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

Impossible proof

  1. May 30, 2010 #1
    A question at the end of a chapter in a book I'm reading asked me to prove something which I'm thinking is a mistake because it seems to be impossible to prove.

    'Prove that if two homogeneous systems of linear equations in two unknowns have the same solutions, then they are equivalent.'

    The book defines equivalence to be 'two systems of linear equations are equivalent if each equation in each system is a linear combination of the equations in the other system'.

    I can think of two systems which share the same solution (the trivial solution) that are not equivalent. That is the second system of homogeneous linear equations are not a combination of the first system of homogeneous linear equations but they both share the same solution.

    system 1,
    x - y = 0
    2x + y = 0

    system 2,
    3x + y = 0
    x + y = 0

    Am I right in believing that their question is malformed or am I wrong in believing that you can not get x + y as a linear combination of c1(x - y) + c2(2x + y).

    Thank you,
    -= Dave
     
  2. jcsd
  3. May 30, 2010 #2
    itd be best to see whats going on geometrically ...a system of 2 linear equations in 2 variables represents 2 lines through the origin
     
  4. May 30, 2010 #3
    The latter. [itex]x+y=(-\frac{1}{3})(x-y)+(\frac{2}{3})(2x+y)[/itex].
     
  5. May 30, 2010 #4
    Thank you Martin. I spent a long time thinking about how to represent x + y as a linear combination of those two and couldn't figure it out. Now that I see the answer its pretty obvious.

    Is there a systematic approach to figuring it out or did you just do it in your head.

    Thank you,
    -= Dave
     
  6. May 30, 2010 #5

    Hurkyl

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    Just solve the equation
    x+y = c1(x-y) + c2(2x+y)​
    for c1 and c2.

    (It may help to collect all of the x's together and all of the y's together)



    P.S. it may be interesting to rewrite everything in terms of matrices and vectors.
     
    Last edited: May 30, 2010
  7. May 30, 2010 #6
    Thanks. In fact, after I wrote this post I went to lunch and I couldn't stop thinking about it and came up with the solution on a napkin. I formulated in terms of vector and matrices and solved for it.

    Now that I know how to solve for it I just have to figure out how to prove it.

    Thanks very much!
    -= Dave
     
  8. May 30, 2010 #7
    Don't want to sound stupid but is not x=y=0 the solution? :)
     
  9. May 31, 2010 #8
    No, you are wrong.

    The solution space are all vectors orthogonal to each row vector, which means that the row vectors must span the orthogonal complement to the solution space. This in turn means that for any one solution space each set of row vectors can produce each other set of row vectors for equivalent equations.

    I put a small proof in spoiler white above.
     
  10. May 31, 2010 #9

    HallsofIvy

    User Avatar
    Science Advisor

    x= y= 0 is a solution to any homogeneous set of equations but if the determinant of the coefficient matrix is 0, the set of all solutions is a subspace of [math]R^n[/math].
     
  11. May 31, 2010 #10

    No it isn't, x + y = 4

    does not have a solution of x=y=0.

    What is this matrix? I see no matrix.
     
  12. May 31, 2010 #11
    That is not a homogeneous equation.
     
  13. May 31, 2010 #12
    Why not?
    Whats wrong with it?
    Homogeneous has loads of different meanings in wikipedia, can you explain what you mean by it in simple (laymans) terms?
     
  14. May 31, 2010 #13
    Try checking the relevant maths page?
    http://en.wikipedia.org/wiki/System_of_linear_equations#Homogeneous_systems

    Now I don't really know your background so maybe you haven't studied linear algebra, then I would have taken a bit more time explaining things.
     
  15. Jun 2, 2010 #14

    uart

    User Avatar
    Science Advisor

    Yes Alice that is totally correct for both of the examples given by the OP. Both of the examples given are "full rank" so the only solution is zero.

    Homogenious in this sense simply means that the RHS is the zero vector, so a better example would be x + y = 0. Now we see that x=y=0 is still a solution, but it is no longer the only solution.
     
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook