Determining Required Degree for Given Accuracy
You'll learn how to figure out how many terms of a Taylor polynomial you need to guarantee your approximation is accurate enough.
The Core Idea
When you approximate a function with a Taylor polynomial, you want to know: "How many terms do I need to keep my error below some tolerance?" This is the flip side of bounding error—instead of calculating the error for a given polynomial, you're solving for the degree that ensures your error stays within acceptable limits.
The process uses the Lagrange Remainder formula:
|R_n(x)| ≤ M · |x - c|^(n+1) / (n+1)!
Where M is the maximum value of |f^(n+1)| on your interval.
Working Backward from Error Tolerance
Here's the strategy:
- Start with your error requirement: For example, you want
|error| < 0.0001 - Set up the inequality:
M · |x - c|^(n+1) / (n+1)! < 0.0001 - Find M: Determine the maximum of the (n+1)-th derivative on your interval
- Solve for n: Often by testing successive values until the inequality holds
Think of it like packing a suitcase: You know the weight limit (error tolerance), and you're figuring out how many items (terms) you can bring before exceeding it. Unlike packing, though, each additional term makes your "bag" lighter (more accurate), not heavier!
Practical Example
To approximate e^x at x = 0.5 with error less than 0.001, you'd test increasing values of n until e · (0.5)^(n+1) / (n+1)! < 0.001, where e ≈ 2.718 is the bound on all derivatives of e^x.
Key Takeaway: To find the required degree for a given accuracy, use the Lagrange Remainder bound and test successive values of n until your error tolerance is satisfied.