We are deeply passionate about science and believe in scientific answers from first principles. Our ambitions have the potential to fundamentally change the course of human history. We employ state of the art numerical methods to solve systems of equations and simulate dynamic aspects of those systems. Our inhouse algorithms are both distributed and precise, thus removing the need to choose between speed of execution and accuracy. When analytical methods and standard implementations of numerical methods fall short, we bring advanced and scalable numerical methods to bear
We are focused on the intersection between high performance computing and scientific computing. This is achieved through a combination of efficient binaries and a distributed computing design. Our distributed computing platform is optimzed to use commodity hardware and our algorithms are distributed and paralellized. The choice to target commodity hardware eliminates the need for supercomputers and reduces operational costs
At our core we are engineers and we love solving scientific problems. Fundamentally we are focused on solving problems as oppose to selling products and services. When it comes to ambitions, we are relentless and uncompromising and we intend to share our findings with all of humanity. These principles are inspired by those of GNU, Free Software Foundation and the Open Source movement.
GNU/Linux is by far the most dominant operating system on the planet. We build on software from GNU and the Free Software Foundation and use Linux from the Debian family of distributions. This enables us to have full control of all aspects of the operating system and without binding ourselves to any particular commercial or political interests. We care about transparency and privacy and believe in the spread of free software as a tool to help facilitate these principles
In principle, we are language agnostic and view programming languages as mere tools. However, some tools are better at solving specific tasks than others and it is important to choose the right tool for the right job. Our primary programming language is Rust. This is because Rust is focused on high performance through zero cost abstractions, relative safety through a novel borrow-checker and with good support for the functional style of programming which, in turn, lends itself well to scientific computations. While relatively young, it has excellent library support and tooling and most 3rd party libraries written in C++ has language bindings to Rust. It supports both low level access to the underlying hardware as well as high level abstraction facilities
Most graphics cards have hardware support for tensor calculus. Therefore, Graphics cards are excellent performance accelerators for problem solving when the problem can be expressed in tensor form. This is true in the case of reverse-mode automatic differentiation - one of the work horses in scientific computing. This makes graphics cards good candidates for scientific computing. Open Compute Language (OpenCL) is a standard for cross-platform, parallel programming of diverse accelerators found in supercomputers, cloud servers, personal computers, mobile devices and embedded platforms. OpenCL is an easily accessible programming model with an API primarily written for C and C++ but with Rust language bindings.
This project is an effort to implement advanced numerical methods for use in solving problems numerically. We emphasize accuracy to numerical precision, generally avoiding algorithmic approximations found in many basic numerical methods. Our numerical methods implement higher orders which many basic numerical methods do not. This is because the calculations we need to perform are rather sensitive to errors and because we need to calculate in many dimensions
This project is focused on the infrastructure required to run simulations and perform large scale scientific computations. The primary output of this project is a platform for distributed high performance computing and strategies for dividing and conquering large scale scientific compuations
Beyond ambitious project with the potential to fundamentally impact the course of human history. The details of this project have been withheld - not because they are proprietary of secret, but because we do not yet have enough data to warrant publication. Rest assured that our findings will be published when we have sufficient data to justify a path to our ambitions