Sam Jaques

699 posts

Sam Jaques

Sam Jaques

@sejaques

Assistant prof at U Waterloo. Loves maths, maps, algorithms, and flashcards. Aspiring full-stack cryptographer. He/him. @[email protected]

Waterloo, Canada Katılım Kasım 2014
548 Takip Edilen877 Takipçiler
Bas Westerbaan
Bas Westerbaan@bwesterb·
We need to track both the performance of quantum computers and the improvements of algorithms. It’s quite hard to estimate the actual performance of an algorithm for a computer that’s not designed yet, but that doesn’t mean the goalpost isn’t getting closer and closer every day.
JP Aumasson@veorq

eprint.iacr.org/2024/222.pdf (logical qubits count excluding error-correction)

English
2
9
28
5.8K
Sam Jaques retweetledi
Michael Newman
Michael Newman@MikeNewmQuantum·
Houston, we are below the quantum error correction threshold! 🚀 In “Quantum error correction below the surface code threshold” (arxiv.org/abs/2408.13687), we implement a 101-qubit surface code. Each time we increase the distance by two, the logical error rate is cut in half!
Michael Newman tweet media
English
14
132
598
119.7K
Sam Jaques
Sam Jaques@sejaques·
It's actually a reasonable heuristic to expect that chemicals that life had millions of years to adapt to will be safer than totally new ones. Is it always right? Of course not. But it's not ignorant.
English
0
0
2
197
Sam Jaques
Sam Jaques@sejaques·
Pet peeve: scientists who say "oh you don't like 'chemicals'? Well water is a chemical! Dihydrogen monoxide!" You know damn well that "chemicals" means "chemicals that never/rarely existed on Earth until humans synthesized them at scale"
English
1
0
2
283
Craig Gidney
Craig Gidney@CraigGidney·
@sejaques I don't think that case was tried. Honestly factories is where the tool is weakest, because it doesn't model reducing distance due to protection from the distillation (or TELS). Also can't search over Clifford correction strategies. You want that stuff optimized for a good CCZ.
English
1
0
1
61
Craig Gidney
Craig Gidney@CraigGidney·
I hosted a student researcher Daniel Tan for a few months. He wrote a tool for verifying and generating lattice surgery constructions: arxiv.org/abs/2404.18369 It can generate some wild constructions; smaller than I was able to make by hand. I've learned some tricks from it.
Craig Gidney tweet media
English
3
4
57
3.8K
Sam Jaques
Sam Jaques@sejaques·
Silly idea: university course catalogs should include "alpha" and "beta" for the course: the slope and intercept of a linear regression between grades in that course and a student's overall average
English
0
0
3
276
Sam Jaques
Sam Jaques@sejaques·
@CraigGidney Did you optimize CCZ factories automatically with that tool? I couldn't find one in the example files.
English
1
0
0
58
Craig Gidney
Craig Gidney@CraigGidney·
Eg. I learned to try switching init basis. In arxiv.org/abs/1812.01238 I surgery'd a 8T→CCZ factory. The encoding measurements were X basis, and T gates are Z basis, so I needed half-distance rotations of the qubits. Switching init basis avoids that! Saves time *and* distance.
Craig Gidney tweet media
English
2
1
2
1.1K
Sam Jaques
Sam Jaques@sejaques·
@itranneo @hashbreaker How familiar are you with lattice security generally? Larger q => easier to break the underlying lattice problem(s). If q is exponentially large I think it's totally broken.
English
1
0
2
45
AntoineR
AntoineR@itranneo·
Looking into NTRU for the 1st time... - are there any security gotchas with the selection of q other than the fact that is need to be much much bigger than (and coprime with) 3? Can it be set to a ridiculously large prime number without impacting security? @hashbreaker
English
2
0
3
211
Sam Jaques
Sam Jaques@sejaques·
Bless whoever added this photo to the "All horses are the same colour" paradox Wikipedia page
Sam Jaques tweet media
English
0
0
2
240
Sam Jaques
Sam Jaques@sejaques·
Singapore will make new orchid hybrid varieties and name them after visiting dignitaries. I guess you could call that... soft flower diplomacy 🥁🥁💥
Sam Jaques tweet media
English
0
0
3
231
Sam Jaques
Sam Jaques@sejaques·
Me to a colleague: I think it's fine if students address us informally in email, the new generation has different norms Me when a student starts an email with "Hi Sammy": the lord is testing me
English
1
1
14
952
Sam Jaques retweetledi
Tanja Lange
Tanja Lange@hyperelliptic·
To those in (a timezone compatible with) Pacific Daylight Time:: I will be speaking at the @Visa Quantum day on 21 May along with Sam Jaques (@sejaques ), Renato Renner, & Brian Coyle (@BrianC2095 ). Registration is open and free at teams.microsoft.com/registration/E… (sorry, MS Teams link).
English
1
3
11
1.7K
Sam Jaques
Sam Jaques@sejaques·
8 hour train ride, fresh mug of coffee, let's goooo
Sam Jaques tweet media
English
3
2
67
10.5K
Sam Jaques
Sam Jaques@sejaques·
@jjaron Hm, I wonder if a log plot would match wealth intuition better
English
0
0
0
145
Jacob Aron
Jacob Aron@jjaron·
This is obviously an important issue, but I think the fact that the average millennial has essentially *no* housing wealth, while the average boomer does, will remain a bigger dividing line. The diff between renting and owning is much bigger than between £100k/£1m house
English
3
2
32
8.9K
Sam Jaques
Sam Jaques@sejaques·
Kind of amazing that Encrochat was used for basically the exact same thing (though the app was originally not backdoored). I guess criminals really want dedicated apps made just for criminals?
Sam Jaques@sejaques

LOL: "Fake messaging app that actually just BCC's all your messages to the FBI" is a real Gordian knot solution to encryption backdoors. (Though, I worry of the long con: if people think services are already backdoored, it will be easier to pass backdoor legislation)

English
0
0
2
1.3K
Sam Jaques
Sam Jaques@sejaques·
@bwesterb I'm being a bit facetious. If I had 2000 engineers it would make sense to assign one to work on efficiency of the key exchange. But generally, there are 1,999 other components with just as much performance impact, so focusing on any one needs some justification
English
0
0
0
81
Bas Westerbaan
Bas Westerbaan@bwesterb·
@sejaques You’d be left with very few, if any, engineer, that you’d allow to work on computational efficiency of their product.
English
1
0
0
115

Keşfet

@sweis @dallairedemers @bwesterb @veorq @CraigGidney @itranneo @hashbreaker @Visa