TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

MemComputing vs. Quantum Computing

29 点作者 velmu大约 2 年前

7 条评论

inasio大约 2 年前
These guys claimed that they could factor semi-primes in polynomial time [0], essentially P == NP. Scott Aaronson predictably had some thoughts on that [1]<p>[0] <a href="https:&#x2F;&#x2F;pubs.aip.org&#x2F;aip&#x2F;cha&#x2F;article&#x2F;27&#x2F;2&#x2F;023107&#x2F;135054&#x2F;Polynomial-time-solution-of-prime-factorization" rel="nofollow">https:&#x2F;&#x2F;pubs.aip.org&#x2F;aip&#x2F;cha&#x2F;article&#x2F;27&#x2F;2&#x2F;023107&#x2F;135054&#x2F;Poly...</a> [1] <a href="https:&#x2F;&#x2F;scottaaronson.blog&#x2F;?p=2212" rel="nofollow">https:&#x2F;&#x2F;scottaaronson.blog&#x2F;?p=2212</a>
评论 #35908531 未加载
marcodiego大约 2 年前
&gt; Somewhat surprisingly, just the emulation of this circuit in software (Virtual Memcomputing Machine) provides solutions to these problems orders of magnitude faster and more accurately than today’s state of the art technologies.<p>Looks like market speech to me.
评论 #35909012 未加载
generalizations大约 2 年前
I didn&#x27;t know that memcomputing had already been solved. That&#x27;s fantastic news...here&#x27;s hoping it doesn&#x27;t remain ridiculously expensive.<p>I take issue with this line, however.<p>&gt; Both MemComputing and quantum computers are physics-based approaches to computation. That is, they set aside many of the foundations of traditional computer science and rethink computation from the ground up. For example, they don’t follow the von Neumann architecture, the computing framework that separates the processing unit from the memory unit, which is employed by all of today’s computers.<p>From what I recall of Von Neumann, his titular architecture was the &#x27;first draft&#x27; of what he was going for, and we&#x27;ve been stuck with it before he could figure out version two. He based his design on the notion of cellular automaton, where of course the processing units are not separated from the memory units.<p>Rather, from what I recall, the Architecture we&#x27;ve been stuck with was just his best attempt at emulating what&#x27;s here called memcomputing, but with the limited engineering resources at his disposal. If he hadn&#x27;t died, I think he would have been trying to do proper memcomputing in the second draft. Only took us the better part of a century to catch up!
评论 #35908351 未加载
inasio大约 2 年前
Their tagline at the end:<p>```MemComputing is available today and is delivering the performance expected of quantum computers across a wide range of valuable optimization problems in industry. ```<p>Quantum computers unfortunately are not expected to do well against traditional OR algorithms in most (all?) OR problems (Grover search only helps in brute force search; Shor it not quite an OR application, and relies on quantum effects), so not really a strong selling point in my view
ftxbro大约 2 年前
&gt; &quot;SOG-based Circuits (SOC) can provide ultra low-power, real-time computation for many challenging computational tasks common in signal processing, graphics, physical simulations, AI, etc. These circuits are also very efficient when solving complex optimization problems. Somewhat surprisingly, just the emulation of this circuit in software (Virtual Memcomputing Machine) provides solutions to these problems orders of magnitude faster and more accurately than today’s state of the art technologies.&quot;<p>It smells like some BS to anyone who knows about this kind of thing, but I think their marketing is for the ones who don&#x27;t know.
thxg大约 2 年前
MemComputing mainly present themselves as optimizers: Their machine can solve real optimization problems, in practice, today. This is very good because, through that lens, we can skip the BS and go straight to the facts. The evaluation criteria for any optimization approach&#x2F;algorithm are:<p>- the quality of the solutions found (in the best case: optimal solutions)<p>- the presence or absence of optimality guarantees (for example, some algorithms provide optimal solutions very often, but cannot guarantee it 100%)<p>- the time (or computational cost) needed to find such solutions<p>Furthermore, the state-of-the-art (SOTA) is well known for most types of optimization problems. In this article, they present a list of real and practical applications, so let us have a look at them one by one:<p>1.1 Traffic flow optimization<p>Simple flow problems can be solved in polynomial time (and quickly in practice), so there is no need for anything fancy. Once you introduce additional constraints or discrete variable, the SOTA is mixed-integer programming for offline problems. For online problems it&#x27;s more complicated. In both cases, I am not aware of any application in which MemComputing can reach SOTA.<p>1.2 Vehicle routing &amp; scheduling<p>Here, depending on your computing time constraints, needs for solution quality and&#x2F;or optimality guarantees, the SOTA can be constraint programming (gecode, OptaPlanner), local search heuristics (LocalSolver), or mixed-integer programming (CPLEX&#x2F;XPress&#x2F;Gurobi) with column generation. MemComputing is nowhere to be seen again.<p>1.3 Supply chain optimization<p>This is a very broad field, but in general mixed-integer programming is king here.<p>2.1 Protein folding<p>Here there is quite an objective measure: the biennal CASP competition. This is where AlphaFold made a splash in 2022. MemComputing has never participated.<p>2.2 Genome sequencing<p>I am not knowledgeable enough to comment here.<p>2.3 Radiotherapy treatment<p>I am not very knowledgeable here either, but last I looked mixed-integer programming approaches were favored.<p>3.1 Portfolio risk optimization<p>Various types of branch-and-bound solvers. Mixed-integer linear&#x2F;quadratic&#x2F;convex programming. No MemComputing.<p>3.2 Detecting market instabilities<p>No idea.<p>3.3 Optimizing trading trajectories<p>No idea.<p>4.1 Training neural networks<p>Many people here know how this is done. Stochastic gradient descent on GPUs or TPUs. No MemComputing involved. How can they even claim to be active in this field?<p>4.2 Detecting statistical anomalies<p>Vague.<p>4.3 Classifying unstructured datasets<p>No idea.<p>The problem is that if you invent a new optimization algorithm, it is very easy to find one instance of one problem for which your algorithm works well. They did literally that in a paper [1]: They took a library of mixed-integer programming problem instances containing 270 benchmark problems, and published a whitepaper showing that they beat a SOTA solver on one of them. A single instance out of 270!<p>The really hard part is the opposite: given a class of problems, find an algorithm that beats the SOTA. MemComputing has never done that. Combined with their propensity for grand claims backed by misleading evidence, MemComputing have accumulated a lot of badwill from the academic community over the years. My suspicion is that, while on the surface this post seems to put their approach in contrast to quantum computing, what they really try to do here is ride on the quantum computing hype wave.<p>[1] <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2003.10644" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2003.10644</a>
IIAOPSW大约 2 年前
Oranges vs Apples.
评论 #35908546 未加载