Open Standard Codes and Routines (OSCAR) is an initiative started in 1997. Its main goal is to provide standardized, documented, and tested open-source numerical tools and packages - primarily for heavy-ion physics applications. For most people, OSCAR is synonymous with its code repository, which contains parton and hadron transport packages, hydrodynamic codes, classical Yang-Mills solvers, and other useful tools. Another key component of OSCAR is the standardization of output formats for heavy-ion physics codes.
OSCAR was spear-headed by Yang Pang and Miklos Gyulassy at Columbia University, with strong backing from Tsung-Dao ("Tee-Dee") Lee. It was kick-started with the first OSCAR workshop in June 1997, organized at the newly formed RIKEN BNL Research Center (RBRC) at Brookhaven National Laboratory. Yang single-handedly created a website and the first repository, which were served from his Linux desktop machine 'rhic' at Columbia. The next few years brought a lot of activity with two more OSCAR workshops (June 1999 at RBRC, and June 2000 at Subatech/Nantes) and working groups that defined standard output formats for transport and ideal hydrodynamic codes.
With Yang's departure* from Columbia the enthusiasm lessened but OSCAR lived on, nevertheless. When Yang's machine was decommissioned in May 2000, the OSCAR site was transferred to the Columbia Nuclear Theory Group, where it was maintained by Denes Molnar. New packages were added such as the first open-source 2+1D ideal hydrodynamic code and the AMPT transport model.
In June 2008, OSCAR moved to Purdue University, where it is being maintained by the High-Energy Nuclear Theory Group. The code repository continues expanding further, most recently with open-source codes that provide dissipative hydrodynamic solutions and classical gluon field evolution (Yang-Mills).
Why OSCAR is needed
Heavy-ion physics involves complex numerical models. Because of the many phenomenological ingredients involved, there is a danger that models become essentially black boxes defined by their source code - or as Yang coined astutely
This makes it very difficult, short of deciphering thousands of lines of source code, to assess and test the physics assumptions inherent in a model. Moreover, as our understanding improves and/or more experimental data become available, models also evolve, often silently. The identification of successes and shortcomings of a given model, or even the reproduction of published results, can therefore be plagued by ambiguities about which version or variant of the model was used.
The paradigm shift OSCAR proposed was
Namely, codes are numerical techniques that are first tested and validated without any reference to data. Once a code is demonstrated to produce reliable numerical solutions, its model assumptions (the equations) can be tested against real experimental results. Another corner-stone of OSCAR is version control. OSCAR serves as a repository of code packages. Each package has a distinct version number, which can uniquely identify in publications the variant of the model that has been utilized. Because the ultimate definition of a numerical package is inevitably the source code, OSCAR also stands for accessibility of the source codes, together with documentation. Finally, OSCAR strives for standardization, primarily for input and output formats to make it easy to interface different packages. OSCAR also encourages an open-source, modular design in code development so that components can be freely reused across packages.
* Yang eventually left physics and moved to finance.