Lattice cryptography: A new unbreakable code

Computer science researchers are creating a new standard with lattice cryptography for a post-Moore's law world, where quantum computing is standard and traditional algebraic security methods would be easily broken.
While traditional cryptography methods have relied on algebra, lattice cryptography utilizes the geometric realm-going in different dimensions in many different directions. The result is a much more difficult to crack system that provides greater security for everyone.
Christopher Peikert, the Patrick C Fischer Development Professor of Theoretical Computer Science and associate professor of electrical engineering and computer science, has research interests including cryptography, lattices, coding theory, algorithms, and computational complexity. To learn more, visit his group's website - web.eecs.umich.edu/~cpeikert/
-----
Watch more videos from Michigan Engineering and subscribe: / michiganengineering
The University of Michigan College of Engineering is one of the world’s top engineering schools. Michigan Engineering is home to 12 highly-ranked departments, and its research budget is among the largest of any public university.
engin.umich.edu
Follow Michigan Engineering:
Twitter: / umengineering
Facebook: / michigan.engineering
Instagram: / michiganengineering
Contact Michigan Engineering:
engin.umich.edu/about/contact/
*This video is an update to a previous version that incorrectly referred to conventional cryptography as relying on the factoring of prime numbers, instead of factoring of numbers that are the product of two large primes.

[@garrettzucker2894]

People like this are the unknown heroes of our world

[@bobafeet1234]

Seriously glad to know big brains are already working on things like this. Amazing!

[@westonemmett1807]

I know I'm pretty off topic but does anybody know a good place to stream new series online ?

[@michaelbrysen2589]

@Weston Emmett flixportal :D

[@westonemmett1807]

@Michael Brysen thank you, signed up and it seems like a nice service :D Appreciate it!

[@michaelbrysen2589]

@Weston Emmett You are welcome :)

[@lancemarchetti8673]

Even a quantum computer will struggle to restore a significant byte that's been altered in a file. Nor have I seen any current LLM archive this. The power of byte order manipulation is so underated.

[@765mbd]

im looking into Zenon Network and their Block-Lattice and meta-Dag approach, thanks great video

[@dreamyrhodes]

Yeah but how does it work now?

[@dimeloklk]

can i invest on their crypto lattice blockchain?

[@philastley8040]

Maybe you will not see it much in the academic world, but I figured this back in 2014 while I was walking to my dead end job 3 miles away from my house 'educating' people of tv licences back in England in a backwards town; based on data points based on a 'birdie on a branch', being effected by gravity, weight distribution, wind and aerodynamics along other vectors. They should probably give me a ring sometime. The only reason I didn't pursue it is I am useless at Maths, even my brother who gave up chasing his PHD in his final years didn't understand what the fuck I was talking about.

[@Ludifant]

But how hard is the lattice problem, really? And are we talking unique SVP? I imagine quadtrees in 2 dimensions, octrees in 3 (or in n dimensions 2n-trees) can help? Would be my first guess for an optimizing algorithm. This strategy will not be helped by direct translation to quantum computing, I get that, but how can we be certain it cannot be helped in another way?
This is an interesting piece of film, but it raises a thousand questions for me.
A little search reveals the Kuperberg algorithm for classical computing and the "average case subset sum routine" in quantum computing (by ODED REGEV) . At first glance this seems to be very promising in being able to outperform Kuperbergs algorithm. It does not seem as clear cut as presented here. The problem that I have is that the problem seems so straightforward, it is hard to imagine it cannot be optimised in ten or twenty different ways.

[@Technoblade1704]

The fact that the problem statement itself is easily understandable is indeed a big plus, because it means that the cryptographic algorithsm are easier to understand and hence it becomes easier to reason about their security, i.e. security by clarity instead of security by obscurity. If you want to know more, Chris Peikert wrote a very good survey about lattice based cryptography, which should answer almost all questions web.eecs.umich.edu/~cpeikert/pubs/lattice-survey.pdf

[@aoe9857]

How can we be sure that there won't be someday a quantum algorithm breaking lattice-based cryptography?

[@ryanhasmanners9997]

Because quantum computers are fundamentally not able to compute that kind of thing, they’re only good in a very few specific areas

[@mikabobo5348]

we really can’t be sure. Sure quantum computers aren’t designed to break lattice cryptography but there might be some new technology that may be able to do so. It might feel impossible or unbreakable right now, but nothing is really impossible when it come to technology (apparently). Im sure people who invented the RSA cryptosystem or people who during the time of invention thought the system was unbreakable but look at what is happening right now. its hard to predict the future, just a thought …

[@kevinmc7993]

I think so all hand created encryption will be broken down by quantum computer. I see real challenges for mathematicians.

[@Ludifant]

That is an interesting statement, do you propose deep learning to find a non-hand created encryption algorythm? I have experimented briefly with having a deep learning network using the GP2 language models to generate javascript-code. And got code that actually worked after just two iterations and interesting solutions after five. That was for a simple animation-path problem, though. It was very hard to control the direction the network was taking the solution. And it arrived at a rather astonishing one, that wasn't really practical, but quite interesting. In what I think you describe, this might be a plus.