Where are we with quantum computers?

Everybody and their neighbor is talking about quantum computers. I actually think it’s going to be either the next hype, or the one right after… and that this will come soon after GenAI falls off the hype mountain it is currently on.

There are many sources of information on quantum computers. In addition to all the press releases of companies announcing things like room-temperature qubits or new machines like this German demonstrator or funding of startups in this space, past, closer, or more recent, or even companies describing use cases, like Pasqal, or flexing about their current or recent client success stories like D-Wave Systems, I was recently pointed to a nice repository of quantum computing related research.

D-Wave engineer on the dashboard of the ‘Advantage’ quantum computer with 5,000+ qubits, where hybrid quantum applications such as those deployed by CaixaBank are executed. — https://www.caixabank.com/comunicacion/noticia/grupo-caixabank-y-d-wave-colaboran-en-innovadoras-aplicaciones-cuanticas-para-el-sector-financiero_es.html?id=43342

While all this shows an extremely active scene, and is very nice… It still feels a bit like in the 1950s and 1960s when transistors were being used to create the first solid state computers… And programming these beasts was still done in binary, with switches or punch cards… Using these machines was limited to a very small set of people with extremely fine tuned and niche skills. It’s only when languages like FORTRAN, COBOL and a few others, were invented that normal humans like you and me were able to start programming these machines to write code for their problems in a reasonable manner, and use them without having to call on scientists or people wearing white blouses to do the coding for them.

Programmers Betty Jean Jennings (left) and Fran Bilas (right) operating ENIAC’s main control panel at the Moore School of Electrical Engineering, c. 1945 (U.S. Army photo from the archives of the ARL Technical Library) — Unidentified U.S. Army photographer — Image from Historic Computer Images

This same situation has appeared several more times during history. I remember very fondly in 1989 when I started working with a small company called Inmos. They had invented a new type of processor called the Transputer. These processors were incredibly smart and powerful for the time. In particular, they had several cores, and communication links, onboard RAM… that allowed to build fancy networks of highly parallel processing. But what really set them apart was that in the CPU they had native multi-tasking, which let one code applications that could launch threads on the CPU, or talk to threads on another CPU (I had imagined and created a routing protocol and algorithm to send messages from one transputer to another on a mesh network of these puppies at the time).

Inmos T414 Transputer

Coding on the transputers using their native assembly language was barely humanly possible… only a few people knew how to do it. Inmos, fortunately, invented a whole new language called OCCAM which has very usable instructions and structures to handle this multi-tasking. You could write code that launched threads, you could synchronize them, set data that could be shared between threads, message… all that stuff… in a very (well, reasonably) easy to understand and use way!

Why am I rambling about this? Because we’re once again, at this point with quantum computers. There are machines available (D-Wave, Pasqal, IBM, Google) and simulators available when you don’t have a machine at hand. The Open Quantum Institute is even now there (hosted at CERN, created by GESDA — check out their amazing Science Breakthrough Radar) to allow researchers, and others users worldwide, to access quantum computing resources to learn and progress with their skills around these technologies.

But guess what, today, to programm these machines, you still need to have an understanding of quantum mechanics, and be able to code wave functions and elements specific to quantum properties. There are some really nice examples here for you to learn from. These are still very much tied to quantum aspects and very close to the hardware they involve… It’s what I call the assembly language of quantum computers. You still need to account for qubit errors, bit flips, specific quantum states like bell states… If you don’t have a minimum of understanding of these things, you can’t really code for a quantum computer today. Just like you need to understand how a shift register or a stack works on a CPU, to code in assembly.

Quantum circuit to create Bell state |Φ+⟩ — https://en.wikipedia.org/wiki/Bell_state

Today you either have dedicated quantum computing languages like Microsoft’s Q#, IBM’s Qiskit, or Google’s Cirq, which are really designed to implement quantum expressions that represent a problem you are trying to solve, or more traditional languages like Python, Julia, C++ and so on, which are used to call libraries of functions that have already coded some quantum specific functions to be used more simply (which, by the way, is the same for using statistical functions, or scientific functions)…

We don’t have a language that is general purpose enough, yet fully includes quantum primitives (like FORTRAN for Science, or COBOL for business, or R for statistics)…

So what will the programming language of the future, that includes quantum primitives, be like? Will we have a new FORTRAN (QUATRAN)? Will we have general languages and libraries of dedicated functions (and everybody will be able to code in Java or C or C++ and just link with quantum libraries — maybe in addition to libc we’ll have a libq or some other name)? We’ll still need dedicated languages to write those libraries in. But I can’t wait until somebody without a degree, or at least a training, in quantum mechanics can program a quantum computer to express the problem they need to solve and execute it with a simple vi — make — ./xxx sequence (yeah, I still use vi).