Tweetovi

Blokirali ste korisnika/cu @taktoa1

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @taktoa1

  1. 27. sij

    Is there an algorithm that runs faster than O(mn^2) for detecting if there is at least one pair of terms of size m chosen from two sets of size n that are unifiable? You can assume that each variable is used only once across all terms (e.g.: terms are linear).

    Poništi
  2. 17. sij

    The linear canonical transform can be computed efficiently (O(n log n)) in the discrete setting:

    Prikaži ovu nit
    Poništi
  3. 17. sij

    In fact, there is a thing called the linear canonical transform that is parameterized by elements of SL₂(C); rotation matrices give the fractional FT, shears give the Fresnel transform, and complex rotation gives the fractional Laplace transform.

    Prikaži ovu nit
    Poništi
  4. 17. sij

    The Fourier transform (FT) has F(F(f)) = -f and F(F(F(F(f)))) = f. Sound familiar? Turns out you can generalize FT to the fractional FT, which does any rotation, not just 90 degree ones.

    Prikaži ovu nit
    Poništi
  5. 12. sij

    Has anyone studied the relationship between classes of linear systems that are easily solved (e.g.: tridiagonal matrices) and the set of graphs that are isomorphic to a graph that has an adjacency matrix in that class?

    Poništi
  6. 2. sij

    The fact that the set of smoothed polynomial-time algorithms coincides with the set of pseudo-polynomial time algorithms is one of the most surprising facts I know.

    Poništi
  7. 2. sij

    Sorry, convexity is f(ax + (1 - a)y) ≤ a f(x) + (1 - a) f(y). And the second derivative is nonnegative everywhere, not positive (that is strict convexity).

    Prikaži ovu nit
    Poništi
  8. 2. sij

    Surprisingly, the polynomial time algorithm for minimizing submodular functions is black-box. it only needs to be able to call the submodular function, despite the fact that minimizing submodular functions is a generalization of many combinatorial optimization problems.

    Prikaži ovu nit
    Poništi
  9. 2. sij

    If f is submodular, then its Lovasz extension is a convex function (i.e.: f(ax + (1 - a)y) = a f(x) + (1 - a) f(y)). Convex functions are easy to minimize, because their local minimum is a global minimum (second derivative positive everywhere)!

    Prikaži ovu nit
    Poništi
  10. 2. sij

    Any submodular function can be thought of as a function defined on vertices of an |X|-dimensional hypercube. The Lovasz extension of a submodular function extends it to a function on the /volume/ of the |X|-dimensional hypercube.

    Prikaži ovu nit
    Poništi
  11. 2. sij

    A function f : P(X) → R is submodular if for all A ⊆ B ⊆ X and x ∉ B, f(A ∪ {x}) - f(A) ≥ f(B ∪ {x}) - f(B). This is a law of diminishing returns -- adding x to B is less effective than adding x to A if A ⊆ B. These functions can be minimized in strongly polynomial time.

    Prikaži ovu nit
    Poništi
  12. 7. pro 2019.

    There is a 0.878-approximation algorithm for max cut using semidefinite programming. Much less well known is that there is a 0.614-approximation algorithm using spectral techniques. I wonder how well the latter does in practice; it seems a lot simpler to implement.

    Poništi
  13. 2. pro 2019.

    Computing the Fiedler vector of a matrix is like 2 lines of code using the eigSH function. Computing the Laplacian of a graph and partitioning a graph based on the sign of the elements of the Laplacian's Fiedler vector is a few more.

    Prikaži ovu nit
    Poništi
  14. 2. pro 2019.

    It is ridiculously easy to implement min-cut via spectral partitioning with hmatrix. Not much reason to bother with the classic min-cut algorithms.

    Prikaži ovu nit
    Poništi
  15. 27. stu 2019.

    where X ≤ Y means "there is a substitution s such that s(Y) = X", of course.

    Prikaži ovu nit
    Poništi
  16. 27. stu 2019.

    I know unification is the meet in the subsumption lattice, but I'm having difficulty disproving "X unifies with Y iff X ≤ Y or Y ≤ X". anyone know a pair of first-order terms that have a unification but don't satisfy that?

    Prikaži ovu nit
    Poništi
  17. proslijedio/la je Tweet
    20. stu 2019.
    Odgovor korisniku/ci

    That's a good point, it's actually a homomorphism that preserves two unrelated monoids. Or, if you define degree as going to (N u {-∞}, max, +), where the zero polynomial has degree -∞ (but other constant polynomials have degree 0), then it really is a semiring homomorphism

    Poništi
  18. 19. stu 2019.

    it's neat that the function that computes the degree of a polynomial is a semiring homomorphism from the ring of polynomials to the (N, max, +) semiring

    Poništi
  19. proslijedio/la je Tweet

    The whole country has suffered immensely from the incentive- and culture-shift that caused most schools to remove an array of shop and trade classes

    Prikaži ovu nit
    Poništi
  20. proslijedio/la je Tweet
    17. ruj 2019.

    It’d be neat if you could add flag names to WARNING pragmas so users could control groups of them specifically: {-# WARNING ["importing this other module instead is prolly a good bet"] "fused-effects-import-advice" #-} -Wno-fused-effects-import-advice

    Prikaži ovu nit
    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·