HNHacker News
TopNewBestAskShowJobs

fastneutron

455 karma · joined May 10, 2020

Principal R&D engineer - Deeptech, nuclear energy, scientific computing, quantum technologies, artificial intelligence.

fastneutron235 -at- gmail -dot- com

submissionscomments
fastneutron··on Reasons why Fortran is still used (2021)
Assuming you’re in the US, Check out the various Department of Energy labs, NASA, basically anywhere there’s big computations being done to support science and engineering projects.
fastneutron··on To get kids into science, just do it
If you want to buy anything from Alfa Aesar, Fisher Scientific or any similar “real” supplier, you need to be a corporation or educational institution and have a purchasing account or p-card. Even if it’s distilled water or sodium chloride. Find a kid who can jump through those hoops and you’ll have someone who’d probably make a decent cofounder
fastneutron··on To get kids into science, just do it
There weren’t exactly mass casualties resulting from the chemistry sets being mentioned by the other posters in this thread either. We just got more afraid of being sued.

Snark aside, where is the middle ground? Fizzing and food coloring is interesting to a 5 year old, but that’s about all you can get as a consumer these days. There’s likely a vast gulf of squandered opportunity between that and what can genuinely be done safely and still be inspiring.

fastneutron··on To get kids into science, just do it
I believe it. In the 50s You used to be able to get chemistry sets with potassium chlorate and mail-in coupons for radioisotopes. Now it’s all the same cheap, uninspiring crap for making glue slime or vinegar/baking soda fizzes.

I had sympathetic parents with chemistry degrees, and we were able to find ways to get the cool stuff when I was a teenager, but for the vast majority, it’s no surprise that kids lose interest in STEM fields as they reach their teens.

fastneutron··on To get kids into science, just do it
To the people posting about having good chemistry sets as a kid: what were these sets and where did you/your parents get them? Where are the good sources today?

Everything I remember growing up in the 90s was pretty dumbed down at best, and if you wanted to order any of the interesting chemicals or equipment, you needed a company or school letterhead with a PO account. Things got better in the 00s when eBay was still pretty open in terms of what you could find, but that seems to have been clamped down on more recently. Fear of litigation most likely.

fastneutron··on Surrogates: Gaussian Process Modeling, Design, and Optimization
Dovetailing off some of the discussion in another post [1], I thought I’d share a useful resource I’ve been using to perform surrogate modeling of expensive numerical simulations.

[1] https://news.ycombinator.com/item?id=33135695

fastneutron··on Technical milestone reached: global earth system simulations with 1.2 km resoln
For reference: https://clima.caltech.edu/
fastneutron··on Technical milestone reached: global earth system simulations with 1.2 km resoln
It’s an example of a surrogate model. It’s an ML model trained on the output of large numerical simulations like the OP, rather doing the simulation itself.

Surrogate models are nice because they can emulate the output of the full fidelity calculation in a fraction of the runtime, but they typically are trained within a range of validity outside of which they cannot reliably extrapolate.

fastneutron··on The beginning of the Monte Carlo method (1987) [pdf]
MC tends to be more physically accurate (in the limit of a large number of particle histories) because it can simulate radiation transport in continuous space, energy, and angular distribution. Deterministic methods can be faster, but the discretization process is somewhat of a dark art because the underlying distributions can be highly nonlinear and rapidly varying. A combination of methods with varying levels of fidelity are typically used in real nuclear engineering applications, and are always referenced back to a common suite of experimental benchmarks for validation.
fastneutron··on Underhanded C Contest
I played with a few of the challenges when they were going on at the time, it’s a shame they appear to have stopped. The winning solutions were always mind-blowingly clever!
fastneutron··on Bringing back the golden days of Bell Labs
The national labs are a huge asset I wish more people knew about. You get to work on interesting, meaningful projects, are given access to all the expertise, training, and development resources you could reasonably want, and have a relatively secure job with a good work-life balance.

I often talk to graduate students in STEM fields about their career prospects, many of whom are interested in software development, but are disillusioned with the career paths academia, FAANG, and the standard tech companies have to offer. Many of them just aren't aware of what the labs have to offer when I bring it up, unless their faculty advisor is in some way involved.

fastneutron··on Traveling Wave Reactor Technology
Proliferation resistance in this case means that the level of effort needed to produce a weapon would be prohibitive to the point that it wouldn’t be possible to do without raising red flags to the international community early in the process. One need only look at recent case studies around the world to see that anyone who wants to proliferate is going to do it with the easiest available technologies.

The plutonium produced in a traveling wave reactor is burned more or less in-situ as it generated. Extracting it would be a pretty destructive operation that would render the core inoperable and be visible to safeguards organizations. Beyond that, the isotope mix in reactor grade plutonium is typically unsuitable for anything but the crudest of weapons.

fastneutron··on Ask HN: Who is hiring? (August 2022)
Naval Nuclear Laboratory | navalnuclearlab.energy.gov | Scientific Software Engineer | Full-Time | Hybrid (Pittsburgh, PA; Albany, NY) | US Citizens Only

The Naval Nuclear Laboratory develops advanced nuclear propulsion technology, provides technical support, and trains nuclear operators in support of the US Naval Nuclear Propulsion Program.

We have multiple open positions for scientific software engineers. Particular application areas include nuclear reactor physics (radiation transport) and fluid dynamics (e.g. turbulence, multiphase flow).

Full postings below:

Entry level: https://navalnuclearlab.energy.gov/job-search/job-detail/?jo...

Experienced: https://navalnuclearlab.energy.gov/job-search/job-detail/?jo...

fastneutron··on Ask HN: Is an undergrad research experience in something other than CS useful?
What year of undergrad are these students? Being in the scientific computing field, many of the undergraduate CS interns we’ve mentored have gone on to work full time in scientific, engineering, or adjacent domains. Having the domain-specific focus helps provide valuable perspective and motivation beyond just the generic developer career path, but still help develop the same core skill set.
fastneutron··on Quantum Flytrap – Shaping the Future of Quantum DevTools
Interesting work, best of luck!

In all honesty, the biggest issue I’ve have with quantum development isn’t the interface per se (everyone knows python), but the instability of the underlying APIs. Half the qiskit examples you find are all using deprecated modules, and anything off that beaten path might as well be throwaway code once the corresponding paper is published, assuming the paper had any code to begin with.

I get that this is a consequence of open source development in a bleeding edge field, but just getting stable and reproducible results is hugely important for those of us trying to do applied research in this domain.

fastneutron··on Ask HN: Heavy computation setup for Python?
Ok, so each instance takes a a "reasonable" amount of time and the whole structure optimization is taking hours? If you can get a setup with MPI and/or CUDA you may be able to cut the instance runtimes, but if the optimization is just slow to converge it won't help much. If it's the former then a higher end workstation or a small cluster could get you there. Cloud could be an option here too if you can aggregate the compute resources.

I'm not as familiar with GPAW but if you're able to share a minimal example I could try to take a look.

fastneutron··on Ask HN: Heavy computation setup for Python?
I support some users running DFT, and I have a few questions: What’s your current setup? How big are the models? Which DFT code are you running; VASP, QE, something else? How long are your runs currently taking vs. how long do you want them to be?

(Edited for an autocorrect hiccup)

fastneutron··on Ask HN: Job opportunities using applied mathematics
Anything in scientific computing and engineering software development is a good fit for applied mathematics graduates. The US national labs (DOE, NASA) are heavily involved in this field, as well as software vendors like Ansys, Siemens, etc.
fastneutron··on Ask HN: How can I learn HPC?
I think you're referring to Victor Eijkhout's book(s): https://theartofhpc.com/

These are great references too! They actually popped into to my head only after I replied! Google search was struggling to find the page for me too, I had to look up the author by name.

fastneutron··on Ask HN: How can I learn HPC?
Lawrence Livermore National Lab has some good HPC tutorials here:

https://hpc.llnl.gov/documentation/tutorials

fastneutron··on Ask HN: Who wants to collaborate?
This is interesting! Are there pre-trained GCM emulators out there if someone wants to play around with this?
fastneutron··on Ask HN: What would be an alternative to blockchain and crypto?
You might be interested in raking a look at Holochain:

https://holochain.org/

I'm by no means an expert in the technology, but it's one attempt at a blockchain alternative that I'm aware of, albeit with different philosophies regarding global consensus.

← PreviousPage 5 of 5