Full length articleFeeding an astrophysical database via distributed computing resources: The case of BaSTI
Introduction
The availability of large sets of stellar evolution models spanning a wide range of stellar masses and initial chemical compositions is a necessary prerequisite for any investigation aimed at interpreting observations of Galactic and extragalactic, resolved and unresolved stellar populations. In the last few years, thanks to the new developments in various fields of Physics, our understanding of the structure and evolution of stars improved and consequently the stellar evolution models also increased in complexity and completeness.
Pietrinferni et al. (2004) built a database of evolutionary predictions: The Bag of Stellar Tracks and Isochrones (BaSTI, see Pietrinferni et al., 2004, Pietrinferni et al., 2006, Pietrinferni et al., 2009, Pietrinferni et al., 2013, Cassisi et al., 2006, Cordier et al., 2007). The database collects the results of a large number numerical simulations based on Frascati Raphson Newton Evolutionary Code (FRANEC) (Pietrinferni et al., 2004) and it is specifically devised for population synthesis studies of resolved and unresolved stellar populations.
BaSTI relational database archives a large number of parameters such as chemical composition coverage, improvements in the model input physics and bolometric corrections/colour transformations, coverage of physical parameters, reproduction of empirical constraints, ease of interpolation and inclusion in population synthesis codes. For this reason it is considered as one of the most complete stellar libraries (Conroy, 2013, Gallart et al., 2005, Marín-Franch et al., 2010).
The Database has been designed to account the following main criteria for a reliable and homogeneous stellar evolution library:
- 1.
the input physics employed in the model computations is the most up-to-date;
- 2.
models for all initial chemical compositions are computed with the same evolutionary code and the same physical framework;
- 3.
models and isochrones reproduce a large array of empirical constraints obtained from observations of single stars and local resolved stellar populations.
- 4.
All results have to be easily available to the potential users.
Recently BaSTI has been expanded by adding new important quantities evaluated during the key points of stellar evolution to allow a fast visualisation of these phases. A particular effort has been devoted to port BaSTI to a VO-compliant environment (see Pietrinferni et al., 2014, for more details) and to develop a new simple and user-friendly web interface (Pietrinferni et al., 2014).
From a computational point of view, to populate the BaSTI database with a new isochrone it is necessary to execute a large number of FRANEC runs that are extremely demanding in terms of CPU time. New runs are necessary to update, maintain and extend the database and to keep the accuracy, homogeneity and completeness of the data. Extensions are often requested by Astronomers who need new data for their scientific activities.
To carry out this large set of simulations we need an extremely high computational power, Taffoni et al. (2010) showed that a Distributed Computing Infrastructure (DCI) can be successfully used to address this problem. However, even today’s most powerful DCI such as Grid or Cloud computing infrastructures still have limitations, especially due to the design of the user interfaces. Many sophisticated tools are command line driven and are complex to use. As a consequence, new users have to become familiar, not only with Astrophysical methods and theories, but also with the use of new codes and with the handling of complex computing resources. In order to hide this complexity, it is common to adopt Science Gateways (SG) technologies (Raicu et al., 2006, Wilkins-Diehr et al., 2008a) powered by a workflow management systems to distribute computation on various computing infrastructures (Belloum et al., 2011, Deelman et al., 2009, Barker and van Hemert, 2008, Curcin and Ghanem, 2008).
A SG or portal is defined here as a community-developed set of tools, applications, and data that are integrated via a portal customised to meet the needs of the Astronomical community. The computational processes supported by SGs are organised as scientific workflows that specify dependencies among underlying tasks for orchestrating DCI resources (such as clusters, grids or clouds). SG technologies allow the scientific research community to create a web-based working environment where researchers can concentrate on scientific problems without facing the complexities of computing, data and workflow infrastructure.
In this paper we discuss the design and development of a SG that allows to execute different types of FRANEC runs giving the Astronomers the possibility of updating the BaSTI database or to make on demand simulations of synthetic stellar evolutionary tracks. In the next section we discuss the physics inputs adopted in the updated version of FRANEC and we describe the global properties of the code. In Section 3 we present the SG technology adopted to develop a FRANEC SG that allows to execute FRANEC runs on local clusters, Grid or Cloud infrastructures. In Section 4 we present the FRANEC SG. Finally we compare this approach with the use of DCI previously adopted by Taffoni et al. (2010). We also discuss the “connection with the virtual observatory”. Final remarks and a short discussion concerning the planned developments in the context of Euro-VO will conclude the paper.
Section snippets
FRANEC evolutionary code
BaSTI has been computed by using a new version of the FRANEC evolutionary code.
FRANEC is a Fortran 77 code, that simulates the evolution of a star on the basis of a number of different physical inputs and parameters. Almost all the adopted physics inputs have been updated in the new version as well as numerical scheme for treating the nuclear burnings and the accuracy of the numerics. The nuclear reaction rates have been updated by using the NACRE compilation (Angulo et al., 1999), with the
Science gateways and workflow technology
To develop FRANEC SG it is necessary to identify a framework that simplifies the web interface design, that allows to easily configure the access to DCIs or local clusters and enables to implement user defined SFR and FIR workflows.
Scientific workflows have become an effective programming paradigm to compose complex computational modules for scientific simulations such as FRANEC. A workflow is a formal specification of a scientific process, which represents, streamlines and automates the
The distributed computing infrastructure
The gUSE environment allows to implement different gateway back-ends where actual computations are executed. In the case of FRANEC we identify two different resource: the European Grid Initiative (EGI) grid and a local computing cluster.
FRANEC science gateway
FRANEC Science Gateway5 is a grid portal that is wrapped around WS-PGRADE providing the execution of a set of FRANEC runs from a simplified web interface (see Fig. 2).
A FRANEC workflow (as depicted in Fig. 3) has been developed using WS-PGRADE native workflow language (Balasko et al., 2010). Moreover, we decided to adopt a modular architecture, each module is a workflow task that can be reused to build other workflows. We developed 7 modules as shown in
Discussion
In the last years, the BaSTI stellar evolution database has been largely used as a test-case for the integration and interoperability between VO and DCI facilities (see e.g. Taffoni et al., 2010). Nowadays BaSTI provides, not only a database of stellar evolutionary tracks, but also a set of tools that facilitate the analysis of observations of stellar populations, providing “on demand” stellar evolution predictions thanks to a SG.
To provide an efficient “on demand service” for the Astronomical
Conclusions
We have received a positive feedback from the BaSTI users, reinforcing our belief that SG technology is a good approach for offering on-demand services to Astronomers. An effective SG framework is crucial to offer an efficient, extensible and maintainable service and we think that our results show that gUSE/WS-PGRADE is a good technology to develop a workflow-based gateway.
Even if extremely successful for FRANEC, according to our experience high level of technical competence is still necessary
Acknowledgements
The research leading to these results has received funding from the European Commission’s Seventh Framework Programme (FP7/2007-2013) under grant agreement no 283481 SCI-BUS (SCIentific gateway Based User Support) and the FP7 project under contract no 312579 ER-flow (Building an European Research Community through Interoperable Workflows and Data).
References (55)
- et al.
A compilation of charged-particle induced thermonuclear reaction rates
Nucl. Phys. A
(1999) - et al.
Workflows and e-science: an overview of workflow system features and capabilities
Future Gener. Comput. Syst.
(2009) - et al.
Modeling the latency on production grids with respect to the execution context
Parallel Comput.
(2009) - et al.
Basti: an updated, advanced and vo-compliant database of stellar evolution predictions
Astron. Comput.
(2014) - et al.
Enabling grid technologies for planck space mission
Future Gener. Comput. Syst.
(2007) - et al.
Enabling scientific workflow sharing through coarse-grained interoperability
Future Gener. Comput. Syst.
(2014) - Aiftimiei, C., Aimar, A., Ceccanti, A., Cecchi, M., Meglio, A.D., Estrella, F., Fuhrmam, P., Giorgio, E., Konya, B.,...
- Balasko, A., Kozlovszky, M., Schnautigel, A., Karóckai, K., Márton, I., Strodl, T., Kacsuk, P., 2010. Converting...
- et al.
Scientific workflow: a survey and research directions
Creating gateway alliances using WS-PGRADE/gUSE