Mathematical Methods (10/24.539)

VII. Power Series Solution Method

Example 7.3 -- Extended Power Series Solution (Dependent Solutions)

Problem Description:

Solve the following variable coefficient linear system:

Problem Solution:

In standard form, the original linear ODE becomes

This form is similar to that observed in Example 7.2, thus we can use the extended power series method, with

and the derivative relations

Substitution of these expressions into the base ODE gives

Collecting like terms gives

and, with a little algebra, this becomes

Now, working on the second term, we let m = p+2 or p = m-2. Then the second term becomes

Therefore, for , the roots of the indicial equation are simply , and it should be noted that these do differ by an integer - thus, one is warned that dependent solutions may result. Also, from the requirement that the coefficient of the xr+1 term be zero, we have a1 = 0, since for either r1 or r2 .

Now to get the desired recurrence relation between am+2 and am, we have

and

Now for r = r1 = 1, we have

Therefore, letting m = 0, 1, 2, etc., gives

In fact, for all odd m, am = 0 since a1 = 0, and for all even m > 0, am = 0 since a2 = 0. Therefore, a0 is the only non-zero term. This gives the simple result that

Performing similar manipulations for r = r2 = -1, we have

but for m = 0, this term becomes undefined (i.e.). This is clearly not allowed. Therefore, r = -1 is not a valid root. Actually, this behavior was not completely unexpected since r1 and r2 differ by an integer. In this case, the phenomenon observed here is not uncommon.

Now, to get a second linearly independent solution, let’s use the reduction of order method. In particular, let . For this specific case, the relationships of interest are

Substitution of these expressions into the defining ODE gives

Expanding and performing the indicated algebraic operations give

or

This equation is separable and it can be simplified considerably using a partial fraction expansion technique, as follows:

Separating variables gives

where the A, B, and C constants are determined from

Therefore, we have the simplified form

This expression can be integrated directly, giving

or

One final integration gives

Therefore, the second linearly independent solution becomes

Finally, the linear combination of the two independent solutions gives the desired general solution for this problem, or

10/24.539 Lecture Notes by Dr. J. R. White, UMass-Lowell (updated October 1998).

Return to Online Courses