Google Releases Cirq 1.0 For Quantum Programming Framework

[phoronix]

Phoronix: Google Releases Cirq 1.0 For Quantum Programming Framework

Google engineers have released Cirq 1.0 as their first full version -- and stable API -- of this open-source programming framework for quantum computing and written in Python...

https://www.phoronix.com/scan.php?page=news_item&px=Google-Cirq-1.0

[sabriah]

What made them choose Python here? What benefits would it have over other, alternative languages?

[mb_q]

What made them choose Python here? What benefits would it have over other, alternative languages?

It's for hype-driven people who can't program (;

[schmidtbag]

What made them choose Python here? What benefits would it have over other, alternative languages?

It's fast and simple to implement. To my understanding, the computer running it isn't doing much of the heavy lifting, so the downsides of Python (namely performance issues) are largely irrelevant.

[kpedersen]

What made them choose Python here? What benefits would it have over other, alternative languages?

Indeed. I am 99.9% certain that the first successful commercial quantum compiler will just be a C or C++ compiler with extensions, similar to OpenMP.

[zexelon]

It's for hype-driven people who can't program (;

HEY! I resemble that remark!

Also as schmidtbag said its simple to implement and lets be honest the non-programming researchers that are the only people using these things only know python anyway! All other languages are for the comp-si department.

Also this thing will never be playing crysis... I mean as far as I know the only game (and one of the only applications so far! (as in not just an algorithm)) to run on a quantum computer is battle ship ... and even on that simple game the system would generate errors after enough turns due to quantum effects they haven't figured out yet.

These machines take us back to the UNIVAC days... they are the size of a room... weigh in at tons... require extreme cooling systems and currently cant perform much better than an abacus for anything but some really esoteric algorithms.

[coder]

It's for hype-driven people who can't program (;

Python is really quite powerful, in competent hands. Though it's not my primary language, I've found it's very rewarding of programming aptitude.

It's fast and simple to implement. To my understanding, the computer running it isn't doing much of the heavy lifting, so the downsides of Python (namely performance issues) are largely irrelevant.

Yes. It's used for setup & readout. The quantum computer doesn't work like a normal Turing machine, and thus isn't running any Python.

[nagerst]

Quantum as in 4D programming on a mostly single threaded language? Huh.... How are you supposed to do quantum programming on cpu cycles in the past and future (4th dimension is time) in the present?
Tod hunter explains "quantum" best for the layman: "The stasis room creates a static field of time. See, just as X-rays can't pass through lead, time cannot penetrate a stasis field. So, although you exist, you no longer exist in time and for you time itself does not exist. You see, although you're still a mass, you are no longer an event in space-time, you are a non-event mass with a quantum probability of zero."

[Palu Macil]

It's for hype-driven people who can't program (;

I'm far more fond of Go than Python, and I've more often worked on great C# Typscript codebases than great Python codebases, but when people talk about Python being new (or they reference "hype"), it seems very strange to me considering Python was released in 1991. That's 5 years before Java and 10 before C#. I've worked with people whose Python code hygiene makes me cringe, but even those people are often some of the best data scientists and cybersecurity experts, and some Python codebases are actually very well maintained, organized, and wonderfully well typed.

[Palu Macil]

What made them choose Python here? What benefits would it have over other, alternative languages?

Python is the defacto number one language for data science. In the sub category of Machine Learning specifically, it's even more dominant. It's a lot of work to compete with the massive body of highly complicated, well researched machine learning libraries already written in C++. Since C++ is less accessible to many academics, statisticians, analysts, hobbyists, etc, the ability of Python to control C++ code allows for people to use.

Implementing something for ML related work in another language without great FFI would be a lot of work and probably have less performance since it's hard to compete with C++ on speed, and it's basically impossible to compete with C++ on talking to graphics cards (not important for quantum computing, but the libraries and communities will be the the same, and I'm sure the outputs will be used together). Rust is a good candidate language for something like that, but the community is already built upon C++ and Python, so it's hard to get much momentum outside Python for ML and data science libraries.

In the specific example here, this library is providing the footwork for some of the other libraries mentioned in the article, and those are very Python centric comminuties.