Chris Peikert
@chrispeikert.bsky.social
Cryptographer (lattices/post-quantum), Professor at U-Michigan Computer Science and Engineering , Head of Cryptography Research at Algorand, PhD from MIT CSAIL. Previously faculty at Georgia Tech School of CS. Here I speak for myself.
What’s the story with the buggy formally verified library?
November 4, 2025 at 12:13 AM
What’s the story with the buggy formally verified library?
Reposted by Chris Peikert
Also, maybe we come up with, like, two more holidays, say in the early and late spring, that involve walking around the neighborhood and talking to your neighbors?
October 31, 2025 at 10:56 PM
Also, maybe we come up with, like, two more holidays, say in the early and late spring, that involve walking around the neighborhood and talking to your neighbors?
Our protocol is simple & modular: it works in the "Arithmetic Black Box" model, and inherits the security properties of the ABB instantiation (passive/active, static/adaptive, etc.).
Using SPDℤ₂𝑘 for ABB, we get about 20,000x and 37x better throughput and latency than SoTA for dishonest majority.
Using SPDℤ₂𝑘 for ABB, we get about 20,000x and 37x better throughput and latency than SoTA for dishonest majority.
October 14, 2025 at 3:40 PM
Our protocol is simple & modular: it works in the "Arithmetic Black Box" model, and inherits the security properties of the ABB instantiation (passive/active, static/adaptive, etc.).
Using SPDℤ₂𝑘 for ABB, we get about 20,000x and 37x better throughput and latency than SoTA for dishonest majority.
Using SPDℤ₂𝑘 for ABB, we get about 20,000x and 37x better throughput and latency than SoTA for dishonest majority.
Several prior threshold decryption protocols use "noise flooding" for security, but this has a high cost in parameters and efficiency.
Instead, we give an efficient MPC "rounding" protocol. It prepares special "gates" offline, and has quite low communication and computation in the online phase.
Instead, we give an efficient MPC "rounding" protocol. It prepares special "gates" offline, and has quite low communication and computation in the online phase.
October 14, 2025 at 3:40 PM
Several prior threshold decryption protocols use "noise flooding" for security, but this has a high cost in parameters and efficiency.
Instead, we give an efficient MPC "rounding" protocol. It prepares special "gates" offline, and has quite low communication and computation in the online phase.
Instead, we give an efficient MPC "rounding" protocol. It prepares special "gates" offline, and has quite low communication and computation in the online phase.
Prior work used fairly simple weights. We define much "smoother" ones (e.g., Gaussian or Laplacian), and use Fourier analysis on a certain lattice to show that these weights have large enough "correlation" with any nearby codeword.
Check it out!: web.eecs.umich.edu/~cpeikert/pu...
Check it out!: web.eecs.umich.edu/~cpeikert/pu...
web.eecs.umich.edu
October 13, 2025 at 2:53 PM
Prior work used fairly simple weights. We define much "smoother" ones (e.g., Gaussian or Laplacian), and use Fourier analysis on a certain lattice to show that these weights have large enough "correlation" with any nearby codeword.
Check it out!: web.eecs.umich.edu/~cpeikert/pu...
Check it out!: web.eecs.umich.edu/~cpeikert/pu...
The key technique (as in our prior work with Ethan Mook on decoding in the ℓ₂ metric) is to use the Guruswami--Sudan 𝒔𝒐𝒇𝒕-𝒅𝒆𝒄𝒊𝒔𝒊𝒐𝒏 list decoder.
For this we need to transform the received word into "weights" that specify, for each coordinate, a "confidence level" for each potential codeword symbol.
For this we need to transform the received word into "weights" that specify, for each coordinate, a "confidence level" for each potential codeword symbol.
October 13, 2025 at 2:53 PM
The key technique (as in our prior work with Ethan Mook on decoding in the ℓ₂ metric) is to use the Guruswami--Sudan 𝒔𝒐𝒇𝒕-𝒅𝒆𝒄𝒊𝒔𝒊𝒐𝒏 list decoder.
For this we need to transform the received word into "weights" that specify, for each coordinate, a "confidence level" for each potential codeword symbol.
For this we need to transform the received word into "weights" that specify, for each coordinate, a "confidence level" for each potential codeword symbol.
Most notably: compared to prior work, our algorithm can decode to 𝒂𝒓𝒃𝒊𝒕𝒓𝒂𝒓𝒊𝒍𝒚 𝒍𝒂𝒓𝒈𝒆 distances, for correspondingly small rates. Moreover, we show that our algorithm is a 𝒖𝒏𝒊𝒒𝒖𝒆 decoder for many parameters of interest.
Here are some plots of the distance-rate tradeoffs and the uniqueness thresholds.
Here are some plots of the distance-rate tradeoffs and the uniqueness thresholds.
October 13, 2025 at 2:53 PM
Most notably: compared to prior work, our algorithm can decode to 𝒂𝒓𝒃𝒊𝒕𝒓𝒂𝒓𝒊𝒍𝒚 𝒍𝒂𝒓𝒈𝒆 distances, for correspondingly small rates. Moreover, we show that our algorithm is a 𝒖𝒏𝒊𝒒𝒖𝒆 decoder for many parameters of interest.
Here are some plots of the distance-rate tradeoffs and the uniqueness thresholds.
Here are some plots of the distance-rate tradeoffs and the uniqueness thresholds.
Reposted by Chris Peikert
Yesterday, Kristi Noem’s and Greg Bovino’s masked agents threw chemical agents near an elementary school, arrested elected officials exercising their First Amendment rights, and raided a Wal-Mart.
October 4, 2025 at 6:10 PM
Yesterday, Kristi Noem’s and Greg Bovino’s masked agents threw chemical agents near an elementary school, arrested elected officials exercising their First Amendment rights, and raided a Wal-Mart.
Reposted by Chris Peikert
We will Make this deadline, very strongly, because we are WINNERS—but Europe and the Rest of the world will have to pay up BIGLY with the strongest and Most polished Best Paper awards! That is the least these low IQ individuals can do to Pay back what we Americans so richly deserve!
October 3, 2025 at 3:37 AM
We will Make this deadline, very strongly, because we are WINNERS—but Europe and the Rest of the world will have to pay up BIGLY with the strongest and Most polished Best Paper awards! That is the least these low IQ individuals can do to Pay back what we Americans so richly deserve!
We will Make this deadline, very strongly, because we are WINNERS—but Europe and the Rest of the world will have to pay up BIGLY with the strongest and Most polished Best Paper awards! That is the least these low IQ individuals can do to Pay back what we Americans so richly deserve!
October 3, 2025 at 3:37 AM
We will Make this deadline, very strongly, because we are WINNERS—but Europe and the Rest of the world will have to pay up BIGLY with the strongest and Most polished Best Paper awards! That is the least these low IQ individuals can do to Pay back what we Americans so richly deserve!