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

Solve the difference equation?

  1. Jul 14, 2013 #1
    Solve the difference equation yn+1=sqrt((n+3)/(n+1)) yn in terms of the initial value y0.

    y1=sqrt(3)y0
    y2=sqrt(6)y0
    yn=sqrt(3n)y0

    But that's not the answer. The answer in the textbook is yn=y0sqrt(((n+2)(n+1))/2). Did I do something wrong?
     
  2. jcsd
  3. Jul 14, 2013 #2

    HallsofIvy

    User Avatar
    Staff Emeritus
    Science Advisor

    It is hard to say what you might have done wrong when you do not show what you did!

    While it is a good idea to calculate values until you can guess a pattern, you are not done until you have proved that pattern. At the very least, you should have checked your "guess". [itex]y_3= \sqrt{\frac{2+3}{2+ 1}}\sqrt{6}y_0= \sqrt{\frac{5(6)}{3}}y_0= \sqrt{10}y_0[/itex]. 10 is NOT 3n= 3(3)= 9. Certainly, you should have realized that "[itex]y_n= \sqrt{3n}y_0[/itex]" is not true when n= 0!

    Yes, [itex]y_1= \sqrt{3}y_0[/itex] and [itex]y_2= \sqrt{6}y_0[/itex]. But then [itex]y_3= \sqrt{10}y_0[/itex], as I showed, and then [itex]y_4= \sqrt{(3+3)/(3+ 1)}\sqrt{10}y_0= \sqrt{15}y_0[/itex], etc.6,

    Is that enough to make a better guess at a pattern? I notice that it is always the square root of an integer times [itex]y_0[/itex]. Further, those integers are 3, 6, 10, 15... For me, at least, that is enough to notice that the 'first differences' are 6-3= 3, 10- 6= 4, 15- 10= 5 and then the 'second differences' are 4- 3= 1 and 5- 4= 1- both 1. IF that continued to be true, that the 'second difference' is a constant, that would mean that the integers are a quadratic function of n, [itex]an^2+ bn+ c[/itex]and we can use those values to determine a, b, and c.
    I did that and got that a= 1/2, b= 3/2, and c= 1 so that the integers are [itex](1/2)(n^2+ 3n+ 2)= (1/2)(n+ 2)(n+ 1)[/itex], exactly what your textbook says.

    You can prove that is true using "proof by induction". When n= 0, [itex]y_0= \sqrt{\frac{1}{2}(0+2)(0+ 1)}y_0= \sqrt{1}y_0= y_0[/itex] is correct. Assume that, for some k, [itex]y_k= \sqrt{\frac{1}{2}(k+2)(k+ 1)}y_0[/itex]. Then, by the difference equation, [itex]y_{k+1}= \sqrt{(k+3)/(k+1)}y_k= \sqrt{\frac{1}{2}((k+3)/(k+1))(k+2)(k+1)}y_0[/itex][itex]= \sqrt{\frac{1}{2}(k+3)(k+2)}y_0= \sqrt{\frac{1}{2}((k+1)+ 2)((k+1)+ 1)}y_0[/itex].
     
    Last edited: Jul 14, 2013
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook