Machine learning and statistics

Hatsya Limited uses modern statistical and machine learning algorithms for many of its solutions, typically applied to data analysis or computer vision. We have provided solutions and direction for organisations ranging from quantitative hedge funds to technology startups, and even a polo club.

Although we favour using modern technology, it is instructive to see that deep learning can be performed on even the most restrictive of architectures. One of our researchers demonstrated that Charles Babbage’s Analytical Engine is capable of performing deep learning using a convolutional neural network.

Quantum computation

We are currently working on a compiler to design efficient quantum circuits using the Solovay-Kitaev algorithm in conjunction with a powerful GPU-assisted search. This utilises the latest GPU technology, namely NVIDIATM Tensor Cores, to perform blazingly fast linear algebra.

Artificial life

The field of artificial life arguably began with John von Neumann’s invention of the cellular automaton. Several of our projects are concerned with artificial life, computational evolution, and cellular automata:

  • Collaboratively with Tim Hutton of Microsoft Research, we have made a lattice chemistry simulator for running custom biochemistries on a CUDA-enabled GPU.
  • Hatsya Limited sponsors a popular distributed search for emergent complexity in cellular automata, created by our founder in 2015. It is fuelled by an open-source search program which now supports several new families of cellular automata.
  • Using SAT solvers, our founder collaborated with Tomas Rokicki and discovered a new cellular automaton configuration which had remained elusive since its existence was pondered by mathematicians 48 years ago.