Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Vector Space Axioms

  1. May 13, 2008 #1
    I am trying to shorten and generalize the the definition of a vector space to redefine it in such a way that only four axioms are required. The axioms must hold for all vectors u, v and w are in V and all scalars c and d.

    I believe the four would be:

    1. u + v is in V,
    2. u + 0 = u
    3. u + -u = 0
    4. cu is in V

    I believe 1 and 2 can be used to satisfy:

    u + v = v + u
    (u + v) + w = u + (v + w)

    and 3 and 4 can be used to satisfy:

    c(u + v) = cu + cv
    (c + d)u = cu + du
    c(du) = (cd)u
    1u = u

    Not sure if I am on the right track here so any suggestions or corrections would be appreciated. Thanks to all who reply.
  2. jcsd
  3. May 13, 2008 #2


    User Avatar
    Science Advisor
    Homework Helper

    Actually I think that all the axioms are necessary, and if you leave out for example the commutativity or associativity axiom you don't get what people would ordinarily call a vector space.
    If you think that you would, you should prove for example that "u + v = v + u" indeed follow from the four axioms you have, as you claim, though I wouldn't see how that could be done. In fact, I don't even see how to prove something as simple as 0 + u = u without using at least associativity ((u + v) + w = u + (v + w)) and -(-u) = u.
  4. May 13, 2008 #3


    User Avatar
    Science Advisor
    Homework Helper

    A model for axioms 1 and 2 would be:

    where 1 holds by the closure of integers under subtraction and 2 holds by the additive identity of integers. But in this model [itex]u+v\neq v+u[/itex] for most u and v.
  5. May 13, 2008 #4


    User Avatar
    Science Advisor
    Homework Helper

    Ah! If you define [itex]-v:=v[/itex] and [itex]cv:=v[/itex] in the above model, you can see that 1, 2, 3, and 4 hold but commutativity still fails in general, as does (u + v) + w = u + (v + w). (It doesn't matter here, but let c be drawn from the reals.) With an appropriate step function instead for scalar multiplication (say cv := 0 for c = 0 and v = 1 and cv := v otherwise) you can make the scalar distribution properties fail as well.
    Last edited: May 13, 2008
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook