Three days following Taylor Swift’s release of her 31-track new album The Tortured Poets Department / The Anthology, Pitchfork released their reviews that rated 6.6 / 6 for the first and second half of the album, respectively. I believe writing the Pitchfork review of a Taylor Swift album should earn you the medal of freedom for your courage. –– Louie @DJLouieXIV, X Swifties quickly flooded the comment section under Pitchfork’s twitter post–speaking any slightest negativity unanonymously against the most passionate and gigantic fan base on earth requires extra courage and care. When there isn’t a ground truth but exists multiple conflicting believes, how influencial voices speak become a tricky issue. ...

Professor YaXiang Yuan recently graced SUFE with a compelling talk on Big Data & Optimization, masterfully engaging a diverse audience—from MBA students and enthusiastic undergrads to school leaders. He skillfully navigated the complexities of optimization theory, balancing depth with accessibility, transforming a regular Saturday into an enlightening experience. He began by highlighting key disciplines essential for shaping students into effective professionals: Statistics Computation Optimization Approximation Theory Discrete Mathematics Combinatorial Graph Theory Differential Equations Computational Geometry and more Using real-life examples, Yuan showcased how cutting-edge optimization theories enhance the modeling and solving of complex problems. Professor Yuan infused the session with his unique charm, paying homage to his deep roots in optimization theory while ensuring the talk remained approachable for all attendees. ...

My advisor was somehow less enthusiastic in information design as compared to other general algorithm game theory topics–despite my several failed attempts to lure him into doing some related projects. But his major concern is solid, that information design demands overly strong assumptions–the existence of a common prior, commitment of the signal sender, inference ability of the signal receiver–all poses challenges for direct applications that justify the existing theory. The real world should be something in between the perfectly rational Bayesian persuasion and the noisy cheap talks. ...

The first time I encountered the name “Daniel Kahneman” was in a behavioral economics class, where we were introduced to his seminal 1979 Econometrica paper on “prospect theory.” Initially, I pigeonholed him as an economist—a label he might not dispute, given his significant contributions to the interdisciplinary field of experimental economics, which bridges psychology and economics and has been instrumental in the development of behavioral economics. However, Daniel Kahneman is perhaps more a psychologist than an economist. The recent Freakonomics podcast episode, “The Men Who Started a Thinking Revolution (Update),” features an interview with journalist Michael Lewis about his 2016 book The Undoing Project, which offers a high-level exploration of Kahneman’s impact on our understanding of ourselves. ...

And how many years can some people exist Before they’re allowed to be free? – Bob Dylan, Blowing in the Wind It is becoming easier for non-college graduates to settle for a job in metropolitans, thanks to the growing platform economy that lowers the bar for job entry. As netizen jokes for the go-to job choices, about the three so-called auspicious opportunities: security, cleaning, insurance sellers and the challenging ironman triathlon: takeout delivery, courier, uber driver". ...

Previously, we discussed the challenges faced by apple farmers in ShaanXi, China (see the sweet and sour of apples and cranberries), that the farmers battle against the fluctuating price by strategically storing apple inventories so as to sell later, yet their decisions are far from optimal that brings regret and misfortune that sabotages the families’ living business. My initial thought was that the problem belongs to market failures, that potential solutions lies in redesigning the selling mechanism and boost information transparency etc.. But here’s another way of tackling the problem, which is somehow pretty fancy: ...

unpacking solvers: streamlining complexity Solvers are advanced software tools designed to process specific mathematical programs and deliver solutions, like a linear programming solver. When juxtaposed with algorithms—which are methods crafted for distinct problems—solvers operate on a broader spectrum, targeting classes of problems rather than individual cases. This abstraction not only boosts their flexibility but also their application across industries. Solvers enable a division of labor that heightens efficiency. Industry specialists focus on refining models to encapsulate real-world complexities, while solvers handle the computational tasks. This specialization allows engineers and scientists to enhance solvers’ efficiency and modeling power without getting mired in external complexities. ...

Here’s a note that clarifies two notions in econometric that I found confusing somehow. And mid-term exam is on the way so here’s a note. playground Let $\{Y_1, Y_2, \ldots, Y_n\}$ be a collection of independent and identically sampled random variables out of $f(y; \theta)$, where $\theta \in \Theta$ is a (possibly unknown) parameter associated with the distribution. In general, we want to estimate $\theta$ using our samples $Y$. An estimation of $\theta$ - random variable $W$ can be expressed as ...

Marianela Nunez’s rendition Kitri in The Royal Ballet’s production of Don Quixote is amazing. Kitri, if one doesn’t know, is the lead character in Don Quixote’s ballet-adjusted version. She embodies the Mediterranean dream of men: a flirtive, sexy and charismatic matured woman in her signature red dress. The first time I watched it I was in awe. After almost infinite replays I came to realize that Nela’s Kitri is by far the most nuanced classical artistic performance I’ve ever seen. ...

Prof. Yi Xiong’s Frontiers in Management Science Session covered a crash course on online learning and multi-armed bandits (MAB). A simple way to view MAB is to consider it as a tool to model and study online learning problems. At time $t$, a decision maker chooses an ‘action’ or arm $i_t\in \mathcal I$ ($\mathcal I$ is the known set of distributions). The chosen machine returns a random reward $r_t$ following its distribution $Q_t$. Our goal as the octopus is to maximize the expected cumulative rewards $\sum_{t = 1}^T \mathbb E[r_t]$. ...