If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains *.kastatic.org and *.kasandbox.org are unblocked.

Main content

Power series intro

Power series is a sum of terms of the general form aₙ(x-a)ⁿ. Whether the series converges or diverges, and the value it converges to, depend on the chosen x-value, which makes power series a function. Created by Sal Khan.

Want to join the conversation?

  • primosaur seedling style avatar for user john.doe.13896
    How are there any applications for this? A function, whose x-values are restricted between -1 and 1 doesnt seem to be very useful?
    (10 votes)
    Default Khan Academy avatar avatar for user
    • leafers seed style avatar for user Travis Bartholome
      A couple points on that:

      1. Not all functions have such a small radius of convergence. The power series for sin(x), for example, converges for all real values of x. That gives you a way to calculate sin(x) for any value using nothing but a polynomial, which is an extremely powerful concept (especially given that we can't just evaluate a number like sin(47) because 47 doesn't fit nicely with the periodicity of the sine function; we have to use some other method that we can actually calculate, which generally boils down to polynomial approximations -- power series).

      2. Even for functions with small radii of convergence, power series still give us the ability to calculate values that would otherwise be unapproachable. The series for ln(x) centered at x=1 converges only over a radius of 1, but for calculating a number like ln(0.36), it's obviously still useful.

      3. We can just shift the center of our power series if we want to approximate a value outside the interval of convergence. For example, to calculate ln(5), we could use a power series for ln(x) centered at x=e^2 instead of x=1, which would put x=5 inside our interval of convergence.

      In short, power series offer a way to calculate the values of functions that transcend addition, subtraction, multiplication, and division -- and they let us do that using only those four operations. That gives us, among other things, a way to program machines to calculate values of functions like sin(x) and sqrt(x).

      Hope that helps.
      (39 votes)
  • leaf red style avatar for user Noble Mushtak
    1.) Is a polynomial a type of power series where a_n is 0 for certain n (it has to be 0 because a polynomial has a finite number of terms)?

    2.) Could the geometric series be generalized to the summation from n=0 to Infinity of a*(x-c)^n?
    (7 votes)
    Default Khan Academy avatar avatar for user
  • leaf green style avatar for user Tan Weisiang
    Hey guys, I had trouble of understanding what Power Series is. What is this power series for? is it just expanding a given function or something? your reply will be helpful thanks!
    (10 votes)
    Default Khan Academy avatar avatar for user
  • blobby green style avatar for user Alice Turowicz
    This tells how to find the interval of convergence for a geometric series but doesn't explain how to do so for a power series, which is what i thought this video was supposed to be about......
    (3 votes)
    Default Khan Academy avatar avatar for user
    • leaf blue style avatar for user Stefen
      It was subtle and easy to miss, but at Sal mentions that Geometric Series is a special case of the Power Series where the common ratio is an x, rather than an r. This is important, he is saying that geometric series, while you may not have thought about them as power series, or even as a representation of a function, they are, and that when you analyse a geometric series, it is just a special case of a power series. In fact, this is the usual method of introducing power series, by developing the concept of a geometric series, and applying that when meeting the power series with the "good news" that the general formula a/(1-r) becomes a/(1-x) and can be used to determine the radius of convergence.
      The following videos will make this point clearer.
      (7 votes)
  • piceratops ultimate style avatar for user giobrach
    At , Sal says that the first term in the series will simplify to a_0. Is that still true when x = c? Wasn't 0^0 an indeterminate form?
    (4 votes)
    Default Khan Academy avatar avatar for user
    • piceratops ultimate style avatar for user Monte
      You are correct that 0^0 is an indeterminate form! Three things that come to mind:
      1) the limit as x --> 0 for x^0 is 1.
      2) 0^0 does not exist, because it is indeterminate (only the limit exists)
      3) if x-c = 0 then all of the rest of the series terms would go to zero
      So if I had to make an educated guess, I would say that there are two possible outcomes:
      1) we start our summation at n = 0 and then the series fails, because it contains an indeterminate form
      2) we start our summation at n = 1 and then the series = 0
      (2 votes)
  • winston default style avatar for user John Shahki
    Whats the difference between a power series as a function of x and a geometric series as a function of x?
    (3 votes)
    Default Khan Academy avatar avatar for user
  • blobby green style avatar for user kat
    At around the mark, if the interval of convergence included 1 or -1, would the radius of convergence still be 1?
    (2 votes)
    Default Khan Academy avatar avatar for user
  • mr pants teal style avatar for user SanFranGiants
    So is radius always half the difference in values for the interval?
    (4 votes)
    Default Khan Academy avatar avatar for user
    • piceratops ultimate style avatar for user Matthew Chen
      Yes, assuming the absolute value of the values are equal. For example if the interval of convergence is -2 < x < 1 and c = 0, the radius of convergence is actually 1, since the series converges as long as x is within 1 of 0. The series does not converge as long as x is within 2 of 0, since x = 2 will not converge.
      (1 vote)
  • blobby green style avatar for user donovan.capsimalis
    I'm confused how it simplifies to a/(1-x) because if you run it with say x = .5 it doesn't work
    (3 votes)
    Default Khan Academy avatar avatar for user
  • starky tree style avatar for user Callisto
    What is center of convergence specifically?
    (2 votes)
    Default Khan Academy avatar avatar for user
    • blobby green style avatar for user tammyla101
      The center of convergence is where the distance from the lowest point to a specific number(the center) is the same as the distance from the highest point to a specific number(the center). Another word for the distance is the radius of convergence. Example: the center of convergence of the interval -1<x<1 is 0, because the radius is 1. You can find the center by subtracting the bigger number of the interval by the smaller number of the interval and then dividing by 2. Example: Center of -1<x<1= (1-(-1))/2=2/2=1.
      (3 votes)

Video transcript

We've already seen many examples of infinite series. But what's exciting about what we're about to do in this video is we're going to use infinite series to define a function. And the most common one that you will see in your mathematical careers is the power series. And I'm about to write a general case of the power series. So I could imagine a function, f of x, being defined as the infinite sum. So going from n equals 0 to infinity of a sub n-- so a sub n is just going to be the coefficient on each term-- times our variable x minus some constant c. You could almost imagine this is shifting our function to the n-th power. So if I were to expand this out, I have my first term's coefficient, a sub 0, times x minus c to the 0-th power, plus a sub 1 times x minus c to the first power. This one, of course, will simplify just a sub 0. This would simplify to a sub 1 times x minus c plus a sub 2 times x minus c squared. And I could just keep going on and on and on. Now, when you see this, you might say, aren't our geometric series, don't those look like a special case of a power series if our common ratio was an x instead of an r in that case, or if our common ratio was a variable, I guess I could say? And you would be right. That absolutely would be the case. So a geometric series. So let's just think about defining a function in terms of a geometric series. And of course, we don't have to use x all the time as the independent variable, but this is kind of the most typical convention. I guess we could also use r as an independent variable if we wanted as well. But let's imagine a function g of x. We could have g of r if we wanted, but let's do g of x is equal to the sum from n equals 0 to infinity of a times x to the n. So this is kind of a typical geometric series here. And what's the difference between this and this? Well, the difference is is here, for every term we're going to have the same coefficient a, while over here we have a sub n. We're multiplying by a different thing every time up here. We're multiplying by the same thing over here. And in this case, this particular geometric series I just made, instead of having x minus c to the n, we have just x to the n. So you could say, well, this is a special case when c is equal to 0. And we can expand it out. This is a times x to the 0, which is just going to be a, plus a times x to the first, plus a times x squared. And we just go on and on and on forever. Now, what's exciting about this is we know that this, under certain conditions, will actually give us a finite value. This will actually converge. This will actually, I guess, give us a sensical answer. So under what conditions does that happen? Well, this converges if each of these terms gets smaller and smaller and smaller. And each of these terms gets smaller and smaller and smaller if the absolute value of our common ratio is less than 1. So let me write that down. So this converges if the absolute value of our common ratio is less than 1. Or another way of thinking about it, this is another way of saying that x is in the interval between-- it's less than 1 and it is greater than negative 1. And this term right over here, now x is a variable. x can vary between those values. We're defining a function in terms of x. We call this the interval of convergence. And so we know that if x is in this interval, this is going to give us a finite sum. And we know what that finite sum is. It's going to be equal to-- if it converges. So if it converges, this is going to be equal to our first term, which is just a-- this simplifies to a right over here-- over 1 minus our common ratio. What's our common ratio? Our common ratio in this example is x. Going from one term to the next, we're just multiplying by x. We're just multiplying by x right over there. Now, this is pretty neat, because we're going to be able to use this fact to put more traditionally-defined functions into this form, and then try to expand them out using a geometric series. And this whole idea of using power series, or in this special case, geometric series to represent functions, has all sorts of applications in engineering and finance. Using a finite number of terms of these series, you could kind of approximate the functions in a way that's simpler for the human brain to understand, or maybe a simpler way to manipulate in some way. But what's interesting here is instead of just going from the sum to-- instead of going from this expanded-out version to this kind of finite value, we're now going to start being able to take something in this form and expand it out into a geometric series. But we have to be careful to make sure that we're only doing it over the interval of convergence. This is only going to be true over the interval of convergence. Now, one other term you might see in your mathematical career is a radius. Radius of convergence. And this is how far-- up to what value, but not including this value. So as long as our x value stays less than a certain amount from our c value, then this thing will converge. Now in this case, our c value is 0. So we could ask ourselves a question. As long as x stays within some value of 0, this thing is going to converge. Well, you see it right over here. As long as x stays within one of 0. It can't go all the way to 1, but as long as it stays less than 1, or as long as it stays greater than negative 1. It can stray anything less than one away from 0, either in the positive direction or the negative direction. Then this thing will still converge. So we could say that our radius of convergence is equal to 1. Another way to think about it, our interval of convergence-- we're going from negative 1 to 1, not including those two boundaries, so our interval is 2. So our radius of convergence is half of that. As long as x stays within one of 0, and that's the same thing as saying this right over here, this series is going to converge.