Near-Field RF Coupling (as compute substrate)

last updated Fri May 22 2026 00:00:00 GMT+0000 (Coordinated Universal Time)
Coherent Ising MachineFDTD-to-Hardware Calibration WallMature Foundry PositioningNear-Field …

Why hasn’t this been done before

Six reasons (descending importance):

  1. Cultural orthogonality. Antenna engineers minimise coupling. CIM researchers came from optics. Nobody crossed both.
  2. Wired-coupling lower resistance. The non-optical CIM community (Roychowdhury Berkeley, Chris Kim Minnesota, Shukla Notre Dame) all went wired electrical because SPICE-model-friendly and VNA-testable. Radiative coupling needs full-wave FDTD simulation which is computationally expensive and outside standard IC EDA flow.
  3. Reconfigurability harder for radiative. Wired J is set by switchable lumped components. Radiative J is set by fabrication geometry + per-tile resonance modulation; the inverse design problem (find varactor states such that realised J ≈ target J) is non-trivial.
  4. Adjacent RIS field optimises for radiation. Reconfigurable Intelligent Surfaces (RIS) and metamaterial antenna arrays do use varactor-tuned dipole/patch arrays, but the objective is radiation (steer a beam, shape a wavefront) not internal energy minimisation. Mapping RIS to Ising solver has not been formally published.
  5. Tooling cost. Full-wave EM simulation at array scale only became affordable at 16-tile / 64-tile in the last few years.
  6. Calibration wall. Anyone who tried hit the FDTD-to-Hardware Calibration Wall and didn’t publish the failure.

Prior art to verify in DD

Aggressive prior-art search needed in three places:

If those return clean, the “no one’s been here before” claim survives. If a 2022 Chinese paper or DARPA final report turns up doing exactly this, the IP wedge collapses fast.

Adjacent prior art (known)

Why now (technology readiness)

Sources

Related concepts

Frontier questions