Quantum Computing · Part B — 08

Shor, Fully Assembled

Now we bolt the toolkit together. Shor’s algorithm factors a big number — the bedrock of internet encryption — by a surprising detour: it turns factoring into finding a repeating period, and finds that period with phase estimation. Almost all of it is ordinary arithmetic; only one step is quantum.

↩ before you start · keep these handy
·From QC 07: phase estimation reads the eigen-phase of a unitary as a binary fraction.
·From QC 06: the inverse QFT inside QPE turns a winding into a readable number.
·From QI 8d: you’ve seen the outline — here is the full machine, end to end.
🔑 symbol decoder · every new mark, in plain words
aᵪ mod Nthe remainder of aᵪ divided by N. As x grows this sequence eventually repeats. period r (order)the smallest r with a^r mod N = 1 — the length of that repeat. gcd(p, q)greatest common divisor — the largest number dividing both. Fast and classical (Euclid). coprimetwo numbers sharing no factor: gcd = 1. Uₐ (modular mult.)the gate |y⟩ → |a·y mod N⟩. Its eigen-phases are exactly the fractions s/r. continued fractionthe classical trick that recovers r from the measured fraction s/r.
feel

Factoring in disguise

Nobody knows how to factor a huge number quickly — but there’s a back door. Pick a random number a and look at the powers a, a², a³… modulo N. They cycle with some period r. A little algebra turns that period into a factor of N. Finding the period is the only hard part — and that’s exactly what a quantum computer does exponentially faster.

🎲 everyday picture

A locked safe is hard to open by guessing the combination. But suppose pressing a key makes the dial advance by a fixed secret step, and you notice it returns to start every r presses. That repeat length quietly reveals the lock’s structure. Shor doesn’t guess the factors — it measures the rhythm of a related cycle, and the rhythm gives the factors away.

recapFactoring reduces to finding the period of aᵪ mod N; the period hands you a factor.
play

Factor a number, step by step

Choose N and a base a. The powers of a (mod N) are listed until they loop — that loop length is the period r the quantum step would find. The classical wrapper then turns r into the factors. Some choices of a fail (odd r, or a trivial root); that’s expected — you just retry.

▸ period-to-factor benchperiod r = quantum · rest = classical
N =
base a = {{ a }}
powers of {{ a }} (mod {{ N }}) — until the cycle closes
{{ s.v }}
a{{ s.exp }}
{{ ln.icon }}{{ ln.text }}
{{ verdict }}
recapFind r, then factor = gcd(ar/2 ± 1, N) — the quantum step supplies r, Euclid does the rest.
math

The five steps, and the one that’s quantum

1.Pick a random a < N. If gcd(a, N) > 1 you’ve already found a factor — lucky, done.
2.[QUANTUM] Run phase estimation on Uₐ|y⟩ = |a·y mod N⟩. Its eigen-phases are s/r, so QPE returns a fraction ≈ s/r.
3.Use continued fractions on that measured fraction to recover the period r.
4.If r is odd or ar/2 ≡ −1, discard and retry from step 1 (succeeds >½ the time).
5.Otherwise the factors are gcd(ar/2 − 1, N) and gcd(ar/2 + 1, N).

Only step 2 needs a quantum computer, and only because no fast classical period-finder is known. Everything else — gcd, continued fractions, retries — runs on your laptop in microseconds.

✎ worked example · factor N = 15 with a = 7
1.powers: 7¹=7, 7²=49≡4, 7³≡13, 7⁴≡1 (mod 15) → period r = 4
2.r is even, and ar/2 = 7² = 49 ≡ 4 (not −1) — good
3.gcd(4 − 1, 15) = gcd(3, 15) = 3;   gcd(4 + 1, 15) = gcd(5, 15) = 5
4.15 = 3 × 5. ✓
recapFive steps; only the period-finding (step 2) is quantum, via QPE on the modular-multiply gate.
⚠ common misconceptions

“The quantum computer tries all possible factors at once.” It never touches factors directly. It finds a period; a short classical calculation converts that period into factors.

“Shor always works on the first try.” It’s randomized. A bad a (odd period, or ar/2 ≡ −1) just means retry — the success chance per attempt is comfortably above ½, so a few tries suffice.

“So today’s computers can break RSA.” The algorithm is proven, but factoring a real 2048-bit key needs millions of high-quality, error-corrected qubits — far beyond current hardware. That gap is the whole story of Parts C and D.

✓ you can now
explain why factoring reduces to period-finding
run the five steps of Shor by hand on a small N
identify the single quantum step (QPE on the modular-multiply gate)
say why it threatens RSA in principle but not yet in practice
← 07 Phase Estimation next · 09 Hamiltonian Simulation