I still can’t believe I performed Stravinsky’s Rite of Spring when I was like 16 or so —the flute part, no less. It felt impossible at the time (still does). The beat counting alone is a waking nightmare for any flutist. You don’t “understand” the piece as much as survive it. Now years later as I’m watching a recreation of its riotous 1913 premiere—courtesy of the film Coco Chanel & Igor Stravinsky (2009)—I’m still stunned. The film opens with a faithful reenactment of that infamous night when Paris’s upperclass literally lost their mind and etiquette over, divine art: (fun fact: the movie is (loosely) about a love affair between Chanel and Stravinsky. With Mads Mikkelsen playing Stravinsky, honestly, I’d believe he can have an affair with anyone…) ...

I’ve came across some interesting stories about Tchaikovsky and his music, from the official website of the Berliner Philharmoniker: What you (might) not know about Pyotr Ilych Tchaikovsky Lino Knocke | Link: https://www.berliner-philharmoniker.de/en/stories/what-you-might-not-know-about-pyotr-ilych-tchaikovsky/ The Porcelain Child “Even as a boy, he responded intensely to his surroundings – and especially to music. Certain harmonies moved him so deeply that he would burst into tears, or collapse sobbing onto his bed after playing the piano, overwhelmed by the physical force of the experience.” ...

Our reading group recently wrapped up a sequence of papers under the theme Beyond Bayesian Bandits — or as I like to think of it, “3B.” (Yes, like the composers: Bach, Beethoven, Brahms. Also sometimes a bit boring.) Anyway, here’s what we’ve covered so far. For some books and papers I don’t attach links but they should be google-able. 📚 Completed Readings Kleinberg: Introduction to Multi-Armed Bandits Slides (Cornell CS6840, 2017) Dumitriu, Tetali, Winkler: Playing Golf with Two Balls Whittle (1980): Multi-armed Bandits and the Gittins Index More useful Gittins Index books: Gittins, Glazebrook and Weber (2011) Multi-armed Bandit Allocation Indices (Second Edition) Qing Zhao (2019) (Section II and III of) Multi-Armed Bandits: Theory and Applications to Online Learning in Networks Hadfield-Menell & Russell (UAI 2015): Multitask Inverse Reinforcement Learning PDF Guha, Munagala, Shi: Restless Bandits with Constraints FOCS 2007 / SODA 2009 Doval & Scully (2024, under review): Local Hedging in Bandits arXiv Chawla, Christou, Harlev, Scully (2025, in submission) arXiv 🔍 Future Readings Gupta, Jiang, Scully, Singla: The Markovian Price of Information arXiv Hajiaghayi, Krysta, Mahdavi, Shin (EC 2025): Delegation with Costly Inspection Banihashem, Hajiaghayi, Krysta, Shin (EC 2025): Delegated Choice with Combinatorial Constraints Ziv Scully & Alexander Terenin: Tutorial: The Gittins Index as a Design Principle (Seems like a solid conceptual anchor for everything above.)

The Tchaikovsky Pas de Deux is bouncy, full of energy. Full sheet music can be downloaded at IMSLP. Reading the complete score is really not so convenient (cause all instruments are stacked together): I’m always awed by how fast conductors read… The female variation has light, crisp and joyful and consistent melody that really make up nice orchestra audition snippet. I assemble them, just for reference (although you can probably just memorize the melody after several pass of the music anyway, but just in case): ...

A brief summary and proof sketch of the core theorem behind the Universal Singular Value Thresholding (USVT) estimator for matrix completion.

This is the arguably the most helpful and standard exercise set for the flute: 17 Grands exercices journaliers de mécanisme (17 grand daily mechanism exercises). IMSLP link. About the authors: Claude-Paul Taffanel (his Wikipedia) was a French flutist and composer and was regarded as the founder of the French Flute School that dominated much of flute composition and performance during the mid-20th century. Taffanel’s student Philippe Gaubert (the other composer) was also a distinguished flute performer, with the Paris Opera and was also Professor of flute in the Conservatoire de Paris (teacher of Marcel Moyse, who is the teacher of Sir James Galway). Basically, they are the history of the flute instrument. ...

Our reading group covered this paper today: Local hedging approximately solves Pandora’s box problems with nonobligatory inspection Ziv Scully and Laura Doval (2025) | Link: https://arxiv.org/pdf/2410.19011 Abstract: We consider search problems with nonobligatory inspection and single-item or combinatorial selection. A decision maker is presented with a number of items, each of which contains an unknown price, and can pay an inspection cost to observe the item’s price before selecting it. Under single-item selection, the decision maker must select one item; under combinatorial selection, the decision maker must select a set of items that satisfies certain constraints. In our nonobligatory inspection setting, the decision maker can select items without first inspecting them. It is well-known that search with nonobligatory inspection is harder than the well-studied obligatory inspection case, for which the optimal policy for single-item selection (Weitzman, 1979) and approximation algorithms for combinatorial selection (Singla, 2018) are known. ...

The Economist has a really interesting July 24 story and relevant articles about AI and its effect on (main macro) economics: The economics of superintelligence | If Silicon Valley’s predictions are even close to being accurate, expect unprecedented upheaval https://www.economist.com/leaders/2025/07/24/the-economics-of-superintelligence (The cover is very adorable, click the above link and check it out) The article says, it is believed that superintelligence that outwits (at least average) humanity AI would soon emerge. And the article did a really refreshing thought experiment: ...

I’m practicing the adapted flute solo piece for Lensky’s Aria from Tchaivkosky’s Opera Eugene Onegin. Here’s the background check for this piece: The opera has three acts and Lensky’s aria takes place near the end of Act II: Act I: Tatyana, a bookish and romantic young woman, falls in love with the dashing but aloof Eugene Onegin, a friend of the poet Vladimir Lensky. She writes Onegin a heartfelt letter confessing her feelings. He coldly rejects her, advising her to control her emotions. Act II: At Tatyana’s name-day party, Onegin flirts with Olga, Lensky’s fiancée, partly out of boredom and partly to spite Lensky. Lensky, deeply hurt, challenges Onegin to a duel. Before the duel, Lensky sings his aria (“Kuda, kuda vï udalilis…”), reflecting on his life, his love, and the possibility of death. Onegin kills Lensky in the duel. He is devastated by his own actions and begins a life of guilt and wandering. Act III: Years later, Onegin returns to society and meets a transformed Tatyana—now elegant and composed, married to Prince Gremin. He falls in love with her, but she rejects him, though she confesses she still loves him. She stays loyal to her husband. Lensky’s Aria remorse on his memories of youth, quiet acceptance of possible death and a mixture of sorrow, nostalgia, and peace. This website organized the OG Russian poem (which was directly applied in Tchaikovsky’s opera) with pronunciation and English translation: https://classicalmusicandmusicians.com/2018/03/21/lenskys-aria-from-eugene-onegin-act-ii-sergei-lemishev-at-age-35/. ...

I first encountered matrix completion duing undergrad in Advance Operations course. So I thought it was history well-solved problem. Turns out well, amazing papers still emerge over the past years: Matrix Estimation by Universal Singular Value Thresholding (USVT) Sourav Chatterjee The Annals of Statistics (2015) Consider the problem of estimating the entries of a large matrix, when the observed entries are noisy versions of a small random fraction of the original entries. This problem has received widespread attention in recent times, especially after the pioneering works of Emmanuel Cand`es and collaborators. ...