Excited that some nice new work led by @riddhisw from my team @ARC_EQUS @Sydney_Science has just been published!

This is about how to develop optimal architectures for the integration of sensors into #quantumcomputing used to stabilize the hardware! https://link.aps.org/doi/10.1103/PhysRevA.102.042611

This was very cool work, developed quite independently by @riddhisw and our collaborator @Luke_Govia at BBN.

Let me tell you about it

We considered how to embed special "spectator qubits" into an underlying grid in a #quantumcomputer. The objective is to use these special qubits as sensors to locally detect what's happening in the hardware and then stabilize the other qubits

We already learned how to perform #machinelearning inspired data inference to take the signals from these sensors and combine them to learn about background noise.

https://www.nature.com/articles/s41534-020-0286-0

Our next question was if the physical layout could be modified to improve the inference

The answer is yes! You can choose to locate the sensors on so-called "Padua points" within the #quantumcomputer architecture to enable optimal interpolation of continuous fields in 2D. This lets you measure the spectators & best estimate what's happening at the data qubits.

The Padua points are very cool and if you look closely you'll see they are quite similar to the @MITLL logo!

More importantly they mathematically enable provably optimal Lagrange interpolation in 2D. This is useful in case, e.g. a background B field varies in space.

. @riddhisw performed a wide variety of numerical simulations to determine that in general Padua-based interpolation structures would outperform just putting the spectator qubits on the regular 2D grid of the data qubits.

We found the optimality of interpolation only persisted when the field being estimated met certain "smoothness" conditions.

What to do? NMQA to the rescue!

NMQA (an adaptive filter @riddhisw designed) always works and is insensitive to grid arch. https://www.nature.com/articles/s41534-020-0286-0

The summary message is that it's always wise to orient spectator qubit sensors on a Padua grid within a #quantumcomputer. If the field being estimated is sufficiently smooth, you can perform optimal estimation. If not, you can use NMQA without loss of performance!

#Paduarocks!