Je suis fasciné par la proportion de copies (genre 90%) qui veulent utiliser le théorème de convergence dominée pour justifier que si f_n → f dans L² et g ∈ L², alors ∫(f_n · g) → ∫(f·g). En lisant l'énoncé, je n'avais même pas envisagé une possible difficulté à cet endroit.
Et le pire, c'est que des malins s'en tirent quand même (la question était en fait que si ∫(f_n · g) = 0 pour tout n alors ∫(f·g)=0: ils prennent une suite extraite des f_n qui converge p.p. et qui est dominée dans L², et appliquent Hölder puis le thm de convergence dominée)!
Je ne dis pas ça pour dire «ces élèves sont nuls»: là c'est clairement un problème d'enseignement, nous leur avons donné l'impression que quand on veut passer à la limite dans une intégrale c'est forcément le thm de convergence dominée qu'il faut appliquer, …
… et du coup ils ne savent rien faire des autres convergences, dans L² ou L¹ disons. Soit ils les confondent sans rien dire de plus, soit ils voient la différence et se sentent obligés de prendre une suite extraite pour se ramener à la convergence p.p.
Précisons que la réponse «attendue» était 🔽 (je donne trois variantes équivalentes, dans le tweet cité et les deux au-dessus de lui; en tout cas le point-clé était Cauchy-Schwarz ou Hölder):
Looking at Ukrainian words from the perspective of someone who knows just a little bit of Russian gives the superficial impression that:
⁃ Russian ‘и’ and ‘ы’ match to Ukrainian ‘і’ and ‘и’,
⁃ Russian ‘е’ and ‘э’ match to Ukrainian ‘є’ and ‘е’.
But It's Complicated™. •1/25
Here's what I THINK I understood. First, Slavic vowels historically came into two flavors: normal and “iotated”, pronounced with a /j/ before them (disclaimer: this is probably a huge oversimplification). In Old Slavonic, they are (sometimes) written with a iota-ligature. •2/25
The iotated vowels, but also some front vowels that were not iotated (‘i’, ‘ь’, ‘e’, ‘ѣ’, ‘ѧ’) caused palatalization (“softening”) of the previous consonant in East Slavic languages, inc. Russian and Ukrainian (exc. ‘e’?). But some vowels also changed, and others dropped. •3/25
The latest xkcd comic xkcd.com/2585/ leads me to raise the following question: suppose u_0,…,u_n (we can assume w.l.o.g. u_0=1) are positive real numbers (“units of measure”) and we consider the following graph (“integer measurements with these units”): …
… vertices of the graph are pairs (i,k) where 0≤i≤n and k∈ℤ, and we have an edge (i,k) → (j,ℓ) iff ℓ is the closest integer to k · u_i / u_j (conversion of k units u_i to the unit u_j), maybe assuming u_i/u_j are all irrational so we never have doubts on rounding.
A question might then be: what are the strongly connected components of this graph like? Are they of bounded size? If so, how can we bound their size from u_0,…,u_n?
(Yeah, I know, “mathematicians are like Frenchmen”. 😅)
Interesting question on MathOverflow: can we find a nonnegative Borel measure μ on ℝ^d which is not discrete and whose Fourier transform stays bounded away from 0 (meaning inf |μˆ(u)| >0)? (NB: if μ is abs. continuous, then |μˆ(u)|→0: Riemann-Lebesgue.) mathoverflow.net/q/416939/17064
Of course if μ is discrete this is easy to find (just take μ=δ₀, so δ₀ˆ = 1). But note that in ℝ the Fourier transform δ_a + δ_b + δ_c (i.e., exp(2iπ(a−u)) + exp(2iπ(b−u)) + exp(2iπ(c−u))) comes arbitrarily close to 0 if (b−a)/(c−a) is irrational.
In another related (and linked) question, I learned that if μ is the “uniform” measure on a Cantor set with dissection ratio θ, so that μˆ(u) = ∏_k cos(θ^k·u), then |μˆ(u)|→0 except if θ is a Pisot-Vijayaraghavan number (Erdős-Salem, no ref given). mathoverflow.net/q/98405/17064
So, here's a good approximation of a regular pentagon on a square lattice (to save you the bother of counting, the vertex positions are, up to translation: (0, 55), (−29, 34), (−18, 0), (18, 0), (29, 34)). How did I find it? 🧵⤵️ •1/18
Easy, you say: just pick any regular pentagon in the plane, and replace each one of its vertices by the closest lattice point! Well… yes, but that will give a very shitty approximation. To be honest, I didn't bother to compute how shitty, … •2/18
… but if you don't use the freedom to choose the size and orientation of your near-regular pentagon arbitrarily in the lattice, you won't do a good job (much like if you try to approximate ξ by rational p/q, you shouldn't just pick q arbitrarily and take p close to qξ). •3/18
We’d like to inform you that [redacted dot com], a leading academic platform for researchers, has just released the 2022 Edition of our Ranking of Top 1000 Scientists in the field of Mathematics.
The ranking is based on the H-index metric provided by Microsoft Academic and includes only leading scientists with an H-index of at least 30 for academic publications made in the area of Mathematics.
The full world ranking is available here: [redacted URL]
The annual release of our ranking is regularly featured by leading universities worldwide, including the University of Maryland, Stanford Robotics Lab, or the University of Texas, and it is a great opportunity to explore where leading experts are heading.
Full disclosure: at the time, I may have been part of the cult-like following around RMS that Laurent mentions.
In particular, I remember leading a group of free software geeks around a (ahem) “Linux” expo in the late 1990's (still fairly rare then, at least in France), chanting: “Join us now and share the software: you'll be free, hackers, you'll be free!” gnu.org/music/free-sof…