Research with Octeract Engine

Using a Solver for Research

Optimisation technology has been employed by academic researchers around the world to produce some of the most groundbreaking publications. From finance to engineering, academics have made use of this type of technology and continue to do so.

Optimisation problems, found in academic research, require computing power and solutions that researchers can actually use. With these considerations in mind, Octeract Engine fits the profile for a solver that can be used as a research tool.

How the Engine can be used for Research

Academic research often involves large problems that are nonlinear in nature. Octeract Engine is built to tackle these types of problems efficiently so that valuable research time is maximised. Here’s why:

  • The Engine can be installed on High Performance Computing (HPC) clusters at any institution. This gives you the power to tackle large research problems in less time.

  • Octeract Engine addresses the issue of providing a solution researchers can work with by offering configuration suggestions if your problem doesn’t solve out of the box.

These two features form the pillars that make Octeract Engine an ideal research tool. They focus on mitigating the main issues encountered by academics when running optimisation problems.

The power of parallel scaling gives you the ability to solve the hardest problems in less time. To demonstrate, the graph below shows the speedup in problem solving time for some QPLib problems with 20 times more computing power. With this additional power, it is clear that some of these problems are solved more than 20 times faster.

optimisation problem

With the suggestion system in place, the way MINLP problems are solved has been transformed. Octeract Engine is the only solver with this system and an online Knowledge Base which is designed to teach.

By having insight into the solving process and recommended actions, research time is no longer wasted on trying to track down an explanation as to what went wrong in solving. The user then knows which actions to take.

Furthermore, the online knowledge base offers researchers the opportunity to learn more about the algorithms and how they can be tweaked to make the problem more solvable. This type of customisation is a breakthrough for researchers who have spent time and efforts to find a way forward in solving MINLPs, often to no avail.

academic research

The image above shows an example of the configuration suggestions that the Engine displays after a MINLP problem has been run. In this case, the user is advised to switch off both USE_FBBT and USE_CP. The links offer more insight into each of these options as well as how they can be changed.

Why the Engine is Research-ready

Octeract Engine is built for academic research. Not only does it have the power of parallel scaling and a suggestion system, it:

  • can solve problems that can’t be solved otherwise (applied research)
  • saves decades of code-writing, for any type of research. Additional research is often needed to understand what went wrong when a problem didn't solve
  • enables Principal Investigators (PIs) and PhD/postdoc students to push papers out much more easily and quickly which means more funding in the long-term

This extended functionality means that the Engine is prepared to partner with academics to solve research problems and aid in the publication of research papers. Octeract Engine is used by academics in over 50 universities across the globe.

It is also installed on a number of HPC clusters at universities around the world including at Imperial College London and Arizona State University.

Was this helpful?
Please make sure you fill the comments section before sending.

Thank you for your comments.
Please contact us if you need any further support.