This post was first sent to my newsletter on July 23rd, 2021.
You really ought to subscribe :)
New month, new work letter.
Before I begin though, please hit reply to these letters and let me know what you think, whether you like them or not and if there is anything you’d want me to write / hunt & research about!
Like I wrote in the introduction to April’s letter, both the letter and the website have zero tracking.
So I have no way of knowing, whether you like stuff or not, whether it resonates or not, whether I am preaching to throngs or shouting into the void, unless you tell me!
Let’s dig into it. As usual, click the headers, to wander off to the orginal articles :)
Excuse me a moment—I am going to be bombastic, over excited, and possibly annoying. The race is run, and we have a winner in the future of quantum computing. IBM, Google, and everyone else can turn in their quantum computing cards and take up knitting.
One key to quantum computing (or any computation, really) is the ability to change a qubit’s state depending on the state of another qubit. This turned out to be doable but cumbersome in optical quantum computing. Typically, a two-(or more) qubit operation is a nonlinear operation, and optical nonlinear processes are very inefficient. Linear two-qubit operations are possible, but they are probabilistic, so you need to repeat your calculation many times to be sure you know which answer is correct. A second critical feature is programmability. It is not desirable to have to create a new computer for every computation you wish to perform. Here, optical quantum computers really seemed to fall down. An optical quantum computer could be easy to set up and measure, or it could be programmable—but not both.
So, what has changed to suddenly make optical quantum computers viable? One is the appearance of detectors that can resolve the number of photons they receive. A second key development was integrated optical circuits. performance has gotten much, much better. Integrated optics are now commonly used in the telecommunications industry, with the scale and reliability that that implies.
The researchers, from a startup called Xanadu and the National Institute of Standards, have pulled together these technology developments to produce a single integrated optical chip that generates eight qubits. The internal setting of the interferometer is the knob that the programmer uses to control the computation. In practice, the knob just changes the temperature of individual waveguide segments. But the programmer doesn’t have to worry about these details. Instead, they have an application programming interface (Strawberry Fields Python Library) that takes very normal-looking Python code. This code is then translated by a control system that maintains the correct temperature differentials on the chip,
What is more, the scaling does not present huge amounts of increased complexity. In superconducting qubits, each qubit is a current loop in a magnetic field. Each qubit generates a field that talks to all the other qubits all the time. Engineers have to take a great deal of trouble to decouple and couple qubits from each other at the right moment. The larger the system, the trickier that task becomes. Ion qubit computers face an analogous problem in their trap modes. There isn’t really an analogous problem in optical systems, and that is their key advantage.
When considering data privacy and protections, there is no data more important than personal data, whether that’s medical, financial, or even social. The discussions around access to our data, or even our metadata, becomes about who knows what, and if my personal data is safe. Today’s announcement between Intel, Microsoft, and DARPA, is a program designed around keeping information safe and encrypted, but still using that data to build better models or provide better statistical analysis without disclosing the actual data. It’s called Fully Homomorphic Encryption, but it is so computationally intense that the concept is almost useless in practice.
So whether that means combining hospital medical records over a state, or customizing a personal service using personal metadata gathered on a user’s smartphone, FHE at that scale is no longer a viable solution. Enter the DARPA DPRIVE program.
DARPA: Defense Advanced Research Projects Agency
DPRIVE: Data Protection in Virtual Environments
Intel has announced that as part of the DPRIVE program, it has signed an agreement with DARPA to develop custom IP leading to silicon to enable faster FHE in the cloud, specifically with Microsoft on both Azure and JEDI cloud, initially with the US government. As part of this multi-year project, expertise from Intel Labs, Intel’s Design Engineering, and Intel’s Data Platforms Group will come together to create a dedicated ASIC to reduce the computational overhead of FHE over existing CPU-based methods. The press release states that the target is to reduce processing time by five orders of magnitude from current methods, reducing compute times from days to minutes.
No wonder we’re a bit dizzy. We just multiplied our minds by many orders of magnitude. It’s easy to confuse someone else’s memory (or manipulation) with our hard-earned ability to remember things that actually happened to us.
And we’re now realizing that we have the power (and perhaps the obligation) to use shared knowledge to make better, more thoughtful decisions. And to intentionally edit out the manipulations and falsehoods that are designed to spread, not to improve our lives.
Until next time …
Don’t forget to write! :)