In a world that is becoming increasingly complex, the Network-Enabled Optimization Server (NEOS), a leading provider of free internet-based service for solving numerical optimization problems, has been able to add features while making computing simpler for scientists and citizens alike. NEOS is based primarily at the University of Wisconsin–Madison and administered by the Core Computation Technology Group at the Morgridge Institute for Research (MIR) and the Optimization Theme at the Wisconsin Institute for Discovery (WID).
NEOS, accessed by more than 50,000 people in over 50 countries, uses distributed computing to find solutions for anything from simple Sudoku puzzles to complex Bioengineering and Image Processing problems. Within the past year NEOS has taken steps to make the interface more resilient, sophisticated and diverse.
To become more resilient and handle the increasing number of jobs that are submitted into the system, an HTCondor™ infrastructure was employed. The underlying servers were enhanced through the addition of new computational resources and improved software engineering design. Users are now enjoying more feedback and uptime.
“New and enhanced features enable users to experiment with a broad class of state-of-the-art optimization codes.”
New solvers including CPLEX and BARON added new capabilities to the set of solvers and new interfaces now enhance the interaction with the system. According to Rosemary Berger, NEOS optimization and support specialist, “The new and enhanced features enable users to experiment with a broad class of state-of-the-art optimization codes, to write models in a convenient way for prototyping solution mechanisms and to exploit computational resources on a large scale to enhance the impact of optimization on science, engineering and society.” Users may now test their models on multiple solvers to compare results.
A diverse selection of case studies is available on the NEOS Guide website. The site illustrates the breadth of applications the service provides. Most of the case studies are interactive and appeal to anyone from seasoned optimizers to curious citizen scientists dabbling with logic puzzles.
The increase in users and jobs run within the past six months are a testament to the usability of the system. It also pointed out an opportunity for the newest enhancement for a streamlined experience.
Collaborating with MIR’s Software Assurance Marketplace (SWAMP), NEOS developed secure user authentication. As of its introduction on March 9, 2016, users have the option to sign in with their email address. This virtual “fast pass” will allow those clients’ jobs to be prioritized on the NEOS servers. Previously, jobs were submitted anonymously and run as servers became available. Clients are not required to authenticate, but the NEOS team is looking forward to offering their authenticated users more services and better customized experiences. It will also help NEOS track who and what kind of entities they are serving. By incorporating SWAMP’s expertise, NEOS and their users are assured of ongoing data security.
Whether researchers want to run iterations on a single problem or try solvers before making a commitment, NEOS makes sense and stays current for its continuously growing client base.