Algorithms
Recipes for complex decision-making.
That's what we do best.
Digisalix = complex algorithms
Most software relies on algorithms that are already well understood. Sometimes, though, the algorithm is the hard part. That is the core of Digisalix expertise: algorithms for complex problems.
Scheduling trains across a national network, optimising how goods are packed into limited space, or finding out how a robot should move all involve a huge number of possibilities, competing goals, and real-world constraints. Finding a good solution, quickly enough to be useful, requires an algorithm designed specifically for the problem.
That’s where Digisalix excels. We design algorithms for complex problems in the physical world and turn them into software people can actually use.
Got a hard optimisation problem?
Tell us what’s slow, expensive, or hard to decide. The first conversation costs nothing.
Designing the right algorithm
This is roughly how we approach it
- 1. Understand the environment We start with the real world around the problem: the people using the system, the surrounding computing architecture, the available data, and the constraints that have to be respected.
- 2. Define what success means The goal might be using fewer cargo pallets, finding the fastest route, or designing a supply network with guaranteed backup routes. Usually, there are several competing goals and a few annoying corner cases for good measure. We work out what matters and turn it into neat mathematics.
- 3. Build a baseline First, we build a straightforward solution that does the right thing and is fast enough to be useful. This gives us something solid against which to measure everything that follows. Depending on the problem, the first working baseline can take as little as two weeks from the start of the project.
- 4. Build and test the whole system An algorithm becomes useful when it works as part of an actual system. We build an end-to-end version that can be tested and benchmarked, including the interfaces people need to work with it.
- 5. Make it better Then we improve the quality of the solutions and the speed of the algorithm until they meet the real requirements. Throughout the process, we compare improvements against the baseline. It keeps optimisation measurable and, more importantly, honest.