Search results for: metaprogramming

Authors: Eashan Hatti

Abstract:

In software development, programmers often must make a choice between high-level programming and high-performance programs. High-level programming encourages the use of complex, pervasive abstractions. However, the use of these abstractions degrades performance-high performance demands that programs be low-level. In a compiler, the optimizer attempts to let the user have both. The optimizer takes high-level, abstract code as an input and produces low-level, performant code as an output. However, there is a problem with having the optimizer be a built-in part of the compiler. Domain-specific abstractions implemented as libraries are common in high-level languages. As a language’s library ecosystem grows, so does the number of abstractions that programmers will use. If these abstractions are to be performant, the optimizer must be extended with new optimizations to target them, or these abstractions must rely on existing general-purpose optimizations. The latter is often not as effective as needed. The former presents too significant of an effort for the compiler developers, as they are the only ones who can extend the language with new optimizations. Thus, the language becomes more high-level, yet the optimizer – and, in turn, program performance – falls behind. Programmers are again confronted with a choice between high-level programming and high-performance programs. To investigate a potential solution to this problem, we developed Peridot, a prototype programming language. Peridot’s main contribution is that it enables library developers to easily extend the language with new optimizations themselves. This allows the optimization workload to be taken off the compiler developers’ hands and given to a much larger set of people who can specialize in each problem domain. Because of this, optimizations can be much more effective while also being much more numerous. To enable this, Peridot supports metaprogramming designed for implementing program transformations. The language is split into two fragments or “levels”, one for metaprogramming, the other for high-level general-purpose programming. The metaprogramming level supports logic programming. Peridot’s key idea is that optimizations are simply implemented as metaprograms. The meta level supports several specific features which make it particularly suited to implementing optimizers. For instance, metaprograms can automatically deduce equalities between the programs they are optimizing via unification, deal with variable binding declaratively via higher-order abstract syntax, and avoid the phase-ordering problem via non-determinism. We have found that this design centered around logic programming makes optimizers concise and easy to write compared to their equivalents in functional or imperative languages. Overall, implementing Peridot has shown that its design is a viable solution to the problem of writing code which is both high-level and performant.

Keywords: optimization, metaprogramming, logic programming, abstraction

Procedia PDF Downloads 56

Authors: Gautier Viaud, Paul-Henry Cournède

Abstract:

Plant growth models have been used extensively for the prediction of the phenotypic performance of plants. However, they remain most often calibrated for a given genotype and therefore do not take into account genotype by environment interactions. One way of achieving such an objective is to consider Bayesian hierarchical models. Three levels can be identified in such models: The first level describes how a given growth model describes the phenotype of the plant as a function of individual parameters, the second level describes how these individual parameters are distributed within a plant population, the third level corresponds to the attribution of priors on population parameters. Thanks to the Bayesian framework, choosing appropriate priors for the population parameters permits to derive analytical expressions for the full conditional distributions of these population parameters. As plant growth models are of a nonlinear nature, individual parameters cannot be sampled explicitly, and a Metropolis step must be performed. This allows for the use of a hybrid Gibbs--Metropolis sampler. A generic approach was devised for the implementation of both general state space models and estimation algorithms within a programming platform. It was designed using the Julia language, which combines an elegant syntax, metaprogramming capabilities and exhibits high efficiency. Results were obtained for Arabidopsis thaliana on both simulated and real data. An organ-scale Greenlab model for the latter is thus presented, where the surface areas of each individual leaf can be simulated. It is assumed that the error made on the measurement of leaf areas is proportional to the leaf area itself; multiplicative normal noises for the observations are therefore used. Real data were obtained via image analysis of zenithal images of Arabidopsis thaliana over a period of 21 days using a two-step segmentation and tracking algorithm which notably takes advantage of the Arabidopsis thaliana phyllotaxy. Since the model formulation is rather flexible, there is no need that the data for a single individual be available at all times, nor that the times at which data is available be the same for all the different individuals. This allows to discard data from image analysis when it is not considered reliable enough, thereby providing low-biased data in large quantity for leaf areas. The proposed model precisely reproduces the dynamics of Arabidopsis thaliana’s growth while accounting for the variability between genotypes. In addition to the estimation of the population parameters, the level of variability is an interesting indicator of the genotypic stability of model parameters. A promising perspective is to test whether some of the latter should be considered as fixed effects.

Keywords: bayesian, genotypic differentiation, hierarchical models, plant growth models

Procedia PDF Downloads 279