blog1_img

A Virtualized Holographic Cloud Ising Machine

Blog post 1. UBITECH, 01/01/2024

HoloCIM behind the Scenes

The modern world is increasingly data and information-driven, while the growth of our interconnected societies is underpinned by the continuous improvement of digital computational technologies and architectures. This comprehensive digitization will allow for the optimization of energy and resource usage, helping to alleviate the strain on planetary resources. The ever-growing number of variables and constraints in real-life optimization problems, however, requires improved computational efficiency, that conventional von Neumann architectures struggle to provide as they are reaching their scalability and power efficiency limits. Beyond traditional computing platforms, several optical classical and quantum spin simulators have been proposed as alternatives and are being actively developed, promising superior speeds along with robust reconfigurability, low power consumption, and hardwired parallel processing. These systems aim to provide faster and cost-effective computational solutions for specialised problem classes and cases of data-intensive (big-data) tasks and optimizations. Optical annealer devices map a system observable to a minimizor variable of a given problem and allow the system to naturally evolve towards the minimum of the mapped cost function. In this case, the inherently fast dynamics of these systems dictate the time to the solution which can be orders of magnitude faster than classical computers, especially for large numbers of optimization elements. Effectively such systems are analogue machines that bootstrap the underlying physics of well-studied physical systems which can be efficiently controlled by employing mature advances of optoelectronic technologies.

In the context of HoloCIM, a novel photonic Ising Machine is designed and developed, that utilizes and enhances newly established holographic and nonlinear photonics principles for the efficient solution of combinatorial optimization problems (COPs), which still cannot be addressed with modern digital computer architectures. The main objectives of the HoloCIM are: 

  1. the augmentation and enhancement of the capabilities of a Spatial Holographic Ising Machine to be able to solve combinatorial optimization problems (COPs), 
  2. support a number of nodes comparable to the current scientific state of the art (N=8 •104)
  3. incorporate faster holographic devices based on digital micromirror devices (DMD) and GPU processor that can in principle reduce, the current system’s processing time by more than two orders of magnitude
  4. implement a dedicated cloud interface for the commercial and scientific exploitation of the system’s baseline capabilities, 
  5. expand the device’s capabilities to accommodate, plug-in additional nonlinear optical elements, enabling higher order spin-spin interactions
  6. create an online user interface to be scientifically and commercially used from interested parties such as research institutions, academia, and industry.

From a technological point of view, HoloCIM proposes an alternative approach to existing photonic simulators exploiting the mature technology of spatial light modulation. The latter introduces a range of advantages that mitigate systemic bottlenecks associated with the scalability and applicability of these devices, with the most pronounced of these being: i) cost effective, ii) easily programmable, iii) environmentally friendly, low power consumption, iv) scalability, v) non-cryogenic operation. We expect that the advances brought forward from the implementation of the project will further inform the photonic technologies community on the use of spatial light modulation systems in order to control, prepare and manipulate light states and can help to further develop hybrid technologies in the emerging fields of quantum communication, quantum computing and quantum sensing.

This blog post is also available in pdf.