Lucas Numbers: Fibonacci's Sibling Sequence, Explained

Fibonacci numbers get all the attention, but they're one member of a much larger family of sequences built from the identical rule: start with two numbers, and let every later term be the sum of the two before it. Change the starting pair from 0 and 1 to 2 and 1, and the sequence that results — 2, 1, 3, 4, 7, 11, 18, 29, 47, 76, 123… — is the Lucas sequence, named for the 19th-century French mathematician Édouard Lucas. It isn't a curiosity riding on Fibonacci's coattails; it satisfies its own identities, has its own real application in prime-number testing, and converges to the identical golden ratio by the identical mechanism. Even its history has its own distinct, slightly grim character, quite apart from anything Fibonacci-related.

Same rule, different starting pair

The Lucas recurrence is L(n) = L(n−1) + L(n−2), starting from L(0)=2 and L(1)=1 — the only difference from Fibonacci is which two numbers seed the process. Édouard Lucas, incidentally, is also the person responsible for the name "Fibonacci sequence" itself, having popularized it in the 1870s in his own writings on recreational and number-theoretic mathematics; the sequence had circulated for centuries under other descriptions before Lucas's name for it stuck. He died in 1891 under an almost absurd set of circumstances for a mathematician remembered for elegant number theory: a shard from an accidentally dropped plate at a banquet cut his cheek, and the resulting infection killed him within days, at 49.

The identity that ties the two sequences together

The cleanest bridge between the two sequences is L(n) = F(n−1) + F(n+1) — every Lucas number equals the sum of the two Fibonacci numbers flanking it at the same index. Checking it directly: L(10) = 123, and F(9) + F(11) = 34 + 89 = 123. L(15) = 1364, and F(14) + F(16) = 377 + 987 = 1364. The identity holds at every single index from 0 through 20 without exception, including the boundary case n=0, which needs the "negafibonacci" extension F(−1)=1 to make F(−1) + F(1) = 1 + 1 = 2 = L(0) come out correctly.

A second identity connects the two sequences multiplicatively rather than additively: F(2n) = F(n)·L(n). At n=10, F(20) = 6765, and F(10)·L(10) = 55 × 123 = 6765 — exact. At n=15, F(30) = 832,040, and F(15)·L(15) = 610 × 1364 = 832,040 — exact again, confirmed at every index from 1 through 15. A third identity, L(n)² − 5·F(n)² = 4·(−1)ⁿ, links the squares of both sequences to a value that alternates between exactly 4 and exactly −4 depending only on whether n is even or odd — verified directly against both sequences at every index from 0 through 15, with no deviation.

Lucas numbers and powers of φ, directly

The relationship goes deeper than shared identities; Lucas numbers are literally built into how φⁿ is expressed algebraically. Since φⁿ = (L(n) + F(n)·√5)/2, the Lucas and Fibonacci numbers at any index n are exactly the two integer ingredients needed to reconstruct φⁿ exactly, with no rounding. Checking it against real values: φ¹₀ ≈ 122.9918693812, and (L(10) + F(10)·√5)/2 = (123 + 55×2.2360679805)/2 works out to the identical 122.9918693812 — agreement to every decimal place double-precision arithmetic can represent. The conjugate root follows the mirror-image formula, ψⁿ = (L(n) − F(n)·√5)/2, which checks out the same way at every n tested. That's not a separate coincidence from Binet's formula; it's the same underlying algebra, just solved for φⁿ directly instead of for F(n).

A cleaner approximation than Fibonacci offers

There's a small, genuine sense in which Lucas numbers relate to φⁿ even more directly than Fibonacci numbers do. Because L(n) = φⁿ + ψⁿ and ψⁿ shrinks toward zero as n grows, L(n) itself is essentially just φⁿ rounded to the nearest whole number, with no division by √5 involved at all — checking it directly, φ¹₀ ≈ 122.991869, which rounds to 123, exactly L(10); φ²₀ ≈ 15126.999934, which rounds to 15127, exactly L(20). Fibonacci's version of the same shortcut needs the extra √5 in the denominator (F(n) ≈ round(φⁿ/√5)) because F(0)=0 and F(1)=1 pin down a different starting scale than L(0)=2 and L(1)=1 do. Both shortcuts rest on the identical vanishing-ψⁿ mechanism; Lucas numbers simply land on the version of it that doesn't need the extra scaling factor.

Lucas primes, and the harder open question next door

Among the Lucas numbers from L(0) through L(20), a genuinely large fraction are prime: 2, 3, 7, 11, 29, 47, 199, 521, 2207, 3571, and 9349 all check out as prime under direct testing — eleven primes out of twenty-one terms, a noticeably higher hit rate than a "random" sequence of similarly sized integers would produce. Fibonacci numbers have their own version of this question (which F(n) are prime), and it happens to be a considerably harder, still partly unresolved area of number theory — whether infinitely many prime Fibonacci numbers exist at all remains an open problem, despite the sequence being simple enough to state to a child. The Lucas sequence's own prime-density pattern is a genuine research topic in its own right, not fully explained by any simple rule, but grounded in a real, checkable observation rather than a folklore claim. Neither the count above nor the open Fibonacci-prime question is a matter of taste or interpretation — both are precise, well-defined mathematical statements, just ones number theory hasn't fully settled yet.

A real application: testing Mersenne numbers for primality

Beyond pure number-theoretic curiosity, the Lucas sequence underlies a genuinely important, still-used piece of applied mathematics: the Lucas–Lehmer primality test, the standard method for checking whether a Mersenne number (one less than a power of two, 2ᵀ − 1) is prime. The test builds a sequence derived from the same Lucas-style recurrence structure and checks a specific divisibility condition after p−2 steps — a method efficient enough that it's the actual technique behind essentially every record-setting largest-known-prime discovery for decades, all of them Mersenne primes found via this Lucas-derived test rather than by generic primality checking. It's a rare case of a sequence popularized through 19th-century recreational mathematics turning out to power genuinely cutting-edge 21st-century computational number theory.

Watching the primality test actually run

The Lucas–Lehmer test's mechanics are worth seeing worked through once rather than just described. To test whether M₁₉ = 2₁₉ − 1 = 127 is prime, start s at 4 and repeatedly square it, subtract 2, and reduce modulo 127: 4, then 14, then 67, then 42, then 111, then 0 — five steps (p−2 = 5 for p=7), landing on exactly 0. Reaching 0 at that specific step is the test's pass condition, and 127 is indeed a genuine, well-known Mersenne prime, confirming the test worked correctly on a case where the answer is already known. Run the identical procedure for p=5 (testing 31) and it reaches 0 after three steps; run it for p=13 (testing 8191, a considerably larger number) and it reaches 0 after eleven steps — both are Mersenne primes, and both get correctly confirmed by the same short, purely arithmetic procedure, with no factoring required at any point.

Part of a bigger family

The classical Fibonacci and Lucas sequences are, in fact, the single best-known example of a much broader mathematical structure that Lucas himself studied systematically: for any two integers P and Q, a pair of sequences Uⁿ(P,Q) and Vⁿ(P,Q) can be built from the recurrence Uⁿ = P·Uⁿ₋₁ − Q·Uⁿ₋₂ (and the same rule for V, with different starting values). Setting P=1 and Q=−1 turns that general recurrence into exactly F(n) = F(n−1) + F(n−2) for the U-sequence and L(n) = L(n−1) + L(n−2) for the V-sequence — meaning Fibonacci and Lucas numbers are simply one specific parameter choice inside a much larger family that Lucas developed the general theory for, decades before either sequence carried his name.

Converging to φ, the same way Fibonacci does

Because the Lucas recurrence is identical to Fibonacci's, just with different starting values, the ratio of consecutive Lucas numbers converges to φ by exactly the same mechanism described in depth elsewhere on this site: L(n+1)/L(n) approaches φ as n grows, with an error that shrinks at the same geometric rate (roughly a factor of φ² per step) governed by the same vanishing ψⁿ term from Binet's formula. L(17)/L(16) = 3571/2207 ≈ 1.6180342, already accurate to six decimal places by n=16 — a reminder that this convergence isn't a Fibonacci-specific fluke, it's a property of the recurrence itself, and the Lucas sequence demonstrates it just as cleanly starting from a completely different pair of numbers.

Try it directly

The Lucas number calculator computes any term and verifies the Fibonacci–Lucas identity for it directly, and the Fibonacci calculator computes the matching Fibonacci terms used throughout the identities above. The Fibonacci & Lucas convergence reference lays both sequences out side by side from n=0 through n=30, making every identity in this post directly checkable at a glance.