Slashdot Mirror
Algorithm Predicts New Superhard Materials
An anonymous reader writes "Researchers in New York have developed an algorithm that can predict new superhard materials — a relatively small class of compounds of which diamond is the most famous. Beyond the pluses this represents for, say, the drilling industry, the physicists claim say their computational approach can be used to think up new materials of all sorts. 'New materials with desired properties will be routinely discovered using supercomputers,' they say, 'instead of the expensive trial-and-error method that is used today.'"
Superhard materials like my penis?
or does someone have to make it...
and if they 'claim' to make it-- in say one molecule in size-- is that enough to patent?
N/T
Welcome to the Panopticon. Used to be a prison, now it's your home.
Is avoiding the trial and error step advisable? Certainly it won't always give you a material with the desired property but how many times do those 'errors' have their own unique benefits? For example the search for a synthetic rubber gave us silly putty.
No matter what it looks like, there isn't a
... but I need a more sophisicated algorithm to figure it out.
Come on, people, Rearden Metal.
why do i have nothing smart to say?
Maybe this is an easier proposition than I thought, but I'm half-way through A Novel and Efficient Synthesis of Cadaverine and it looks like getting anything accomplished in chemistry is much harder than I used to think. The hardest things I know of, like diamond or carbides, look like they have simple molecular structures so are they difficult to simulate or are there more complex substances that are similarly hard bulk materials?
I say bulk because even unobtainium is useless if the stuff falls apart in fractions of a second. Knowledge is nice, but add utility and it's so much better.
If God forks the Universe every time you roll a die, he'd better have a damned good memory.
According to TFA, they developed an "evolutionary algorithm"; that means it is still trial and error.
Everything is information; there is no fundamental difference between doing something in a computer simulation or with beakers and ovens.
Performing trial and error on a supercomputer just happens to be much faster than performing trial and error in the laboratory.
The quest for a new refrigerant gas gave us teflon...
finding new hard materials? this was done at least 15 years ago:
http://users.skynet.be/intelligence/annexe/creativity.htm
(i first read this in new scientist way back when, but their article is paywalled)
I wrote up a plan for something like this about 2 1/2 years ago and posted on my blog about 9 months ago when it became obvious to me that as cool of an idea as it was, it wasn't something I wanted to work on.
The basic idea is to take a computational chemistry package and run it through a genetic algorithm to search for suitable candidates that solve certain problems.
Better solar cells, dielectrics for supercaps, or materials with specific properties.
The physics quickly went over my head and I was never able to get funding or grants for this without a PhD.
I am glad to see this is starting to happen.
Project - Mc Lab / Magic Chemist, in a Box.
http://johnsokol.blogspot.com/2010/12/project-mc-lab-magic-chemist-in-box.html
http://thegreentank.blogspot.com/2010/12/project-mc-lab-magic-chemist-in-box.html
I am always doing that which I can not do, in order that I may learn how to do it. - Pablo Picasso
So, I work in this field (computational condensed matter physics). I was going to do a PhD with one of his competitors in the random-structure field but eventually chose another. Weirdly, like, earlier today before I saw this announced, Prof. Oganov added me on Facbeook. So, questions: a) Why did he add me? b) Did he know I've got vague connections to his field? Curiouser and curiouser.
what impact will this have on the adult entertainment industry?
will be handled by the forces controlling each dimension
Transuranic, heavy elements may not be used where there is life
medium atomic weights are available
Gold, Lead, Copper, Jet, Diamond, Radium Sapphire Silver and Steel
Sapphire and Steel have been assigned
Ah, the 1980's when Joanna Lumley was hot and David Macullum was cool.
mention of adamantium, this is slashdot right?
What complex molecule formula does it spit out for the property of superconductivity at room temperature?
Because really, if it can't find formulas for things we haven't figured out on our own yet, just how useful is it?
File under 'M' for 'Manic ranting'
It could probably figure it out eventually. I don't know if it could beat us though. But I imagine we would tell it everything we knew about superconductors vs temperature for various molecules, then it would learn what's important to make the superconducting temperature go up, then learn how to make the superconducting temperature go up, then figure out what molecules could exist, then run simulations to see what would happen in the real world. Alot of computing.
The basic idea is to take a computational chemistry package and run it through a genetic algorithm to search for suitable candidates that solve certain problems.
Here, try mine.
Chembench is a web-based computational chemistry tool, runs genetic algorithm based models (among others).
The physics were over my head, too, but that wasn't a problem. We used commercial descriptor calculation tools for a while. Now the open-source chemical descriptors provided in CDK are getting good enough to replace those.
What I'd love to see, and more in the realm of sci-fi, are materials that can somehow 'pretend' to be multiple times harder than diamond rather than actually be that hard on a purely physical level.
Hey, if a material can 'pretend' to be invisible to light then why not 'pretend' to be semi-invisible to kinetic energy or pressure?
I know it's a reach but a fun concept to try for, no?
Actually, superconductivity has not been properly modeled in quantum mechanics. There are theories about it, but proper models are Not There Yet (TM).
So, they can do hardness and such, but not superconductivity. But this is pioneering work so...
From TFA,
The suggestion that a high-pressure form of TiO2 is the hardest oxide was made by Swedish researchers in a highly-cited paper published in 2001 in Nature. However, calculations show that all possible forms of TiO2 are much softer than common corundum, Al2O3, and therefore the experimental data from 2001 has to be reconsidered. The latest experiments done at Yale University and the University of Tokyo point in the same direction.
I'm not certain who wrote this, but experimental data always trumps calculation. If calculation does not match experiment, it is calculation that is wrong. Only experimental data, as in the last sentence, can counter experimental data... Who writes these things?
What is probably meant is that the *conclusions* drawn from the experimental data from 2001 have to be reconsidered.
Because even a statement like "the hardness of AL2O3 is X" is a conclusion drawn from data and a model built on those conclusions, not actual data itself.
Knowing the right combination is always easy in hindsight. Once found, we can always awe at any solution's simplicity and elegance. But to find it is like trying to guess a PIN number. The possibilities are always exponential with every new variable. So that is why we need super computers. To say trial and error is a viable approach is highly underestimating the scale of the problem. Trial and error is for production, not discovery.
A bit before 1990 I read the same really good idea which was implemented to an extent by using computers to interpolate between known phase diagrams to find possible intermediate materials with good properties. I'm not sure who was doing it but it made it into a major journal and some mentions in the mainstream press. The specifics would be very different to what you mentioned but computers have been used to identify possible useful materials for some time.
Here's how that could be useful:
Find a pile of materials that match a likely model of high temperature superconductivity and see which which actually work and thus refine the model. Then use the better model to find more materials and maybe find something that will superconduct in a room that is not on the ISS.
This of course has been keeping computers busy for a bit over two decades and makes me wonder why the summary was written as if using computers to find likely materials is fresh news.
http://en.wikipedia.org/wiki/Computational_Materials_Science
In my experience with modeling, trial-and-error will still be needed. However, it will be greatly reduced. The computer will merely provide "better guesses" for good materials. Those guesses will have to be tested, and the results fed into the computer to improve the model.
The number of trials will go from hundreds or thousands to around ten.
One of our competitors trademarked the term "hypothesis". From now on, we will call them "boneheaded ideas".
So, when do we get our mithril and adamantium?
According to TFA, they developed an "evolutionary algorithm"; that means it is still trial and error.
Everything is information; there is no fundamental difference between doing something in a computer simulation or with beakers and ovens.
Performing trial and error on a supercomputer just happens to be much faster than performing trial and error in the laboratory.
Somewhere, a really smart German named Goedel is laughing his ass off at you. Please let us know when you can simulate emergent phenomenon like consciousness in your computer. (Here's a hint: You can't, so don't try. Just concede your error and press on with a new theory, ok?)
The benefit of this idea is that we may be able to use similar computer models to look for materials with other valuable properties. Rather than just happen upon a useful material, such as teflon, (or maybe just entertaining, such as silly putty) we can look for something that specifically suits the end purpose. No, we probably wouldn't find teflon by accident when looking for alternative refrigerants, but we might find it (or something far better) when specifically looking for a tough and slippery material.
The way I look at it is that we're moving on from semi-randomly combining things, hoping to get something that suits our purpose, to being able to determine ahead of time what will work. It's like the difference between alchemy and chemistry.
AL2O3 is corundum, which has an experimentally verified hardness (9 on the Mohs scale, making it the second hardest non-synthetic material found on earth). The molecule whose hardness is in doubt is TiO2.
File under 'M' for 'Manic ranting'
If you want to know if super hard materials exist, just ask me....Yup.
/facepalm
Here's a hint: You don't understand the point of Gödel's incompleteness theorems.