Why a tiny change in a math problem can ruin your computer's answer

Why a tiny change in a math problem can ruin your computer's answer
Image: Intellect (generated)

In short: James H. Wilkinson, a pioneer of numerical analysis, revealed how digital computers struggle with instability. When a problem is ill-conditioned, even a microscopic adjustment to the data can lead to wildly incorrect results. Understanding these errors is the difference between a reliable simulation and a total computational failure.

James H. Wilkinson was an English mathematician and a foundational figure in numerical analysis, the field that bridges mathematics and computer science to solve complex real-world problems. Awarded the A.M. Turing Award in 1970, Wilkinson was specifically recognized for his groundbreaking work in numerical linear algebra and his development of 'backward' error analysis. His career, which included working on the ACE computer project with Alan Turing at the National Physical Laboratory, focused on ensuring that digital computers could produce accurate results despite the inherent limitations of finite precision arithmetic.

Numerical analysts often face problems that cannot be solved directly, requiring them to replace a complex task with a 'nearby problem' that is easier to compute. A major challenge in this process is stability, which measures how sensitive a solution is to small changes in input data. Wilkinson famously illustrated this through the study of unstable polynomials. In one example, a polynomial with roots at 1 through 7 becomes highly sensitive when a single coefficient is altered by a tiny fraction. Such a shift can transform simple integer roots into complex numbers, demonstrating why the formulation of a problem must be as robust as the algorithm used to solve it.

This sensitivity is critical in modern scientific computing, where sophisticated models—such as those for atmospheric dynamics or mechanical systems—rely on solving large systems of equations. Because these systems often involve millions of rounding errors, numerical analysts must carefully monitor the size and analytic form of errors to maintain accuracy. By understanding these hidden mechanisms, researchers can create reliable software that powers everything from engineering simulators to climate models, ensuring that the underlying mathematics remains transparent and effective for the user.

Opens the Intellect+ feed with this card first. Swipe up for the next one.

Sources

To cite this card, link to https://intellectplus.co.uk/app/science-james-h-wilkinson.

How we source, check and correct cards: Editorial standards.

General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.

Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?

Related

Finding something worth knowing…