• iAvicenna@lemmy.world
    link
    fedilink
    English
    arrow-up
    1
    ·
    edit-2
    1 day ago

    So I call an infinite dimensional vector space of countable/uncountable dimensions if it has a countable and uncountable basis. What is the analytical definition? Or do you mean basis in the sense of topology?

    • gandalf_der_12te@discuss.tchncs.de
      link
      fedilink
      English
      arrow-up
      2
      ·
      1 day ago

      Uhm, i remember there’s two definitions for basis.

      The basis in linear algebra says that you can compose every vector v as a finite sum v = sum over i from 1 to N of a_i * v_i, where a_i are arbitrary coefficients

      The basis in analysis says that you can compose every vector v as an infinite sum v = sum over i from 1 to infinity of a_i * v_i. So that makes a convergent series. It requires that a topology is defined on the vector space fist, so convergence becomes well-defined. We call such a vector space of countably infinite dimension if such a basis (v_1, v_2, …) exists that every vector v can be represented as a convergent series.

      • iAvicenna@lemmy.world
        link
        fedilink
        English
        arrow-up
        2
        ·
        22 hours ago

        Ah that makes sense, regular definition of basis is not much of use in infinite dimension anyways as far as I recall. Wonder if differentiability is required for what you said since polynomials on compact domains (probably required for uniform convergence or sth) would also work for cont functions I think.

        • gandalf_der_12te@discuss.tchncs.de
          link
          fedilink
          English
          arrow-up
          1
          ·
          9 hours ago

          regular definition of basis is not much of use in infinite dimension anyways as far as I recall.

          yeah, that’s exactly why we have an alternative definition for that :D

          Wonder if differentiability is required for what you said since polynomials on compact domains (probably required for uniform convergence or sth) would also work for cont functions I think.

          Differentiability is not required; what is required is a topology, i.e. a definition of convergence to make sure the infinite series are well-defined.