Thierry Faure

Declarative task delegation in OpenMOLE

In this paper we present OpenMOLE, a scientific framework providing a virtualized runtime environment for distributed computing. Current distributed execution systems do not hide the hardware and software heterogeneity of computing and data resources whereas OpenMOLE provides generic services to develop distributed scientific algorithms independently from the execution environment architecture. OpenMOLE uses abstraction layers to delegate computing tasks with the same high level interface for the major underlying architectures: local processors, batch systems, computational grids, Internet computing and cloud computing. The file access abstraction layer is another key feature helping a generic usage of the computation power provided by grids and clusters. The OpenMOLE framework has been tested with the exploration of a bacterial biofilm simulation with an individual-based model.

EasyABC: performing efficient approximate Bayesian computation sampling schemes using R

Summary Approximate Bayesian computation (ABC), a type of likelihood-free inference, is a family of statistical techniques to perform parameter estimation and model selection. It is increasingly used in ecology and evolution, where the models used can be too complex to be handled with standard likelihood techniques. The essence of ABC techniques is to compare simulation outputs to observed data, in order to select the parameter values of the simulations which best fit the data. ABC techniques are thus computationally demanding. This constitutes a key limitation to their implementation. We introduce the R package ‘EasyABC’ that enables one to launch a series of simulations from the R platform and to retrieve the simulation outputs in an appropriate format for post-processing. The ‘EasyABC’ package further implements several efficient parameter sampling schemes to speed up the ABC procedure: on top of the standard prior sampling, it implements various algorithms to perform sequential (ABC-sequential) and Markov chain Monte Carlo (ABC-MCMC) sampling schemes. The package functions can furthermore make use of parallel computing. The R package ‘EasyABC’ complements the package ‘abc’ which enables various post-processing of simulation outputs. ‘EasyABC’ makes several state-of-the-art ABC implementations available to the large community of R users in the fields of ecology and evolution. It is a freely available R package under the GPL license, and it can be downloaded at http://cran.r-project.org/web/packages/EasyABC/index.html.

How to conduct a proper sensitivity analysis in life cycle assessment: taking into account correlations within LCI data and interactions within the LCA calculation model.

Sensitivity analysis (SA) is a significant tool for studying the robustness of results and their sensitivity to uncertainty factors in life cycle assessment (LCA). It highlights the most important set of model parameters to determine whether data quality needs to be improved, and to enhance interpretation of results. Interactions within the LCA calculation model and correlations within Life Cycle Inventory (LCI) input parameters are two main issues among the LCA calculation process. Here we propose a methodology for conducting a proper SA which takes into account the effects of these two issues. This study first presents the SA in an uncorrelated case, comparing local and independent global sensitivity analysis. Independent global sensitivity analysis aims to analyze the variability of results because of the variation of input parameters over the whole domain of uncertainty, together with interactions among input parameters. We then apply a dependent global sensitivity approach that makes minor modifications to traditional Sobol indices to address the correlation issue. Finally, we propose some guidelines for choosing the appropriate SA method depending on the characteristics of the model and the goals of the study. Our results clearly show that the choice of sensitivity methods should be made according to the magnitude of uncertainty and the degree of correlation.

A Multidimensional Model for Data Warehouses of Simulation Results

This paper examines the multidimensional modeling of a data warehouse for simulation results. Environmental dynamics modeling is used to study complex scenarios like urbanization, climate change and deforestation while allowing decision makers to understand and predict the evolution of the environment in response to potential value changes in a large number of influence variables. In this context, exploring simulation models produces a huge volume of data, which must often be studied extensively at different levels of aggregation due to there being a great need to define tools and methodologies specifically adapted for the storage and analysis of such complex data. Data warehousing systems provide technologies for managing simulation results from different sources. Moreover, OLAP technologies allow one to analyze and compare these results and their corresponding models. In this paper, the authors propose a generic multidimensional schema to analyze the results of a simulation model, which can guide modelers in designing specific data warehouses, and an adaptation of an OLAP client tool to provide an adequate visualization of data. As an example, a data warehouse for the analysis of results produced from a savanna simulation model is implemented using a Relational OLAP architecture.

SimExplorer: Programming Experimental Designs on Models and Managing Quality of Modelling Process

This article describes Simexplorer, a computer framework for managing simulation experiments and, to some extent, the scientific quality of the modelling process. An information system, included in the framework, insures the traceability of the experiments and their reproducibility and thus contributes to the modelling process quality management. Moreover, this information system provides facilities for sharing and exchanging components of experiment scenarios. The authors illustrate the use of the framework on a simple example of modelling process.

The Combined Use of Correlative and Mechanistic Species Distribution Models Benefits Low Conservation Status Species.

Species can respond to climate change by tracking appropriate environmental conditions in space, resulting in a range shift. Species Distribution Models (SDMs) can help forecast such range shift responses. For few species, both correlative and mechanistic SDMs were built, but allis shad (Alosa alosa), an endangered anadromous fish species, is one of them. The main purpose of this study was to provide a framework for joint analyses of correlative and mechanistic SDMs projections in order to strengthen conservation measures for species of conservation concern. Guidelines for joint representation and subsequent interpretation of models outputs were defined and applied. The present joint analysis was based on the novel mechanistic model GR3D (Global Repositioning Dynamics of Diadromous fish Distribution) which was parameterized on allis shad and then used to predict its future distribution along the European Atlantic coast under different climate change scenarios (RCP 4.5 and RCP 8.5). We then used a correlative SDM for this species to forecast its distribution across the same geographic area and under the same climate change scenarios. First, projections from correlative and mechanistic models provided congruent trends in probability of habitat suitability and population dynamics. This agreement was preferentially interpreted as referring to the species vulnerability to climate change. Climate change could not be accordingly listed as a major threat for allis shad. The congruence in predicted range limits between SDMs projections was the next point of interest. The difference, when noticed, required to deepen our understanding of the niche modelled by each approach. In this respect, the relative position of the northern range limit between the two methods strongly suggested here that a key biological process related to intraspecific variability was potentially lacking in the mechanistic SDM. Based on our knowledge, we hypothesized that local adaptations to cold temperatures deserved more attention in terms of modelling, but further in conservation planning as well.

Using the Reliability Theory for Assessing the Decision Confidence Probability for Comparative Life Cycle Assessments.

Comparative decision making process is widely used to identify which option (system, product, service, etc.) has smaller environmental footprints and for providing recommendations that help stakeholders take future decisions. However, the uncertainty problem complicates the comparison and the decision making. Probability-based decision support in LCA is a way to help stakeholders in their decision-making process. It calculates the decision confidence probability which expresses the probability of a option to have a smaller environmental impact than the one of another option. Here we apply the reliability theory to approximate the decision confidence probability. We compare the traditional Monte Carlo method with a reliability method called FORM method. The Monte Carlo method needs high computational time to calculate the decision confidence probability. The FORM method enables us to approximate the decision confidence probability with fewer simulations than the Monte Carlo method by approximating the response surface. Moreover, the FORM method calculates the associated importance factors that correspond to a sensitivity analysis in relation to the probability. The importance factors allow stakeholders to determine which factors influence their decision. Our results clearly show that the reliability method provides additional useful information to stakeholders as well as it reduces the computational time.