How To Calculate A Continued Fraction

Convert a decimal or fraction into a continued fraction, compute convergents, and visualize approximation error.

Tip: for an exact rational result, use Fraction mode. Decimal mode builds a finite approximation.

How to Calculate a Continued Fraction: Complete Expert Guide

A continued fraction is one of the most powerful ways to represent numbers, especially when you want excellent rational approximations with small denominators. If you are learning number theory, numerical methods, or just need a practical way to convert decimal values into fractions, continued fractions give a mathematically optimal method in many cases. This guide explains exactly how to calculate a continued fraction by hand and with software, how to read convergents, and how to avoid common mistakes.

What is a continued fraction?

A simple continued fraction has the form:

a₀ + 1/(a₁ + 1/(a₂ + 1/(a₃ + …)))

It is usually written in compact notation as [a₀; a₁, a₂, a₃, …]. For simple continued fractions, each a_i after a₀ is a positive integer. Rational numbers always produce finite continued fractions. Irrational numbers produce infinite continued fractions, and many famous constants have highly structured patterns:

• √2 = [1; 2, 2, 2, 2, …] repeating forever.
• e = [2; 1, 2, 1, 1, 4, 1, 1, 6, …] with a known pattern.
• π = [3; 7, 15, 1, 292, 1, 1, …] irregular but extremely useful for approximations.

Why continued fractions matter in practice

Continued fractions are not just theoretical. They are used when you need compact and accurate fractional approximations. They appear in cryptography, Diophantine equations, control systems, signal processing, and algorithm design. In practical engineering and finance tools, they help convert floating-point values into human-friendly fractions while controlling approximation error.

One key fact: convergents from continued fractions are often the best possible rational approximations for bounded denominators. That means if you want the best p/q with q not too large, continued fractions are usually where you start.

Step-by-step method: rational number to continued fraction

If your number is a fraction N/D, use the Euclidean algorithm. This is exact and finite.

1. Divide N by D: N = qD + r, where q is the integer quotient and r is the remainder.
2. The first continued-fraction term is q.
3. Replace (N, D) with (D, r).
4. Repeat until remainder becomes 0.

Example with 415/93:

1. 415 = 4×93 + 43, so a₀ = 4
2. 93 = 2×43 + 7, so a₁ = 2
3. 43 = 6×7 + 1, so a₂ = 6
4. 7 = 7×1 + 0, so a₃ = 7

Therefore 415/93 = [4; 2, 6, 7].

Step-by-step method: decimal number to continued fraction

For decimal values, especially irrationals or measured data, use repeated “integer part and reciprocal” operations:

1. Set x to your decimal value.
2. Take a = floor(x). Record a.
3. Compute fractional part f = x – a.
4. If f is very small (below your tolerance), stop.
5. Set x = 1/f and repeat.

Because computer decimals are finite precision, this method must use a stopping rule: either a tolerance or a maximum number of terms. The calculator above includes both controls so you can balance precision and stability.

Choosing the right stopping rule

• Exact fractions: use fraction mode if you already know numerator and denominator.
• Noisy measurements: stop early to avoid fitting noise in the least significant digits.
• Symbolic constants: use more terms, then inspect error reduction by convergent index.

How to compute convergents

Given coefficients [a₀; a₁, …, a_n], convergents p_k/q_k are built recursively:

• p_-2=0, p_-1=1
• q_-2=1, q_-1=0
• p_k = a_kp_k-1 + p_k-2
• q_k = a_kq_k-1 + q_k-2

Each convergent is a rational approximation to the target value. The sequence usually oscillates around the true number and converges quickly compared with many naive methods.

Comparison table: continued fraction efficiency vs decimal truncation

The numbers below illustrate a practical pattern: continued-fraction convergents often reach a given error target using much smaller denominators than decimal-place based fractions.

Constant	Method	Approximation	Denominator	Absolute Error
π	Convergent	355/113	113	2.6676×10^-7
π	4-decimal truncation fraction	31416/10000	10000	7.3464×10^-6
√2	Convergent	99/70	70	7.2152×10^-5
√2	4-decimal truncation fraction	14142/10000	10000	1.3562×10^-5
e	Convergent	193/71	71	2.8031×10^-5
e	4-decimal truncation fraction	27183/10000	10000	1.7154×10^-5

Important interpretation: decimal truncation can sometimes have smaller absolute error at a specific decimal-place cut, but it usually requires very large denominators. Continued fractions deliver exceptional denominator efficiency, which is critical in constrained systems and symbolic work.

Detailed convergence statistics for π convergents

The table below shows how quickly errors can collapse as you add terms to π’s continued fraction [3; 7, 15, 1, 292, …].

k	Convergent	Decimal Value	Absolute Error vs π
0	3/1	3.0000000000	1.4159×10^-1
1	22/7	3.1428571429	1.2645×10^-3
2	333/106	3.1415094340	8.3219×10^-5
3	355/113	3.1415929204	2.6676×10^-7
4	103993/33102	3.1415926530	5.7789×10^-10

Common mistakes when calculating continued fractions

• Mixing floor and rounding: always use floor for each step in simple continued fractions.
• Ignoring floating-point noise: in decimal mode, tiny residuals can produce huge artificial terms.
• Using too many terms on measured data: this overfits noise and creates misleading complexity.
• Forgetting sign handling: negative values need careful integer-part extraction.
• Assuming every sequence is simple: general continued fractions allow non-integer parts, but this calculator uses simple continued fractions for stability and interpretability.

Practical interpretation: what does each term mean?

Large terms indicate that the current convergent is unusually close to the target. For example, the 292 term in π explains the remarkable jump in accuracy from 355/113 to the next convergent. In optimization and approximation workflows, this helps you detect where precision gains are concentrated.

Denominator growth matters as much as error. If your system can only store denominators up to a threshold, the best usable convergent is often the last one under that threshold. This is exactly why continued fractions are favored in clock ratio design, rational resampling, and bounded-complexity hardware representations.

Algorithmic complexity and implementation notes

For rational inputs, complexity is linked to the Euclidean algorithm and is very efficient in practice. For decimal inputs, each iteration requires subtraction, reciprocal, and floor operations, which are constant-time at fixed machine precision. The chart in the calculator visualizes error by convergent index so you can decide if additional terms are worthwhile.

For production code, use these safeguards:

1. Clamp maximum terms to prevent runaway loops.
2. Use a tolerance such as 1e-12 for double precision.
3. Normalize fraction signs so denominator stays positive.
4. Validate denominator not equal to zero.
5. Display both fraction and decimal outputs for user trust.

Authoritative references for deeper study

If you want rigorous definitions, proofs, and advanced applications, these sources are strong starting points:

• NIST Digital Library of Mathematical Functions: Continued Fractions (dlmf.nist.gov)
• MIT OpenCourseWare, Theory of Numbers (ocw.mit.edu)
• Lamar University Mathematics Notes on Continued Fractions (lamar.edu)

Final takeaway

To calculate a continued fraction, repeatedly extract the integer part and invert the remainder, or use Euclidean division directly for rational inputs. Then compute convergents to get high-quality fractional approximations and inspect approximation error term by term. If your goal is accurate approximation with small denominators, continued fractions are one of the best tools available. Use the calculator above to run exact and approximate cases quickly, compare convergents, and visualize where precision improves most.