Quantum computing, briefly and sceptically

Bits and qubits

A classical bit is 0 or 1. A qubit can be in a superposition of both, and a register of n qubits can represent a combination of all 2ⁿ states at once.

The tempting conclusion — that it therefore does 2ⁿ computations simultaneously — is wrong, and the correction is the whole point. You cannot read 2ⁿ answers out. Measuring collapses the register to a single state. A quantum algorithm has to be constructed so that interference between the states makes the wanted answer overwhelmingly likely to be the one that survives measurement.

That construction is hard, and it has been achieved for a small number of problem types. It is not a general speed-up applied to whatever you were doing before.

Which problems

The problems where a genuine advantage is expected share a shape: a combinatorial search over an enormous discrete space, where classical methods have nothing better than trying possibilities.

Which makes the structural connection, if there is one, optimisation rather than analysis — Modules 15 to 17 rather than Module 9. A discrete design space of 8²³ combinations is exactly the shape of problem the class covers.

Why not analysis

A finite element solve is a linear system: assemble a matrix, apply boundary conditions, solve. Classical methods for this are extremely good — sparse direct solvers and iterative methods are close to optimal for the structure these matrices have — and there is no known quantum algorithm that changes that picture for the sizes structural engineers work at.

There are quantum linear-algebra algorithms with attractive asymptotic properties. They come with preconditions about how the matrix is supplied and how the answer is read out that are severe, and reading a full solution vector out is precisely the thing quantum measurement does not let you do cheaply. Since a structural engineer wants displacements at every node, that limitation is directly in the way.

A quantum computer would not make a finite element analysis faster. If it helps structural engineering at all, it will be somewhere in the optimisation, and that is a statement about problem classes rather than a prediction.

How to read a claim

This course's currency register marks quantum computing as requires verification, and splits the claim deliberately:

• Quantum computers are not in practical use for structural analysis — a statement about current practice that this work has not verified against a current authoritative source.
• The problems they are expected to help with are combinatorial rather than linear-algebraic — a statement about problem classes, which is stable.

The register keeps them apart because they have different lifetimes. And it carries one instruction: do not repeat any projection about timescales from any source, including this one. Predictions about when quantum computing will be practical have been made confidently and wrongly for three decades, and there is no reason to think the current batch is different.