The Nobel Prize in Chemistry 2016 | Hacker News Reader

The Nobel Prize in Chemistry 2016

nobelprize.org

116 pointsdnetesn19 comments

I am reaching Feynman's "pleasure of finding things out". Gotta say his nanotechnology lecture was ver inspiring. I'm glad we've made much progress on it. I hope this ends up creating tiny super bots some day. The kind surgeons can "hangout" inside a body and perform a delicate surgery.

That was a great read, so many different kind of small "machines"! I think for the first time I feel hopeful about the relatively "near" future (next 50 years) of nanotech.

They're not really machines, as a machine is defined as a contraption that converts one form of energy to another (e.g. a lever), these are just molecules that look like machines but have no handles with which to use as the analogous machine.

Converting UV light into rotary motion doesn't count?

It definitely does, by any reasonable definition of "machinery", which is why "cellular machinery" is a term in fact.

I am super proud as I had lectures with Ben at University of Groningen whilst doing master and PhD courses. Super cool guy, straight to the point.

What about Drexler, Merkle and Freitas?

I wonder if software could assist progress here.

Anyone know if tools exist for modeling chemical behavior on a level that would be useful for simulating the assembly of 'machines' like in the article?

I see these researchers are jumping up a level of abstraction, thinking in terms of new primitives (e.g. 'rotaxanes')—maybe the software could begin its modeling in terms of rules obeyed by these new primitives (rather than starting from quantum mechanics or whatever).

Edit: added italics.

Computational chemistry (and/or computational solid state physics) is a field with 50+ years of history and a ton of exceptionally active research groups.

Your most likely starting point is density functional theory; https://en.wikipedia.org/wiki/Density_functional_theory, https://en.wikipedia.org/wiki/Kohn–Sham_equations. Classical molecular mechanics is even older than that.

Thanks! The article on density functional theory is very interesting so far.

Any idea on the second part of my question, though? Whether it might be possible to explicitly model the behavior of these higher-level constructs, e.g. modeling catenane interactions rather than molecular interactions.

what you're describing is essentially what people try to do with molecular dynamics and kinetic Monte Carlo methods, which are even older methods than density functional theory. There's currently a lot of interest in the materials community in developing ways to link these abstractions -- often highly abstracted models just aren't good enough for quantitative results, because they are difficult to parameterize, or they abstract away important physics.

https://en.wikipedia.org/wiki/Multiscale_modeling

Perfect! That's exactly the kind of thing I was thinking. Sounds like the different scales aren't really 'sealed off' as much as I was hoping, though, and that underlying details are still critical to doing actual work in the field :/ Ah, well.

My understanding is that a lot of the molecular interactions involved become exponentially complex to model using classical computers --- so this would be one of the major new areas of exploration with quantum computers.

People I've spoken with in the industry are excited about simulating fertilizer synthesis processes (which are super energy inefficient the way we do it at scale, but are theoretically much cheaper to perform, since anaerobic bacteria do it in the soil at normal temperature and pressure)

And also excited about simulating battery design -- being able to test millions of molecules without actually having to make a battery.

Of course, these are the early-stage speculative things, I'm sure more fancy nanotech could be explored with quantum computers as well.

http://blogs.microsoft.com/next/2015/11/13/with-quantum-comp...

Part of why I was interested in the idea is that it seems like what's being done in the article deals with vastly smaller numbers of molecules than what traditional chemical simulations would be targeted at.

Some of the discussed 'machines' are comprised of just a few molecules, and I imagine the more complex ones are still under 1000. So as long as the software was specialized for this application, it seems like it would be feasible on classical computers.

Software can probably assist here, at least that is the assertion of the research group I am working in.

Simulating the assembly of molecular machines is very hard, more so than simulating the machines themselves. Chemical reactions are hard to simulate because we don't have a good way to model atoms, phenomenally large amounts of atoms are involved in chemical reactions, and chemical reactions take incredibly long amounts of time relative to what the individual atoms are doing. The last two make it so that events that are incredibly rare relative to individual atoms occur often enough in the real world to be of significant importance.

Quantum computers might fix the first issue, the last two we'll need to get clever. And indeed we have some clever approaches for dealing with these two issues, but there is still work that needs to be done.

Thinking in primitives isn't that useful. For years we have know that if we could daisy chain rotaxanes together we could make amazing artificial muscles. We get greater strains and relative forces than real muscles, but we have yet to realize such things. One can draw a structure with interlocking loops that could be daisy chained, but when we try to make in the real world, it ends up being huge(synthetically) and doesn't necessarily move in ways we thought it would.

Fascinating work. Now I wish I had continued on my PhD in chemistry 35 years ago. Oh well... this field would have been close to what I would have worked on, or at least enough that it might have attracted me. Building things no one else ever built is pretty inspiring.

Great stuff. Sadly, we're still a really far away from Drexler's vision of nanotechnology.

Reply on news.ycombinator.com