Related WPs WP3, WP4 / Related Tasks: T3.1, T3.2, T4.1, T4.2

In solid state physics the mean field Ising spin model is one of the most well studied NP-hard textbook examples and as such a number of implementations have already been demonstrated with a number of different algorithms and architectures. In this model, the spins are all to all coupled and have the same positive interaction strength ξi=ξ0 , which effectively means that the range of the interaction is infinite and thus doesn’t depend on the distance between the nodes of the system. We aim to be able to demonstrate a successful implementation of the model on our device with several nodes greater than 40000 and crosscheck for repeatability as well as system convergence irrespective of the initial randomized state of the spin matrix.

Related WPs: WP4 / Related Tasks: T4.2

Experimental noise acts as a detrimental factor in many optoelectronic devices. However, in recurrent algorithms and artificial neural networks for Ising problems, noise has been scientifically shown to be able to promote and in special cases even speed up the iteration procedure to the ground state and can thus be seen as a free parameter if it can be adequately controlled and tuned. The effect has been recently observed in a platform with few photonic spins and various competing interactions. Noise-tolerant settings are especially important when scaling the device to solve systems with many units by providing a mechanism to escape from local minima and thus enhance the minimization of the Hamiltonian to its global minimum. Nevertheless, the impact of noise in large-scale photonic Ising machines remains largely unexplored while the precise impact of noise depends on each problem’s particular features.

Related WPs: WP4 / Related Tasks: T4.3

Yet, there are physical systems and numeric models whose dynamics cannot be fully captured by two-body interactions, and proper descriptions of multi-body interaction are required. Some of the NP hard problems are rooted in the multi-body interactions present in the spin system. For example, the Boolean satisfiability (k-SAT) problem is NP hard when k ≥ 3, where k prescribes the order of interactions or constraints. This poses a significant computational challenge, whose complexity and volume exceed by far that of Ising problems with only two-body interaction, even for a moderate number of spins. Although a small class of many-body interaction can be decomposed into a series of two-body interactions via some recursiveor algebraic means, they, often, subject to strict constraints or require tedious error corrections. For simulating complex systems and processing data with high-order correlation, suitable Ising machines remain desirable that simultaneously support high connectivity, multi-body interaction, and a large number of spins. Towards this direction the Department of Physics of UCY (UCY/PHY) will study and develop the experimental configuration exploiting its expertise in non-linear systems and processes. The UCY/PHY partner will work towards the development and implementation of an appropriate experimental set up to include higher order, (four-body) spin interactions by exploiting the second harmonic nonlinear process.

Related WPs: WP3 / Related Tasks: T3.1

The iteration time of a spatial Ising machine is limited by three factors namely: 1) SLM response, 2) camera rate and 3) data processing. To overcome this speed limitation a two-fold approach will be followed.1) The SLM will be replaced by a Digital Micromirror Device (DMD) and 2) the control of the whole data processing and iteration procedure will be performed through a GPU processor. The use of the DMD instead of the SLM will significantly increase the iteration procedure since the DMD’s refresh rate scales up to 23kHz while the corresponding one for a state-of-the-art SLM reaches only 60Hz. Both device improvements will considerably speed up the iteration process by more than two orders of magnitude, according to our estimations, notably suppressing the iteration speed.

Related WPs: WP3 / Related Tasks: T3.2

A very crucial factor towards the development of a market-oriented device is the precise range control of its operational temperature. Random and uncontrollable fluctuations of the ambient temperature may affect the operability of the various optical and electronic components from which the Ising machine is composed and thus degrade the system’s performance and reliability. Detailed temperature dependent experiments will be conducted in a precise manner to estimate the effect of each individual optoelectronic component on the device performance. 
The goal of this study will be to elucidate the exact operating range under which the Ising machine remains unaffected, hence defining the temperature operating range for future integration of the system.

Related WPs: WP5 / Related Tasks: T5.1, T5.2, T5.3


The cloud annealing service will be designed and developed in parallel with the hardware implementation, incorporating every hardware’s mature technological advance, and serving also as an online educational portal to Ising computational machines. Offering an Ising simulator “as-a-service” will provide easy access to hardware computational advances of quantum computing facilitating the virtual access on the HoloCIM Ising machine for the execution of combinatorial problems. Through a dashboard, users will be able to define tasks for computation on the Ising machine, monitor in real-time the computation process, compare performance between the Ising service and conventional cloud offerings, and gain access to collected experiences from previous runs. The ultimate goal of this service is to speed up the scientific research and software development for new computational paradigms beyond the von Neumann architecture.