Browse posts by tag
The quantum-from-scratch series is now a ten-episode animated playlist: a NumPy simulator built from nothing, amplitude bars you can watch interfere, and a classical on-ramp with the coin under the cup.
The very first useful thing our simulator did, back in post 0, was turn a vector of amplitudes into a vector of probabilities.
Function approximation, why one linear unit cannot learn XOR, and what a hidden layer actually buys. The opening of a from-scratch tour of inductive bias.
Across the series I kept saying the simulator was small: a qubit is an array, a gate is a matrix, the whole thing is a few hundred lines.
Last post was the bad news: a real qubit leaks its coherence into the environment and forgets what it was doing.
Last post ended with a warning: the off-diagonal coherences of a density matrix are the fragile part, and a real qubit loses them on its own.
Every state in this series so far has been a single vector with definite amplitudes.
This post does not add a quantum idea.
This is the one everyone has heard of: the algorithm that factors integers in polynomial time and, if a big enough quantum computer is ever built, breaks RSA.
The last post built the Quantum Fourier Transform and promised it was a readout instrument.
The last three posts built circuits whose payoff was a single global fact read out by interference.
Deutsch-Jozsa and Bernstein-Vazirani solved artificial promise problems.
So far the qubits have mostly sat still.
In post 0 a qubit was a unit vector in $\mathbb{C}^2$, and everything about it fit in a length-two array.
I wanted to understand quantum computing properly, which for me means building the thing rather than driving a framework that does the linear algebra in the basement and hands back an answer.
Before the qubit, the ordinary object it edits: classical probability as vectors and stochastic matrices, and the one rule (weights never go negative) that quantum information breaks.