1 / 27

Recurrence Relations

Recurrence Relations . An equation that allow us to compute the n th term of a sequence from preceding terms. Example1: The selection sort In the selection sort algorithm the total number of comparisons is given by. 0 if n = 1 C n = C n-1 + n-1 for n > 1.

santiago
Download Presentation

Recurrence Relations

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Recurrence Relations • An equation that allow us to compute the nth term of a sequence from preceding terms. • Example1: The selection sort • In the selection sort algorithm the total number of comparisons is given by 0 if n = 1 Cn = Cn-1 + n-1 for n > 1 Recurrence Relations

  2. Example2: The subset recurrence • The subset recurrence: let Sn be the number of subsets of N = {1, 2, 3,….., n-1, n}, then Sncan be computed from Sn-1 (which is the number of subsets of {1, 2, …..,n-1} as follows: • Let N’ = {1, 2, …..,n-1}, Sn-1 is the number of subsets of N’. Now subsets of N are either contain N or does not. • The subsets of N that do not contain n are the subsets of N’. Their number is Sn-1. • The subsets of N that contain n are the subsets of N’ with n added to each subset. Their number is Sn-1. Recurrence Relations

  3. Example2: The subset recurrence 2 * Sn-1 for n > 1 • Thus Sn = S0 = 1 for n= 0 • where S0 is the number of subsets of the empty set. • Can we find out the first 5 terms of Sn? Recurrence Relations

  4. Example3: The Bijection Recurrence • What is the number of one-to-one onto functions from a set A = {1, 2,…., n} to a set T with n elements? • Does it depend on n? if so? • Let it be bn. We would like to find it. Recurrence Relations

  5. Example3: The Bijection Recurrence • bn can defined recursively since • for a function f • f(n) has n choices let it be t & the numbers {1, 2,…., n-1} have to be mapped to the remaining n-1 elements of T. • The number of bijections is bn-1 • Thus by the product principle there are n*bn-1 choices. That is bn=n*bn-1 • Also notice that b1 = 1 Recurrence Relations

  6. The order of the recurrence relation • In the previous examples the nth term is computed from the nth-1 term without using other previous terms such as the nth-2 term. • Such a recurrence relations is called a first order recurrence relation. • A recurrence relation is said to be of the rth order if computing the nth term requires computing the preceding r terms. • What is the order of the Fibonacci number? • f(n) = f(n-1) + f(n-2) and both f(0) & f(1) = 1 Recurrence Relations

  7. Linear recurrence relations • A first order recurrence relation is called linear recurrence relations if the nth term can be computed as • an = b(n) * an-1 + d(n) for b and d are functions of n or constants. • In general the linear recurrence equation is of the form an= b1(n)*an-1+b2(n)*an-2 + …..b0(n)*a0+ d(n) Recurrence Relations

  8. First order linear homogenous recurrence equation • A first order linear recurrence equation is called homogenous equation if d(n) = 0. • Examples: • an = n*an-1 • an = an-1+ an-2 • an = nan-1+ n2 Recurrence Relations

  9. Constant coefficient recurrence • A first order linear recurrence relation is constant coefficient recurrence if b(n) is constant. • If d(n) = 0 then the relation becomes homogenous constant coefficient recurrence. • Example: the number of subsets example is a homogenous relation • Are there other examples? Recurrence Relations

  10. Solutions of recurrence equations • A function f is a solution to a recurrence equation if substituting f for an gives a valid equation for all n. • Example: Cn = n(n-1)/2 is a solution to the recurrence relation in example1. • Notice C1 = 1*(1-1)/2 = 1*0/2 = 0 • Now Cn = Cn-1 + n-1 is also obtained since • Cn-1= (n-1)(n-1-1)/2 • Adding n-1 to both sides we get • Cn-1 + n-1 = Cn & • (n-1)(n-2)/2 + (n-1)=(n-1)(n-2)/2 + 2(n-1)/2 = (n2-n)/2 = n(n-1)/2 Recurrence Relations

  11. Solving recurrence Relation • First order linear homogenous constant coefficient recurrence relation • Let an= b*an-1 for n > m and c for n=m or am = c, then an = c*bn-m for n > m. • Proof by induction: • Base step: the statement is true for n=m • Assume it is true for n = k-1, show it is true for n = k. • Now ak-1=c*bk-1-m implies when multiplying by b • b*ak-1 = ak & b*c*bk-1-m= c*bk-m • Thus ak = c*bk-m • What is the solution for relation in example2? • Sn = 2*Sn-1 and S0=1. Thus the solution is sn = 2n since c = 1, m = 0 and b = 2 Recurrence Relations

  12. Solving recurrence Relation • Second order linear homogenous constant coefficient recurrence relation • Substitute in the given relation and simplify to get an equation in terms of r, called the Characteristic Equation of the recurrence relation. • Solve the characteristic equation for its roots let the roots be distinct say r, s. • The solution for the recurrence relation will be: an = b*rn + d*sn where b & d are constants depending on the initial conditions. Recurrence Relations

  13. Example to proof the solution to second order recurrence relation • Solve the recurrence relation: an = c1*an-1+c2*an-2 • The characteristic equation is: tn =c1*tn-1 +c2*tn-2…...** • If the solution of ** is r & s, then the solution to an = c1*an-1+c2*an-2 is an = b*rn + d*sn • Proof: substitute for an, an-1, an-2 their values and verify its correctness Recurrence Relations

  14. Using the characteristic equation for first order relation • an = 2an-1 for n > 0 & 1 for n = 0 • The characteristic equation: rn = 2rn-1 • Finding the roots are rn-1 (r – 2) = 0 • r = 0 and r = 2 • The solution is 2, thus an = c2n • c is constant obtained from the initial condition when n = 0, a0 = c*20, thus c=1 Recurrence Relations

  15. Example on second order recurrence relation • Solve the relation an = an-1 + 2*an-2 for n > 1, 4 for n = 0, and 5 for n = 1. • The characteristic equation is: tn – tn-1 – 2*tn-2 = 0 • The solution for the characteristic equation are t = 2 or -1, thus the solution for the recurrence relation is: an = b*2n + c*(-1)n • If n = 0: 4 = b * 20 + c*(-1)0 • If n = 1: 5 = b * 21 + c*(-1)1 • Solve for b & c we get b = 3 & c = 1 Recurrence Relations

  16. The sum of two solutions is a solution • If r & s are two solutions of a linear recurrence relation, then b*r + c*s is a solution of the relation where b & c are constants with values obtained from the initial conditions. • This can be verified by plug in the values of the solutions in the recurrence relation. Recurrence Relations

  17. One more example on second order recurrence relation • Solve the relation an = 6*an-1 - 9*an-2. • The characteristic equation is: tn - 6tn-1 + 9*tn-2 = 0 Its solution is t = 3 • Is 3n a solution to the relation? Try it. • How about an = b*3n?Try it. How about n*3n is it a solution? How about b*n*3n? • The solution is an = b*3n + c*n*3n where b & c are constants depending on the initial conditions Recurrence Relations

  18. Solving recurrence Relation • Solving general Linear Homogeneous Recurrence Relations with Constant coefficients where s are known constants, and there are K initial conditions where s are known constants. Recurrence Relations

  19. Solving Linear Homogeneous Steps • Substitute in the given relation and simplify to get an equation in terms of r, called the Characteristic Equation of the recurrence relation. • Solve the characteristic equation for its roots. • Foreach different root r of multiplicity m, add (to the solution of an) a new expression of the form • Solve for the unknown constants Ai s from the given initial conditions. Recurrence Relations

  20. Example • Find the solution of an = 3an−1 − 4an−3, n >2, a0 = 0, a1 = 6, a2 = 24 • First, find characteristic equation: 0 = [r3 − 3r2 + 4]*rn-3 • Find the roots =(r − 2)(r − 2)(r + 1) • We get the roots r1 & r2 = 2, r3 = −1 • The solution is: an = A02n+A1n2n+A2(-1)n • Now solve for n = 0, 1,& 2 we get the equations: 0 = A0+A2, 6 = 2A0+2A1-A2, 24 = 4A0+8A1+A2 • Solve for A0, A1, A2: 0, 3, 0. Recurrence Relations

  21. Solution of first order non-constant recurrence • The bijection recurrence bn= n*bn-1 is non-constant recurrence. This relation is similar to the inductive step in the n! definition • With b1=1, we deduce that bn= n! In General • If an= f(n) an-1 for n> m & a1 = c, then an= c* • Proof by induction. Is the solution Recurrence Relations

  22. Solution of first order non-Homogenous recurrence • If an = c*an-1 + f(n), then the solution: an= an(h) + an(p) • Depending on f(n), the particular an(p) solution can be found. The following table shows some cases for f(n) and the corresponding solution. Recurrence Relations

  23. Example on non-Homogenous recurrence relation • Given the following recurrence relation an=2an-1 + 3 • So the particular solution is: an(p)= D • Substitute this back into the recurrence relation and solve for D. • Namely, D = 2D + 3. Thus D = -3. • The solution for the homogenous equation is 2 • If the particular solution is not identical to the homogenous then the solution is: an=c2n-3 Recurrence Relations

  24. A second example • Given an=2an-1+ 2n2 what is its solution? • Let an(p)=c1n2+c2n+c3 be the particular solution, solve for c1, c2, &c3 by substituting the solution in the relation. • C1*n2+c2*n+c3-2*c1*(n-1)2-2*c2*(n-1)-2*c3-2*n2 = 0 • In order for the solution to be independent of n both coefficients of n2 & n must be Zero in addition the constant term should be Zero too this yield to the following set of equations Recurrence Relations

  25. Continue the solution an=2an-1+2n2 • The equations are: • c1-2c1-2=0, • c2+4c1-2c1=0, and • c3-2c1+2c2-2c3 = 0 • Thus C1 = -2, c2 = -8, and c3 =-12 • The particular solution is: • an(p) = -2n2-8n -12 • The homogenous solution is: an(h)= c2n • Thus the solution is: an= an(p)+ an(h) Recurrence Relations

  26. One more example • Let an=2an-1+3*2n. • an(h) = c2n, an(p)=d*2n, but 2 is a root for the characteristic equation corresponding to the homogenous equation, so an(p)shouldbe = d*n*2n. • Solve for d in the initial recurrence relation to obtain d*n*2n=2d(n-1)2(n-1)+32n =dn2n-d2n+32n Thus d = 3 & hence an(p)= 3n2n. • The solution is an= c2n + 3n2n Recurrence Relations

  27. Converting to first order • Let an= c*an/2+n • Is the relation of first order? Second? • Can we solve it? If not can we convert it to some order we know? • Let n = 2k, then n/2 = 2k/2 = 2k-1. Thus • If we let an= tk, then an/2= tk-1 & hence • The relation can be rewritten as: tk = c*tk-1+2k a first order recurrence relation. Recurrence Relations

More Related