C.T.S.G. Alembic

Coding for computation based on liquid metal oscillator

Our neurons can process information with incredibly high energy efficiencies. Neurons show oscillatory behavior in the form of spike trains (burst of action potentials). In this context, there is a global effort to mimic neuron-like oscillators to explore their potential in computation. We are developing EGaIn (a type of liquid metal) droplet oscillators that show very rich dynamics suitable for data processing.

Therefore, the following research questions rise involving coding, 3D-farbication or mechanistic studies:

1. How can we control the coupling mechanisms between different droplets oscillators? This can be achieved by changing the surface chemistry/energy (with monolayers for example), type of electrolyte, or voltage sequences.
2. How can we optimize data processing, such as classification, speak recognition, or time-series prediction? This can be explored codes, or by changing delay times (in hardware).
3. EGaIn droplet oscillator’s movement is aligned with its electrical signals. How can the visual information of EGaIn droplet oscillator be used to advance identifying ideal states of oscillator for specific tasks? (Investigating this could help understanding how brain activity shifts in different states during leaning and problem-solving.)

These questions aim to explore and optimize the computation potential of EGaIn oscillators with help of real-time visualizations of the computation process. It could reveal new methods for energy-efficient and adaptable computational systems.

These questions form starting points, the project can always be tailor-made depending on your experience and interests.

Contact info:

Supervisor:
Christian A. Nijhuis, c.a.nijhuis@utwente.nl

Daily Supervisor:
​​​​​​​Yi Li, y.li-8@utwente.nl