Generic access to symbolic computing services
Abstract
Symbolic computation is one of the computational domains that requires large computational
resources. Computer Algebra Systems (CAS), the main tools used for symbolic
computations, are mainly designed to be used as software tools installed on standalone
machines that do not provide the required resources for solving large symbolic computation
problems. In order to support symbolic computations an infrastructure built upon
massively distributed computational environments must be developed.
Building an infrastructure for symbolic computations requires a thorough analysis of
the most important requirements raised by the symbolic computation world and must
be built based on the most suitable architectural styles and technologies. The architecture
that we propose is composed of several main components: the Computer Algebra
System (CAS) Server that exposes the functionality implemented by one or more supporting
CASs through generic interfaces of Grid Services; the Architecture for Grid
Symbolic Services Orchestration (AGSSO) Server that allows seamless composition of
CAS Server capabilities; and client side libraries to assist the users in describing workflows
for symbolic computations directly within the CAS environment. We have also
designed and developed a framework for automatic data management of mathematical
content that relies on OpenMath encoding.
To support the validation and fine tuning of the system we have developed a simulation
platform that mimics the environment on which the architecture is deployed.