacts and use of a Bayesian prior probability distribution removes others, resulting in a substantially more accurate reconstruction.

Levine, Zachary H. (ORCID:000000029928276X)↗

Physics-assisted generative adversarial network for X-ray tomography

X-ray tomography is capable of imaging the interior of objects in three dimensions non-invasively, with applications in biomedical imaging, materials science, electronic inspection, and other fields. The reconstruction process can be an ill-conditioned inverse problem, requiring regularization to obtain satisfactory results. Recently, deep learning has been adopted for tomographic reconstruction. Unlike iterative algorithms which require a distribution that is known a priori , deep reconstruction networks can learn a prior distribution through sampling the training distributions. In this work, we develop a Physics-assisted Generative Adversarial Network (PGAN), a two-step algorithm for tomographic reconstruction. In contrast to previous efforts, our PGAN utilizes maximum-likelihood estimates derived from the measurements to regularize the reconstruction with both known physics and the learned prior. Compared with methods with less physics assisting in training, PGAN can reduce the photon requirement with limited projection angles to achieve a given error rate. The advantages of using a physics-assisted learned prior in X-ray tomography may further enable low-photon nanoscale imaging.

47 OTHER INSTRUMENTATION↗

Evolutionary Dynamics of West Nile Virus in the United States, 1999–2011: Phylogeny, Selection Pressure and Evolutionary Time-Scale Analysis

West Nile virus (WNV), an arbovirus maintained in a bird-mosquito enzootic cycle, can infect other vertebrates including humans. WNV was first reported in the US in 1999 where, to date, three genotypes belonging to WNV lineage I have been described (NY99, WN02, SW/WN03). We report here the WNV sequences obtained from two birds, one mosquito, and 29 selected human samples acquired during the US epidemics from 2006–2011 and our examination of the evolutionary dynamics in the open-reading frame of WNV isolates reported from 1999–2011. Maximum-likelihood and Bayesian methods were used to perform the phylogenetic analyses and selection pressure analyses were conducted with the HyPhy package. Phylogenetic analysis identified human WNV isolates within the main WNV genotypes that have circulated in the US. Within genotype SW/WN03, we have identified a cluster with strains derived from blood donors and birds from Idaho and North Dakota collected during 2006–2007, termed here MW/WN06. Using different codon-based and branch-site selection models, we detected a number of codons subjected to positive pressure in WNV genes. The mean nucleotide substitution rate for WNV isolates obtained from humans was calculated to be 5.0661024 substitutions/site/year (s/s/y). The Bayesian skyline plot shows that after a period of high genetic variability following the introduction of WNV into the US, the WNV population appears to have reached genetic stability. The establishment of WNV in the US represents a unique opportunity to understand how an arbovirus adapts and evolves in a naive environment. We describe a novel, well-supported cluster of WNV formed by strains collected from humans and birds from Idaho and North Dakota. Adequate genetic surveillance is essential to public health since new mutants could potentially affect viral pathogenesis, decrease performance of diagnostic assays, and negatively impact the efficacy of vaccines and the development of specific therapies.

59 BASIC BIOLOGICAL SCIENCES↗

Genetic Variability of West Nile Virus in U.S. Blood Donors from the 2012 Epidemic Season

West Nile virus (WNV) is an arbovirus maintained in nature in a bird-mosquito enzootic cycle which can also infect other vertebrates including humans. WNV is now endemic in the United States (U.S.), causing yearly outbreaks that have resulted in an estimated total of 4–5 million human infections. Over 41,700 cases of West Nile disease, including 18,810 neuroinvasive cases and 1,765 deaths, were reported to the CDC between 1999 and 2014. In 2012, the second largest West Nile outbreak in the U.S. was reported, which caused 5,674 cases and 286 deaths. WNV continues to evolve, and three major WNV lineage I genotypes (NY99, WN02, and SW/WN03) have been described in the U.S. since introduction of the virus in 1999. We report here the WNV sequences obtained from 19 human samples acquired during the 2012 U.S. outbreak and our examination of the evolutionary dynamics in WNV isolates sequenced from 1999–2012. Maximum-likelihood and Bayesian methods were used to perform the phylogenetic analyses. Selection pressure analyses were performed with the HyPhy package using the Datamonkey web-server. Using different codon-based and branch-site selection models, we detected a number of codons subjected to positive pressure in WNV genes. Thirteen of the 19 completely sequenced isolates from 10 U.S. states were genetically similar, sharing up to 55 nucleotide mutations and 4 amino acid substitutions when compared with the prototype isolate WN-NY99. Altogether, these analyses showed that following a brief contraction in 2008–2009, WNV genetic divergence in the U.S. continued to increase in 2012, and that closely related variants were found across a broad geographic range of the U.S., coincident with the second-largest WNV outbreak in U.S. history.

59 BASIC BIOLOGICAL SCIENCES↗

Human Mammary Epithelial Cells Exhibit a Bimodal Correlated Random Walk Pattern

Organisms, at scales ranging from unicellular to mammals, have been known to exhibit foraging behavior described by random walks whose segments confirm to Lévy or exponential distributions. For the first time, we present evidence that single cells (mammary epithelial cells) that exist in multi-cellular organisms (humans) follow a bimodal correlated random walk (BCRW). Cellular tracks of MCF-10A pBabe, neuN and neuT random migration on 2-D plastic substrates, analyzed using bimodal analysis, were found to reveal the BCRW pattern. We find two types of exponentially distributed correlated flights (corresponding to what we refer to as the directional and re-orientation phases) each having its own correlation between move step-lengths within flights. The exponential distribution of flight lengths was confirmed using different analysis methods (logarithmic binning with normalization, survival frequency plots and maximum likelihood estimation). Because of the presence of non-uniform turn angle distribution of move step-lengths within a flight and two different types of flights, we propose that the epithelial random walk is a BCRW comprising of two alternating modes with varying degree of correlations, rather than a simple persistent random walk. A BCRW model rather than a simple persistent random walk correctly matches the super-diffusivity in the cell migration paths as indicated by simulations based on the BCRW model.

59 BASIC BIOLOGICAL SCIENCES↗

Daily Sampling of an HIV-1 Patient with Slowly Progressing Disease Displays Persistence of Multiple env Subpopulations Consistent with Neutrality

The molecular evolution of HIV-1 is characterized by frequent substitutions, indels and recombination events. In addition, a HIV-1 population may adapt through frequency changes of its variants. To reveal such population dynamics we analyzed HIV-1 subpopulation frequencies in an untreated patient with stable, low plasma HIV-1 RNA levels and close to normal CD4+ T-cell levels. The patient was intensively sampled during a 32-day period as well as approximately 1.5 years before and after this period (days 2664, 1, 2, 3, 11, 18, 25, 32 and 522). 77 sequences of HIV-1 env (approximately 3100 nucleotides) were obtained from plasma by limiting dilution with 7–11 sequences per time point, except day 2664. Phylogenetic analysis using maximum likelihood methods showed that the sequences clustered in six distinct subpopulations. We devised a method that took into account the relatively coarse sampling of the population. Data from days 1 through 32 were consistent with constant within-patient subpopulation frequencies. However, over longer time periods, i.e. between days 1…32 and 522, there were significant changes in subpopulation frequencies, which were consistent with evolutionarily neutral fluctuations. We found no clear signal of natural selection within the subpopulations over the study period, but positive selection was evident on the long branches that connected the subpopulations, which corresponds to >3 years as the subpopulations already were established when we started the study. Thus, selective forces may have been involved when the subpopulations were established. Genetic drift within subpopulations caused by de novo substitutions could be resolved after approximately one month. Overall, we conclude that subpopulation frequencies within this patient changed significantly over a time period of 1.5 years, but that this does not imply directional or balancing selection. We show that the short-term evolution we study here is likely representative for many patients of slow and normal disease progression.

59 BASIC BIOLOGICAL SCIENCES↗

Developmental Pathway of the MPER-Directed HIV-1-Neutralizing Antibody 10E8

Antibody 10E8 targets the membrane-proximal external region (MPER) of HIV-1 gp41, neutralizes >97% of HIV-1 isolates, and lacks the auto-reactivity often associated with MPER-directed antibodies. The developmental pathway of 10E8 might therefore serve as a promising template for vaccine design, but samples from time-of-infection—often used to infer the B cell record—are unavailable. In this study, we used crystallography, next-generation sequencing (NGS), and functional assessments to infer the 10E8 developmental pathway from a single time point. Mutational analysis indicated somatic hypermutation of the 2nd-heavy chain-complementarity determining region (CDR H2) to be critical for neutralization, and structures of 10E8 variants with V-gene regions reverted to genomic origin for heavy-and-light chains or heavy chain-only showed structural differences >2 Å relative to mature 10E8 in the CDR H2 and H3. To understand these developmental changes, we used bioinformatic sieving, maximum likelihood, and parsimony analyses of immunoglobulin transcripts to identify 10E8-lineage members, to infer the 10E8-unmutated common ancestor (UCA), and to calculate 10E8-developmental intermediates. We were assisted in this analysis by the preservation of a critical D-gene segment, which was unmutated in most 10E8-lineage sequences. UCA and early intermediates weakly bound a 26-residue-MPER peptide, whereas HIV-1 neutralization and epitope recognition in liposomes were only observed with late intermediates. Antibody 10E8 thus develops from a UCA with weak MPER affinity and substantial differences in CDR H2 and H3 from the mature 10E8; only after extensive somatic hypermutation do 10E8-lineage members gain recognition in the context of membrane and HIV-1 neutralization.

60 APPLIED LIFE SCIENCES↗

Phylogenetic diversity, trichothecene potential, and pathogenicity within Fusarium sambucinum species complex

The Fusarium sambucinum species complex (FSAMSC) is one of the most taxonomically challenging groups of fusaria, comprising prominent mycotoxigenic plant pathogens and other species with various lifestyles. Among toxins produced by members of the FSAMSC, trichothecenes pose the most significant threat to public health. Herein a global collection of 171 strains, originating from diverse hosts or substrates, were selected to represent FSAMSC diversity. This strain collection was used to assess their species diversity, evaluate their potential to produce trichothecenes, and cause disease on wheat. Maximum likelihood and Bayesian analyses of a combined 3-gene dataset used to infer evolutionary relationships revealed that the 171 strains originally received as 48 species represent 74 genealogically exclusive phylogenetically distinct species distributed among six strongly supported clades: Brachygibbosum, Graminearum, Longipes, Novel, Sambucinum, and Sporotrichioides. Most of the strains produced trichothecenes in vitro but varied in type, indicating that the six clades correspond to type A, type B, or both types of trichothecene-producing lineages. Furthermore, five strains representing two putative novel species within the Sambucinum Clade produced two newly discovered type A trichothecenes, 15-keto NX-2 and 15-keto NX-3. Strains of the two putatively novel species together with members of the Graminearum Clade were aggressive toward wheat when tested for pathogenicity on heads of the susceptible cultivar Apogee. In planta, the Graminearum Clade strains produced nivalenol or deoxynivalenol and the aggressive Sambucinum Clade strains synthesized NX-3 and 15-keto NX-3. Other strains within the Brachygibbosum, Longipes, Novel, Sambucinum, and Sporotrichioides Clades were nonpathogenic or could infect the inoculated floret without spreading within the head. Moreover, most of these strains did not produce any toxin in the inoculated spikelets. These data highlight aggressiveness toward wheat appears to be influenced by the type of toxin produced and that it is not limited to members of the Graminearum Clade.

59 BASIC BIOLOGICAL SCIENCES↗

Origin and evolution of HIV-1 subtype A6

Background: HIV outbreaks in the Former Soviet Union (FSU) countries were characterized by repeated transmission of the HIV variant AFSU, which is now classified as a distinct subtype A sub-subtype called A6. The current study used phylogenetic/phylodynamic and signature mutation analyses to determine likely evolutionary relationship between subtype A6 and other subtype A sub-subtypes. Methods: For this study, an initial Maximum Likelihood phylogenetic analysis was performed using a total of 553 full-length, publicly available, reverse transcriptase sequences, from A1, A2, A3, A4, A5, and A6 sub-subtypes of subtype A. For phylogenetic clustering and signature mutation analysis, a total of 5961 and 3959 pol and env sequences, respectively, were used. Results: Phylogenetic and signature mutation analysis showed that HIV-1 sub-subtype A6 likely originated from sub-subtype A1 of African origin. A6 and A1 pol and env genes shared several signature mutations that indicate genetic similarity between the two subtypes. For A6, tMRCA dated to 1975, 15 years later than that of A1. Conclusion: The current study provides insights into the evolution and diversification of A6 in the backdrop of FSU countries and indicates that A6 in FSU countries evolved from A1 of African origin and is getting bridged outside the FSU region.

60 APPLIED LIFE SCIENCES↗

An Indirect Search for Weakly Interacting Massive Particles in the Sun Using Upward-going Muons in NOvA

I present the first Dark Matter search results using the full data set collected with the upward-going muon trigger in NOvA. Weakly Interactive Massive Particles (WIMPs) are a theoretical non-baryonic form of Dark Matter. The nature of Dark Matter is one of the most exciting open questions in modern physics. Though its existence can be inferred by astrophysical evidence, its properties are not yet understood. If we assume that Dark Matter particles can produce Standard Model particles through their interactions, an indirect search can help shed light on this mystery.The NOvA collaboration has built a 14 kton, fine-grained, low-Z, total absorption tracking calorimeter at an off-axis angle to the NuMI neutrino beam. Even though the detector is optimized to observe electron neutrino appearance from a muon neutrino beam, it has a unique potential for more exotic searches given its excellent granularity and energy resolution and relatively low-energy neutrino thresholds. In fact, with an efficient upward-going muon trigger and sufficient background suppression offline, NOvA is capable of a competitive indirect Dark Matter search for low-mass WIMPs.The idea of the upward-going muon trigger is first to select high-quality muon tracks, then use the timing information of all of the hits of each track to estimate directionality. In this way, the background flux is suppressed by more than a factor of 10^5 at trigger level to a rate of approximately 1 Hz. To further optimize this search, we use only upward-going muons that point to the Sun, so our search occurs at night when the Sun is on the other side of the Earth. This strategy also allows us to use the time when the Sun is above the horizon as a control region to estimate the background. Ultimately, implementation of a cut based and maximum likelihood analysis provides a powerful tool for rejecting background and selecting a sample of neutrino-induced upward-going muons. The overall background rejection power achieved by the analysis is substantial and impressive. Starting with approximately 150,000 events per second, we reduced it to 40 events per year. Since no statistically significant excess was found, a 90\% C.L. upper limit on the expected muon flux of upward-going muons has been set using the upper limit on the number of events given the number of observed events in the signal region. Lastly, by assuming the theory behind the upward-going muon flux, a limit on the WIMP-nucleon spin-dependent cross-section in the Sun was estimated. Although the limits on the spin-dependent cross-section do not appear to be competitive with previous indirect Dark Matter searches, the upward-going muon flux limits are promising. The upward-going muon flux limits could extend these results to a broader class of models that are not specific to the dark matter theory but produce upward-going muons, leading to competitive results.

Principato, Cristiana↗

Stan : A Probabilistic Programming Language

Stan is a probabilistic programming language for specifying statistical models. A Stan program imperatively defines a log probability function over parameters conditioned on specified data and constants. As of version 2.14.0, Stan provides full Bayesian inference for continuous-variable models through Markov chain Monte Carlo methods such as the No-U-Turn sampler, an adaptive form of Hamiltonian Monte Carlo sampling. Penalized maximum likelihood estimates are calculated using optimization methods such as the limited memory Broyden-Fletcher-Goldfarb-Shanno algorithm. Stan is also a platform for computing log densities and their gradients and Hessians, which can be used in alternative algorithms such as variational Bayes, expectation propagation, and marginal inference using approximate integration. To this end, Stan is set up so that the densities, gradients, and Hessians, along with intermediate quantities of the algorithm such as acceptance probabilities, are easily accessible. Stan can also be called from the command line using the cmdstan package, through R using the rstan package, and through Python using the pystan package. All three interfaces support sampling and optimization-based inference with diagnostics and posterior analysis. rstan and pystan also provide access to log probabilities, gradients, Hessians, parameter transforms, and specialized plotting.

97 MATHEMATICS AND COMPUTING↗