Afterword

The course is over. That is a strange thing to write, after a hundred and fifty notebooks, because a course does not really end — it stops, and what it built goes on working inside whoever built it alongside us.

We began with a small embarrassment: a computer cannot represent \((1+\varepsilon)-1\). That was never a warning to be careful. It was the first instance of the only lesson the course had to teach — that every number we compute is approximate, and that the whole of the craft is knowing, and being able to say, exactly how approximate. Everything after was elaboration. The oscillator we followed through eight volumes; the action that turned out to generate all of it; the exponent that two lattices and a quantum chain agreed on without conferring — these were never results to be collected. They were occasions, each one, to practice a single discipline: compute the thing more than one way and demand the answers meet; separate an error into its sources and name the largest; check the result against a fact you already trust; and when you are wrong, be wrong out loud, on purpose, where the wrongness can still teach you something.

We did not solve everything, and we said so each time. There are places — the fluctuation determinant, the renormalization group’s real machinery, the sum over diagrams — where we walked to the edge, named what lay beyond it, and stopped. That was not modesty. A course that pretends to have no edges is lying about the one thing it most owes you, which is where its knowledge runs out. The horizons are real. They are yours now, if you want them.

What you carry out of here is not the \(\coth\), or the instanton, or \(\beta = 1/8\); those you can always look up. What you carry is the instrument the course was quietly building the whole time — a way of computing a quantity no one has handed you the answer to, and knowing, within a stated tolerance, how far to trust what comes back. That is the difference between calculation and knowledge, and it does not care what you point it at. Materials, markets, the weather, some question no one has thought to ask yet: the discipline is the same.

Thank you for the company. It was a genuine pleasure to compute these things beside you. The error never went away — we simply learned, together, exactly how large it was, which from the first notebook to the last was the whole of the craft.

— R. A.