Guest essay by Eric Worrall
All those long flights to remote observatories ensure Astronomers have an outsized carbon footprint. And Astronomers keen to reduce their carbon footprints are being advised to abandon software languages like Python, in favour of compiled languages like C++.
‘We’re part of the problem.’ Astronomers confront their role in—and vulnerability to—climate change
By Daniel Clery
Oct. 7, 2020 , 11:30 AMAstronomers have a climate problem. Not only is global warming increasing the frequency of wildfires and the strength of hurricanes that physically threaten observatories, but a changing climate could mar their views by bringing higher temperatures, humidity, and turbulent air closer to their mountaintop perches. Astronomers are also adding to the climate problem themselves, with long flights to remote facilities and meetings and heavy use of energy-hungry supercomputers for cosmic simulations. “We’re part of the problem, not of the solution,” says Leo Burtscher of Leiden University.
Those concerns were cast in sharp relief by six papers published last month in Nature Astronomy. One, on the carbon costs of meetings, emerged directly from the 2019 European Astronomical Society (EAS) meeting in France, which took place during a record-breaking heatwave when temperatures exceeded 45°C. “We were sitting with no air conditioning, sweating through all these interesting talks,” Burtscher says. Discussions turned to climate change and the carbon emitted getting everyone to the meeting, and they inspired Burtscher and his colleagues to size up the meeting’s travel emissions. They added up to nearly 1900 tons of carbon dioxide (CO2) equivalent or about 1.5 tons per delegate—roughly the same as emitted by an average resident of India in a whole year.
…
Astronomers are now taking steps to reduce their carbon footprint. In another of the six studies, Simon Portegies Zwart of Leiden University calls for changes in computing strategy. Astronomers should avoid traditional computers and instead use ones that rely on more efficient graphical processor units, Zwart says, although they are harder to program. Astronomers should also abandon popular programming languages such as Python in favor of efficient compiled languages. Languages such as Fortran and C++, Zwart calculates, are more than 100 times more carbon efficient than Python because they require fewer operations. Another option, says MPIA’s Knud Jahnke, is to set up supercomputers in Iceland, with its carbon-free geothermal power and cold climate, which reduces cooling needs, or in other countries with plentiful renewable energy.
…
This month, Lick workers cleared brush and trees around the site to lessen the risk of future fires. Astronomers need to take action, too, Burtscher says. It’s a moral decision—and a practical one, he says. “We need to change in order to continue our professions.”
…
Read more: https://www.sciencemag.org/news/2020/10/we-re-part-problem-astronomers-confront-their-role-and-vulnerability-climate-change
As a software developer who has worked with many people with academic qualifications in Astronomy, Physics and Mathematics, I feel qualified to comment on the advice to Astronomers to switch to C++. Just don’t.
Woke astronomers? That’s probably why they took up astronomy in the first place 🙂
So my 30 years experience in FORTRAN programming are marketable once again?
C++ can do a lot of jobs really well. Don’t switch to save co2 emissions. Switch to save on electricity bills. Even without the warmunist power generation schemes, electricity costs money. If you can save money by using C++, do it, by all means.
I am a C++ programmer and I never think of it as saving CO2 emissions. It’s for getting results faster with a smaller power bill using less hardware.
Please don’t bash C++. It’s not C++’s fault. It just so happens that C++ has advantages which can be viewed this way.
Well said! The energy bill tells the story clearly enough.
Having developed optimizing compilers for 20 years, I need to say that Fortran’s presumption that no aliasing can occur between different arguments to a procedure gives it a compilation advantage over C++, provided that you’ve respected that (unchecked) rule.
Also, the anti-interpreter bias that started this thread has me doubled over laughing.
If you regard an interpreter as the delivery vehicle for state-of-the-art, highly-tuned, domain-specific library functions, they nearly always dance rings around naive “applications programmers” trying to write supercomputing code for the first time. Python, APL, Matlab, and others are a fine way to code number crunching routines unless you have to roll your own inner loops. Then I have to agree with the assembly coders here: dig in way way down deep and write library routines like the craft-coders of old.
Thanks for the reminder that even in intellectually robust fields like astronomy and astrophysics, around half of the the credentialed experts graduated in the bottom halves of their classes.
But who stop with C++? Why not go all the way to Fortran, like the climate modelers do, and further reduce the risk of useful results?
Or even further:
C++ for astronomers? Quite braindead idea – C++ provides too many possibilities for inexperienced programmer to shoot himself in the foot. Try to imagine some 10-20 years old software package that have been extended and “improved” by bunch of post-grads – and most of them have try to learn C++ when they implement new features required for their own thesis. Python is much safer and performance penalties are not so bad if libraries like numpy and scipy are used.
Of course, the ideal way would be using languages and programming paradigms that make making errors harder and support designing better program structures – like functional programming (e.g., with Scala) or languages with better type safety like Rust. However, it is quite unlikely that most scientist would have time or enough interest to develop proper programming skills. So, once again – python is not so bad choice.
The common misnomered ‘Carbon Footprint’ is really a ‘CO2footprint’; the only real Carbon Footprint is when someone walks the coals – I’m off to increase my Hydrogen Footprint, err have a glass of water!
A lot of £$€ are spent searching Skies for possibly likely non-existence Aliens, as my fellow UniofManchester Alumni concludes we maybe the only techno species in this galaxy at the moment; the distances are too vast & Speed of E/M rad too damn slow. Even if we find some signal it’ll take years/decades for comms; InterStellarTravel is impossible anyway IMO, as you wouldn’t be able to carry enough OnBoardSpares/Resources to travel more than InterSolarSystem unless one invents the Matter Generator as in SciFi StarTrek.
Some colleagues work at the local Jodrell observatory & vice versa, recently one came back & described the exaggerated claims made for them to obtain funding for projects such as SKA.
Not against it but they need to be realistic, why so many scientists are funded for this pure science, maybe use the money for real human problems? Resolving Poverty, cutting real pollution, reducing/solving landfill etc.
#Just Saying
That alumni being ProfBECox.
I’ve also heard/seen some ClimateAlarmists use another misnomer of ‘MethaneBudget’ !
On the subject of conservation, I seem to remember that P.J. O’Rourke on his travels was in a country where one of the locals was so obsessed with the topic that he believed there were too many vowels in their language and a few should be deleted in the interest of conservation in all things.
The astronomer just reminded me.
Much ado about nothing.
What a bunch of shite.
Are the authors looking to get an endorsement for promoting C++ or do they just think that a free programming language is just too “vulgar” for astronomy?
There are two concerns mentioned here. Air travel to remote observatories and inefficient computer languages.
Air travel: if gross air travel exists then one must ask why? Most high level astronomy is not by the human eye but by cameras or other sensors, taken automatically across the sky. Other than the maintenance team no one needs to be there. However, if someone offered you an expensed trip to an exotic location would any of us object?
Now they have captured some data, either visual or other, they need to crunch it with a computer program.
If you are not a professional programmer, why would you want to learn a complex language such as C++ with its attendant testing difficulties.
I can see why you would wish to use python to go through many iterations of your ideas to extract the signal hiding in your data.
As to people recommending the use of assembler language! I can only assume they are joking. The development of computer languages followed a distinct path to make it easy for the human to express their ideas to the ‘computer’. The human is the weak link in program development. Anything that makes it easier is welcome.
Seriously? An abacus is is even better at reducing their carbon footprint and while they’re relishing their “woke-ism” why not at the same time deal with Critical Race Theory and decolonization of astronomy too.