Max Aifer (@MaxAifer) / X

Max Aifer

588 posts

Max Aifer

@MaxAifer

Theorist @NormalComputing. Thermodynamic computing for efficient AI.

New York, NY

Joined July 2021

Pinned
Max Aifer
@MaxAifer
Feb 19
Check out our new paper on the application of lattice random walks to bayesian inference
Lars Holdijk
@HoldijkLars
Feb 19
Struggling with minibatch noise in Stochastic Gradient Bayesian Inference? Want your chains to naturally run on stochastic hardware? Introducing — Stochastic Gradient Lattice Random Walk! New work from @NormalComputing in collab with @zierhjmensch, @adnarim066, @wellingmax, and
00:00
1.2K
Max Aifer
@MaxAifer
Oct 3, 2024
A thread on our new paper Thermodynamic Bayesian Inference 250 years later, Bayes’s theorem is still the gold standard for probabilistic reasoning. But for complicated models it’s too hard to implement exactly, so approximations are used. For example, the complexity of Bayesian
174K
Max Aifer
@MaxAifer
Aug 11, 2023
Replying to @MaxAifer
Enter thermodynamic computing. In this preprint, Thermodynamic Linear Algebra (arxiv.org/abs/2308.05660), we show that a system of coupled oscillators in contact with a heat reservoir can be used to solve linear systems in an amount of time proportional to the number of variables.
231K
Max Aifer
@MaxAifer
Oct 3, 2024
Introducing Thermodynamic Bayesian Inference. We propose physical systems whose thermal equilibrium states correspond to Bayesian posteriors. This allows for sampling posteriors with improved time and energy costs. Deep dive coming soon.
38K
Max Aifer
@MaxAifer
Oct 31, 2024
A few months ago we scored a big victory for thermodynamic computing which has been kept quiet until now, and now is the time to talk about it. Based on the ideas in our paper Thermodynamic Linear Algebra (arxiv.org/abs/2308.05660), we wrote a proposal to vastly reduce the cost of
arxiv.org
Thermodynamic Linear Algebra
Linear algebraic primitives are at the core of many modern algorithms in engineering, science, and machine learning. Hence, accelerating these primitives with novel computing hardware would have...
33K
Max Aifer
@MaxAifer
Aug 11, 2023
Finding the point where these three planes intersect is an example of a problem which has been studied for literal millennia: solving a linear system of equations. Billions of dollars have been spent on developing new hardware to solve this and related problems faster.
41K
Max Aifer
@MaxAifer
Nov 13, 2023
A basic fact in linear algebra is that an N-dimensional vector space can have at most N linearly independent vectors. Is there a better statement of non-orthogonality when there are more vectors than dimensions? This is the best I could derive:
35K
Max Aifer
@MaxAifer
Apr 22, 2024
Just turned in my dissertation. 🎉
11K
Max Aifer
@MaxAifer
Feb 11, 2025
Thomas appears to have solved it using some dark magic. thanks @thomasahle :)
18K
Max Aifer
@MaxAifer
Feb 5, 2025
I graduated in December but haven’t gotten around to posting this yet. A lot of people will tell you that a PhD isn’t worth it in the modern job market. In a way, they’re right; and that’s what they’re supposed to say. And what you’re supposed to do, if you love to learn, is
6.6K
Max Aifer
@MaxAifer
Oct 3, 2024
Replying to @MaxAifer
In our new paper Thermodynamic Bayesian Inference (arxiv.org/abs/2410.01793), we propose devices which are physical realizations of the systems simulated by the Langevin algorithm. These systems take the form of electronic circuits with noise sources, and come to thermal
14K
Max Aifer
@MaxAifer
Nov 22, 2023
Quantum parallelisim—the idea that quantum computers gain an advantage by “trying multiple possibilities at once”—is popular but controversial. In fact, quantum mechanics is not even necessary to try multiple possibilities at once: we can achieve this with just thermodynamics.
16K
Max Aifer
@MaxAifer
Aug 11, 2023
Replying to @MaxAifer
Thermodynamic algorithms are of growing interest, and our results represent the first rigorously-proven speedups for thermodynamic algorithms. The results also reveal a deep, and wholly unexplored connection between thermodynamics and linear algebra.
7.4K
Max Aifer
@MaxAifer
Apr 28, 2024
Some good points made here. Some of them are not specific to any one company and more about thermo computing in general, so I’ll try to respond to those:
kyon
@0xKyon
Apr 28, 2024
unpopular opinion: first off this whole idea of using analog circuits to accelerate ai workloads aint exactly new. ibm intel and others have been experimenting with that shit for years tryna leverage the noise tolerance of neural nets to boost efficiency heres the thing analog
32K