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

Finding the elements of a group given two generators and relations

  1. Oct 6, 2013 #1
    Hey everyone

    Let's say I have two generators, [itex]a[/itex] and [itex]b[/itex], with the following relations:

    Where E is the Identity element.

    What I've done so far is this - the number of elements of the group is the product of the exponents of both generators, which is 10. Then I listed all the elements and their products which aren't equal to Identity. So I got the group elements:

    {[itex]{E , a , a^{2} , a^{3} , a^{4} , ab , a^{2}b , a^{3}b , a^{4}b , b}[/itex]}

    So is that right? The only thing that's confusing me is the second relation; not sure why its there really if u can get the group elements without it.

    Thanks guys!
  2. jcsd
  3. Oct 6, 2013 #2


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    You are missing a ton of elements - a and b do not commute, so a2b is not the same as aba for example. Basically if all you had was a5=b2 = E, then you would have everything of the form
    ak1bak2bak3b....aknb for any choice of k1,...,kn (and number n) as distinct elements. The second relation will make some of them equal to each other.
  4. Oct 6, 2013 #3
    okay, so I have two questions:

    (1) How do u know that a and b don't commute? I mean I couldnt infer that from the context of the question
    (2) do I just need elements like ba, ba^2, ba^3 etc...since ab is not = ba?

    I always thought that the order of the group is the product of the powers of its generators - so that's not generally true then?
  5. Oct 6, 2013 #4


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    When they list relations, those are the ONLY relations that a and b satisfy (along with anything that can be generated by stringing together those relations). a and b commuting is a relation:
    ab = ba.

    If they don't tell you that, or things from which you can infer this, then it's not true. And yes, you will also need elements that start with b (and elements that start and end with b, and elements that start and end with a).

    Your statement about the order of a group is only true for an abelian group, if I give you a group generated by a and by with a2 = b2 = E and no other relations, this group is infinitely large, with elements a, ab, aba, abab, ababa,.... and b, ba, bab, baba, etc.
  6. Oct 6, 2013 #5
    Okay, thanks for the explanation :) but....where do I stop with writing the elements?
  7. Oct 6, 2013 #6


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    If there are infinitely many elements in a set, then you just can't write them all down. If the relations happen to make your set finite, then you stop when you've written them all down...

    You have to look at that extra relation that you have and determine how it lets you re-write some of these elements.... as a warmup, how would you use it to re-write ba in terms of a and b only and in that order (a's, then b's)) (no inverses)
    Last edited: Oct 6, 2013
  8. Oct 6, 2013 #7
    I've been trying for ages and I don't know how i can re-write ba in terms of a and b....I keep getting random stuff and I dont even know if its right. Is this an infinite group...?
  9. Oct 6, 2013 #8
    Just to make sure...I have the relation
    So if I multiply both sides by, for example [itex]b^{-1}[/itex], do I get this:

    So basically is the order of my operations correct?
  10. Oct 6, 2013 #9
    I mean I have a lot of questions about how these operations are meant to be...a lot of stuff doesnt make sense, cos unless im wrong I just showed that b is its own inverse...so something's not right
  11. Oct 6, 2013 #10


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    OK, it might help to start even more basic. Given that b2 = E, what is b-1 in terms of b? Can you calculate a similarly?
  12. Oct 6, 2013 #11
    Well it depends...cos if what im doing is wrong then that might be the problem. Check this out:
    [itex]b^{2}=E[/itex] Good so far

    But..that's wrong isnt it :S

    and as far as [itex]a[/itex] goes, I'm not sure if I can do any of the regular power manipulation with it, cos I dont know the composition rule of the group. Can I still assume that, for example, the square root of [itex]a^{5}[/itex] is basically [itex]a^{5/2}[/itex]?
  13. Oct 6, 2013 #12
    I've managed to derive a handful of other relations, which are eliminating elements of the form aba, aba^2 etc from the group......if that's correct, I think ive reached a maximum number of 14 elements in this group, all other things like ababa seem to be simplifying to preceding elements
  14. Oct 6, 2013 #13
    I do wana thank you for your time though, i really really appreciate it. You're helping me a lot :D
  15. Oct 6, 2013 #14
    Okay, this is where I've reached.
    I used the relation [itex]bab^{-1}=a^{-1}[/itex] to show that

    Now, if that is correct (please tell me if it is) then I cant find any more unique elements than the following ones

    [itex]{E, a, a^{2}, a^{3}, a^{4}, b, ba, ba^{2}, ba^{3}, ba^{4}, ab, ab^{2}, ab^{3}, ab^{4}}[/itex]

    which has an order of 14.
  16. Oct 6, 2013 #15


    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    This looks perfectly good for b... can't you do something similar for a?

    a5 = E
    a5 a-1 = a-1
    a4 = a-1.

    Then bab-1 = a-1 turns into
    ba = a-1b = a4 b.

    So this basically tells you how to "commute" a and b, which is another way to approach the problem. However your way also works (modulo a couple mistakes I think you made), once we have
    aba = b

    then we know that any element which has an a to both the right and the left of b we can reduce using this. Therefore the only possible elements are

    where i=0 or 1, k = 0,1,2,3,4 and j=0 or 1

    again it's critical here that a-1 and b-1 can be written in terms of a and b respectively, otherwise we would have elements that have a bunch of inverses in them as well. Over all choices of i,j and k there are 20 possible elements, and I think they're all distinct but I could be wrong.
  17. Oct 7, 2013 #16
    But I don't need to have inverses in the group right? T_T
  18. Oct 7, 2013 #17
    I understand this a ton better now. Watch me show off to my friends Y'ALL ARE WRONG LOLZ
  19. Oct 7, 2013 #18
    ahem. I just woke up, excuse me
  20. Oct 7, 2013 #19
    Most of this is said above, but to synthesize:
    - [itex]a^{-1}=a^4[/itex] and [itex]b^{-1}=b[/itex], so that every element can be expressed using only nonnegative powers.
    - [itex]aba=b[/itex], and so any element can be expressed in the form [itex]b^ia^jb^k[/itex] for some [itex]i,j,k\geq 0[/itex].
    - [itex]a^5=b^2=e[/itex], and so we can always take [itex]i,k\in \{0,1\}, j\in\{0,1,2,3,4\}[/itex].
    - If [itex]j=0[/itex], then there's further redundancy, as [itex]b^ib^k[/itex] depends only on whether [itex]i+k[/itex] is even or odd (given [itex]b^2=e[/itex]).
    Therefore we can write [itex]G=\{b^ia^jb^k: \enspace i=0,1, \enspace j=1,2,3,4, \enspace k=0,1 \}\cup\{e,b\}[/itex].

    I think there's no redundancy here, so that [itex]|G|= 2\cdot 4\cdot 2 + 2 = 18.[/itex]
  21. Oct 7, 2013 #20


    User Avatar
    Homework Helper

    There are 10 elements; the group is [itex]D_5[/itex], the symmetry group of the regular pentagon.

    Briefly, you have [itex]ba = a^4b[/itex], from which it follows that [itex]ba^k = a^{5-k}b[/itex] for [itex]k \in \{1,2,3,4,5\}[/itex]. So every element is expressible in the form [itex]a^jb^k[/itex] and the 10 distinct elements are [itex]\{e, a, a^2, a^3, a^4, b, ab, a^2b, a^3b, a^4b\}[/itex].
    Last edited: Oct 7, 2013
  22. Oct 7, 2013 #21


    User Avatar
    Homework Helper

    Lagrange's theorem and [itex]a^5 = e[/itex] say otherwise.
  23. Oct 7, 2013 #22
    You're right. That's embarrassing.
  24. Oct 7, 2013 #23
    Okay....I'm a little confused now...
  25. Oct 7, 2013 #24
    Comment #20 is exactly correct.
  26. Oct 7, 2013 #25
    Okay, cos I got 14 elements, but let me double check to see if they can be reduced to 10. I have another question, though - is this just a coincidence? I mean that the order of this group is 10 - the product of the powers of its generators - even though it's non-abelian?
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook