frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

Celsius founder Alex Mashinsky sentenced to 12 years for "unbank yourself" scam

https://arstechnica.com/tech-policy/2025/05/celsius-founder-gets-12-years-for-multi-billion-dollar-crypto-lending-scam/
1•tomrod•45s ago•0 comments

Finding Beauty and Truth in Mundane Occurrences

https://www.quantamagazine.org/finding-beauty-and-truth-in-mundane-occurrences-20250509/
1•johnshades•1m ago•0 comments

LLM SQL Generation Benchmark Results

https://llm-benchmark.tinybird.live/
1•sh_tomer•3m ago•0 comments

The "Diet" Smartphone: Minimal Phone Review [video]

https://www.youtube.com/watch?v=atYcpCoghnc
2•lordleft•3m ago•0 comments

Former Supreme Court Justice David Souter Dies at 85

https://www.npr.org/2025/05/09/g-s1-65326/justice-david-souter-dies
1•danso•4m ago•1 comments

VS Code: April 2025

https://code.visualstudio.com/updates/v1_100
1•tosh•4m ago•0 comments

AI on Kubernetes Without the Pain

https://blog.skypilot.co/ai-on-kubernetes/
1•rombr•5m ago•0 comments

Playing repeated games with large language models

https://www.nature.com/articles/s41562-025-02172-y
2•bookofjoe•6m ago•0 comments

Plane caught fire as pilot confused left and right

https://www.bbc.co.uk/news/articles/c9wgjqj4xx4o
1•jeffwass•6m ago•0 comments

Pakistan's Chinese-made jet brought down two Indian fighter aircraft

https://www.reuters.com/world/pakistans-chinese-made-jet-brought-down-two-indian-fighter-aircraft-us-officials-2025-05-08/
2•c420•7m ago•0 comments

What do you think about YC Summer 2025 Request for Startups?

1•BetzyFlores504•7m ago•0 comments

Some Nuanced Thoughts on AI

https://bentsukun.ch/posts/ai-nuance/
2•speckx•8m ago•0 comments

I Hadn't Written a Resume Since 2008–So I Coded One Instead

https://www.richardadleta.com/posts/resume-viewer-intro
1•radleta•9m ago•1 comments

Coffee shops ditch WiFi and laptops to limit remote work

https://www.axios.com/2025/05/09/coffee-shops-no-wifi-laptops-remote-work
3•bikenaga•9m ago•0 comments

Infinite Craft

https://neal.fun/infinite-craft/
2•salzig•12m ago•0 comments

I Visited a Chinese NAS Factory and Here Is What I Saw [video]

https://www.youtube.com/watch?v=PMcucckpkmw
1•rhema•12m ago•0 comments

The Zeigarnik Engine: Turning Open Loops into Momentum

https://www.joanwestenberg.com/the-zeigarnik-engine-using-unfinished-tasks-to-fuel-endless-momentum/
1•NotInOurNames•13m ago•0 comments

Show HN: Building a self collected multiview mocap studio for robot training

https://github.com/rerun-io/pi0-lerobot/tree/hand-kinematic-fitting
1•pablovelagomez•13m ago•0 comments

Debian Looks to Better Address Ill-Maintained / Dormant Packages

https://www.phoronix.com/news/Dormant-Debian-Packages-Issue
3•mikece•15m ago•0 comments

Show HN: Selli – Real-Time AI Assistant for Sales Calls (macOS)

https://www.selli.app
1•milanWiegard•16m ago•0 comments

Yunzii Launches Its QL75 Retro Typewriter Mechanical Keyboard

https://www.techpowerup.com/336578/yunzii-launches-its-ql75-retro-typewriter-mechanical-keyboard
1•speckx•18m ago•0 comments

Radxa Orion O6 Brings Arm to the Midrange PC

https://www.jeffgeerling.com/blog/2025/radxa-orion-o6-brings-arm-midrange-pc
4•ingve•19m ago•1 comments

Tesla Cybertruck inventory goes through the roof

https://www.arenaev.com/tesla_cybertruck_inventory_goes_through_the_roof-news-4680.php
6•mooreds•19m ago•0 comments

I Hacked a Japanese Corporate Exam with a Local AI Model

https://www.rafaelviana.io/posts/line-badge
1•vianarafael•19m ago•0 comments

SveltePlot: A grammar of graphics style framework for Svelte

https://svelteplot.dev/why-svelteplot
1•tomgp•20m ago•0 comments

Long-Term Ultraprocessed Food Intake and Prodromal Features of Parkinson Disease

https://www.neurology.org/doi/10.1212/WNL.0000000000213562
1•nixass•20m ago•0 comments

CRUD APIs Are Poor Design

https://codeopinion.com/crud-apis-are-poor-design/
1•ingve•21m ago•0 comments

ZeroSearch: Incentivize the Search Capability of LLMs without Searching

https://alibaba-nlp.github.io/ZeroSearch/
2•vessenes•23m ago•1 comments

Show HN: I built a visual AI chat with branching and rich content

https://why.new
2•rogutkuba•24m ago•0 comments

Ask HN: How to get good at marketing your product and SEO?

2•flashblaze•26m ago•0 comments
Open in hackernews

Linear Programming for Fun and Profit

https://modal.com/blog/resource-solver
40•hmac1282•6h ago

Comments

ayhanfuat•1h ago
> X = [x1, ..., Xn]: instances of each type to launch

Is this a continuous variable? Seems discrete to me. I am surprised it is solved by simplex.

Frummy•1h ago
It's the answer, a vector of integers
ayhanfuat•55m ago
Simplex cannot give a vector of integers though, unless the constraint matrix is unimodular. Maybe the integrality constraint was relaxed.
cweld510•49m ago
You're right -- we do relax the integrality constraint, gaining performance at the expense of some precision, and we're generally able to paper over the difference at scheduling time. We've investigated integer linear programming for some use cases, but for solves to run quickly, we have to constrain the inputs significantly.
ayhanfuat•37m ago
Thanks for the clarification. I guess it wouldn’t matter much if the numbers are large. Initially I thought they were mostly ones and zeros.
stncls•32m ago
If this is business critical for you, you may want to switch to a faster solver. Glop is very nice, but it would be reasonable to expect a commercial solver (Gurobi, XPress, COpt) to be 60x faster [1]. By the same measure, the best open source solvers (CLP, HiGHS) are 2-3x faster than Glop.

Actually, the commercial solvers are so fast that I would not be surprised if they solved the IP problem as fast as Glop solves the LP. (Yes, the theory says it is impossible, but in practice it happens.) The cost of a commercial solver is 10k to 50k per license.

[1] ... this 60x number has very high variance depending on the type of problem, but it is not taken out of nowhere, it comes from the Mittelmann LP benchmarks https://plato.asu.edu/ftp/lpopt.html There are also benchmarks for other types of problems, including IP, see the whole list here: https://plato.asu.edu/bench.html

petters•31m ago
If you are able to paper over the fractional numbers and get a usable solution, an integer solver should also be able to find a feasible solution easily. Perhaps not optimal, but better than just solving the LP and rounding
cweld510•1h ago
Great to see this post here -- really enjoyed writing it! I think it's really cool how an algorithm from an operational research context can play a critical role in a high-availability large-scale cloud service.
sumtechguy•1h ago
LP is a shockingly good way to optimize a system. If you can put inputs/outputs into the correct form. Had an econ prof that loved these things for doing supply/demand maxima and minimum finding. He didnt outright say it but I think it was his current line of study when I was taking classes from him the 90s. I thought that, as he managed to bring it up in every class he taught.
underanalyzer•39m ago
Neat article. I do wish it mentioned that there are polynomial time algorithms to solve linear programming problems. According to the Google ortools docs it has the option to use those as well (but not with the GLOP solver). Might be good for when simplex is struggling (https://developers.google.com/optimization/lp/lp_advanced)
stncls•10m ago
You're right, but it's very subtle and complicated.

In theory, the simplex method is not known to be polynomial-time, and it is likely that indeed it is not. Some variants of the simplex method have been proven to take exponential time in some worst cases (Klee-Minty cubes). What solvers implement could be said to be one such variant ("steepest-edge pricing"), but because solvers have tons of heuristics and engineering, and also because they work in floating-point arithmetic... it's difficult to tell for sure.

In practice, the main alternative is interior-point (aka. barrier) methods which, contrary to the simplex method, are polynomial-time in theory. They are usually (but not always) faster, and their advantage tends to increase for larger instances. The problem is that they are converging numerical algorithms, and with floating-point arithmetic they never quite 100% converge. By contrast, the simplex method is a combinatorial algorithm, and the numerical errors it faces should not accumulate. As a result, good solvers perform "crossover" after interior-point methods, to get a numerically clean optimal solution. Crossover is a combinatorial algorithm, like the simplex method. Unlike the simplex method though, crossover is polynomial-time in theory (strongly so, even). However, here, theory and practice diverge a bit, and crossover implementations are essentially simplified simplex methods. As a result, in my opinion, calling iterior-point + crossover polynomial-time would be a stretch.

Still, for large problems, we can expect iterior-point + crossover to be faster than the simplex method, by a factor 2x to 10x.

There is also first-order methods, which are getting much attention lately. However, in my experience, you should only use that if you are willing to tolerate huge constraint violations in the solution, and wildly suboptimal solutions. Their main use case is when other solvers need too much RAM to solve your instance.