Professors Hu Fu and Gavin Tang are jointly teaching an online algorithms seminar, following selected content from Borodin and Pankratov’s 2019 textbook Online Algorithms. The course explores how algorithms perform under uncertainty, without prior knowledge of future inputs. Introduction to Online Algorithms Prior-Free Online Setting In online algorithms, decisions are made sequentially and non-revocable, without knowledge of future inputs. The final reward is contigent on the whole sequence of input. ...

This semester, Professor Hu Fu is offering a mechanism design seminar for senior undergraduates. The course is structured like a graduate-level theory class, without lecture notes or a base textbook. The mechanism design seminar adopts an algorithmic perspective on economic topics, focusing on: Combinatorial Auctions Optimal Auctions I will be updating and sharing my notes here for future reference. Let’s dive in. Combinatorial Auctions: Setup Consider an auction with: Items: $m$ distinct items to sell. Agents: $n$ bidders participating in the auction. Each agent $ i $ has a valuation function: ...

Super rich or super slim, at least one should be guaranteed. Douban is a Simplified Chinese online community that mainly aggregates well-literated Millennials and Gen-Zs, discussing movie (and general entertainment) comments, life and gossips. Here’s an interesting post on DouBan: starting today, losing weight w.r.t. A-share Quick shot of the original post on Sep. 10, 2024. A-share (A股) is short for domestic shares in China that are traded mainly in the Shanghai and Shenzhen stock exchanges. The owner of the post declared a ‘contract’ with the underlying asset being her own investment portfolio, that binds with her diet: ...

A publisher, an ad exchange, an advertiser and an ad server walk into a courtroom… source. Google will spend the next four to six weeks in a Virginia courtroom defending itself against allegations that it operates a monopoly over the digital advertising industry. It’s possibly the most influential antitrust case, ever. ad exchanger would be a nice place to follow up with expert pov and in-depth covers: DOJ vs. Google, Day One: Who’s Testifying And Google’s Main Arguments In Brief Your Day One Recap: DOJ vs. Google Goes Deep Into The Ad Tech Weeds The DOJ’s Witness List For The Google Antitrust Trial Is A Who’s Who Of Advertising The DOJ vs. Google, Day Two: Tales From The Underbelly Of Ad Tech There just isn’t enough popcorn in the world…

How to use SSH to authenticate connection to remote servers and services (e.g. GitHub). how it works (the mechanism) You can access and write data in repositories on GitHub using SSH (Secure Shell Protocol). When you connect via SSH, you authenticate using a private key file on your local machine. For more information about SSH, see Secure Shell on Wikipedia. When you set up SSH, you will need to generate a new private SSH key and add it to the SSH agent. You must also add the public SSH key to your account on GitHub before you use the key to authenticate or sign commits. ...

Here’s an interesting problem appeared in the final exam of market mechanism design. And I didn’t figure it out during then💔. The original question is from lecture note of 21 Algorithmic Game Theory, instructed by Michael Dinitz. background Consider combinatorial auction allocating $m$ items to $n$ players. Outcome is each bidder get a set $S_i\subset [m], \forall i \in [n]$ and they don’t overlap. Valuation: in this question we consider Single-Minded bidders. Each player $i$ has a target set $S_i^*\in [m]$ such that ...

This survey comprehensively covers the recent developments over the past two decades in autobidding within the online advertising ecosystem: Most research focuses on specific interactions between two of these components, often through the lenses of mechanism or algorithm design. Alimohammadi (2023). The complexity of the system is evident as it integrates various components—the advertisers, the autioneer, the auto-bidding agents—they are rarely analyzed collectively. ...

Here’s another Nature’s story on slow productivity: Slow productivity worked for Marie Curie — here’s why you should adopt it, too August 2024. Link. The story is immensely inspiring. I archived it here and highly recommend reading it! Here are some excerpts: Slow Productivity is a call to arms to reject the performative busyness of the modern workplace, where frequent virtual meetings and long e-mail chains sap so much of workers’ attention. One exhausted postdoctoral researcher interviewed by Newport defined productivity, as it is currently measured in academia, as “working all the time”. ...

Cal Newport is a computer scientist at Georgetown University in Washington DC. He also writes books about time management. Nature career summarized from his latest book Slow Productivity, the 7 ways to practise slow productivity in the lab: Limit daily goals. Newport recommends working on, and obsessing about, one large project a day rather than switching back and forth between multiple big tasks. This is something Newport learnt from computer scientist Nancy Lynch, his doctoral adviser at MIT. ...

There used to be an interesting debate on whether it is better to leave the notification on or block all notifications—in terms of reducing screen time and improving productivity. Two lovely read to begin the new semester: Nature has the stories: Science must protect thinking time in a world of instant communication July 2024. Link. Productivity is the eternal goal of corporate and academia. Apps—all kinds of apps—has been trying to increase connectivity. But they are not the same. ...