Finding the value of a logarithm by hand can feel impossible without a calculator. Yet, mathematicians routinely approximate complex transcendental functions using polynomials. The Taylor expansion of the natural logarithm centered at $x = 1$—historically known as the Mercator series—turns the intimidating function $ln(x)$ into a clean, easy-to-evaluate sum of polynomial terms.
If you need to calculate logarithmic values near $1$ manually, or if you're trying to understand how infinite polynomial series model non-polynomial functions, this specific expansion is your primary tool. By centering the series at $x = 1$, we can evaluate points very close to $1$ with just a few terms, as the error shrinks rapidly for inputs near the center point.
Understanding the Foundation
To build a Taylor series (or more specifically, a Maclaurin series when shifted), we need a function that is infinitely differentiable at our center point, $a = 1$. Let our function be $f(x) = ln(x)$.
The general formula for a Taylor series centered at $x = a$ is: To explore the bigger picture, we recommend the recent article by The New York Times.
$$f(x) = \sum_{n=0}^{\infty} \frac{f^{(n)}(a)}{n!} (x - a)^n$$
For our specific case, where $a = 1$:
$$f(x) = f(1) + f'(1)(x - 1) + \frac{f''(1)}{2!}(x - 1)^2 + \frac{f'''(1)}{3!}(x - 1)^3 + \dots$$
Let's compute the first few derivatives of $f(x) = ln(x)$ and evaluate each one at $x = 1$:
- $f(x) = ln(x) \implies f(1) = ln(1) = 0$
- $f'(x) = \frac{1}{x} \implies f'(1) = \frac{1}{1} = 1$
- $f''(x) = -\frac{1}{x^2} \implies f''(1) = -1$
- $f'''(x) = \frac{2}{x^3} \implies f'''(1) = 2$
- $f^{(4)}(x) = -\frac{6}{x^4} \implies f^{(4)}(1) = -6$
Notice the alternating signs and the factorial pattern emerging in the numerators. The general pattern for the $n$-th derivative evaluated at $1$ (for $n \ge 1$) is:
$$f^{(n)}(1) = (-1)^{n-1} (n - 1)!$$
Deriving the Series Step-by-Step
Now we substitute our evaluated derivatives back into the Taylor series formula. Remember that $f(1) = 0$, so the constant term disappears.
$$ln(x) = 0 + 1(x - 1) + \frac{-1}{2!}(x - 1)^2 + \frac{2}{3!}(x - 1)^3 + \frac{-6}{4!}(x - 1)^4 + \dots$$
Let's simplify the coefficients by expanding the factorials:
- The second-term coefficient: $\frac{-1}{2!} = -\frac{1}{2}$
- The third-term coefficient: $\frac{2}{3!} = \frac{2}{6} = \frac{1}{3}$
- The fourth-term coefficient: $\frac{-6}{4!} = -\frac{6}{24} = -\frac{1}{4}$
Putting it all together, the expansion becomes:
$$ln(x) = (x - 1) - \frac{(x - 1)^2}{2} + \frac{(x - 1)^3}{3} - \frac{(x - 1)^4}{4} + \dots$$
We can express this compactly using summation notation:
$$ln(x) = \sum_{n=1}^{\infty} \frac{(-1)^{n-1}}{n} (x - 1)^n$$
This representation is clean, elegant, and exceptionally useful for numbers close to $1$. If you want to dive deeper into the broader theory of power series, you can review the Wolfram MathWorld Calculus Resources for additional analytical proofs.
Radius and Interval of Convergence
You can't just plug any positive number into an infinite series and expect a valid answer. We must determine where this series actually converges.
To find the radius of convergence, we apply the Ratio Test. Let's look at the absolute value of the $n$-th term, $a_n = \frac{(x-1)^n}{n}$.
$$\lim_{n \to \infty} \left| \frac{a_{n+1}}{a_n} \right| = \lim_{n \to \infty} \left| \frac{(x-1)^{n+1}}{n+1} \cdot \frac{n}{(x-1)^n} \right|$$
$$\lim_{n \to \infty} |x - 1| \left( \frac{n}{n + 1} \right) = |x - 1|$$
For the series to converge absolutely, the limit must be strictly less than $1$:
$$|x - 1| < 1$$
This inequality gives us our radius of convergence $R = 1$, and the open interval of convergence:
$$0 < x < 2$$
Let's test the endpoints of this interval to see what happens:
- At $x = 0$: The series becomes $\sum_{n=1}^{\infty} \frac{(-1)^{n-1}}{n}(-1)^n$, which diverges (it's the negative harmonic series, or rather, it diverges to negative infinity because $ln(0)$ is undefined). More precisely, substituting $x=0$ yields $\sum \frac{-1}{n}$, which diverges by the Alternating Series Test failure or direct comparison to the harmonic series.
- At $x = 2$: The series becomes $\sum_{n=1}^{\infty} \frac{(-1)^{n-1}(1)^n}{n} = 1 - \frac{1}{2} + \frac{1}{3} - \frac{1}{4} + \dots$, which is the well-known alternating harmonic series. This converges conditionally to $ln(2)$.
Thus, the complete interval of convergence is $0 < x \le 2$. If you try to use this specific expansion for $x = 3$, the terms will grow infinitely large instead of shrinking, rendering the approximation useless.
Practical Calculation Example
Let's see how well this series performs in practice. Say we want to approximate $ln(1.2)$ using just the first three terms of our Taylor expansion.
Here, $x = 1.2$, which means $x - 1 = 0.2$. Let's plug this into our polynomial:
- First term ($n=1$): $0.2$
- Second term ($n=2$): $-\frac{(0.2)^2}{2} = -\frac{0.04}{2} = -0.02$
- Third term ($n=3$): $\frac{(0.3)^3}{3}$ — wait, $(0.2)^3 = 0.008$. So $\frac{0.008}{3} \approx 0.002667$.
Let's sum them up:
$$ln(1.2) \approx 0.2 - 0.02 + 0.002667 = 0.182667$$
Let's check this against the true value of $ln(1.2)$ (which is approximately $0.1823215$). Our simple three-term polynomial gave us a result accurate to three decimal places with almost no computational effort. If you need higher precision, you simply add more terms from the sequence.
Common Mistakes to Avoid
When working with polynomial expansions, small setup errors ruin the entire calculation. Watch out for these pitfalls:
- Forgetting the center shift: Don't confuse this expansion with a Maclaurin series centered at $x = 0$. If you plug $x$ directly into $\sum \frac{(-1)^{n-1}}{n} x^n$, you are actually expanding a different function, or failing to center it correctly. Always ensure you are using $(x - 1)$ terms.
- Sign errors in alternating terms: The first term is positive, the second is negative, the third is positive, and so on. Mixing up the $(-1)^{n-1}$ term will invert your alternating pattern and yield completely incorrect numbers.
- Ignoring the convergence boundary: Trying to evaluate $ln(4)$ using this series won't work because $4$ falls far outside the interval $(0, 2]$. If you need to evaluate logarithms for numbers much larger than $2$, use logarithmic identity rules (like $ln(4) = 2 \cdot ln(2)$) or shift your center point.
For broader mathematical context on series convergence and real analysis, you can cross-reference the OpenStax Calculus Volume 2 Textbook, which offers extensive problem sets on power series.
Frequently Asked Questions
Why can't we center the Taylor expansion of ln(x) at x = 0?
The natural logarithm function is undefined at $x = 0$ (since $ln(0)$ approaches negative infinity), and it has no vertical tangent or derivative there. Because derivatives must exist at the center point $a$, a Taylor series cannot be built around $0$ for this function.
How many terms do I need for high precision?
It depends entirely on how close your $x$ value is to $1$. If $x = 1.05$, just two or three terms will give you remarkable accuracy. If $x = 1.9$ (near the edge of the convergence interval), the terms shrink very slowly, requiring dozens or hundreds of terms to achieve tight precision.
What is the connection between this series and the Mercator series?
They are the exact same thing. Nicholas Mercator discovered this logarithmic series in the seventeenth century before Brook Taylor published his general theorem for polynomial expansions.
Can this series calculate logarithms for bases other than e?
Yes. Once you calculate the natural logarithm using the series, you can change the base using the standard change-of-base formula: $\log_b(x) = \frac{ln(x)}{ln(b)}$.
