Powell's Pi Paradox & Madhava of Sangamagrama — First Principles Study

Source: Mathologer — "Powell's Pi Paradox"

1. The Naive Question

If you add up the series 1 - 1/3 + 1/5 - 1/7 + 1/9 - ... forever, you get π/4. This is a beautiful result. But try computing it: after a million terms, you've only gotten six correct digits of π.

Then something strange happens. Deep inside the decimal expansion — past where the series has gone wrong — the digits start matching π again. How can a series that's hopelessly inaccurate at the surface still carry the correct digits hiding further down?

This is Powell's Paradox. And the solution, it turns out, was already known in 14th-century India.

2. Strip the Assumptions

Assumption 1: "Leibniz and Gregory discovered the π/4 series."
The series π/4 = 1 - 1/3 + 1/5 - ... is called the Leibniz formula or the Gregory-Leibniz series in Western textbooks. Both men published it in the late 17th century. Madhava of Sangamagrama derived it at least 200 years earlier, around 1400 AD, along with the power series for sine, cosine, and arctangent. Kerala's mathematicians were doing calculus before Europe.

Assumption 2: "A slowly converging series is just slow — no hidden structure."
Powell's paradox shows this is wrong. The error in a partial sum isn't random noise. It has structure — specifically, it follows the pattern of Madhava's correction terms. The "paradoxical" matching digits are the correction terms making themselves visible.

Assumption 3: "Corrections to a series are a modern idea."
Madhava derived a sequence of correction terms in the 14th century that transformed his hopelessly slow π/4 series into one with 15-digit accuracy — by adding a single carefully chosen fraction.

3. Build Back Up

The Series and Its Problem

π/4 = 1 - 1/3 + 1/5 - 1/7 + ... (alternating series of odd denominators)

To get 6 correct digits: ~1,000,000 terms needed. Useless for practical computation.

Madhava's Correction Terms

Madhava noticed that adding/subtracting half the last term included brought the partial sum much closer. He then systematically refined:

These weren't guesses — they were derived by comparing partial sums to the known high-precision approximation 355/113 and spotting the pattern in the residuals.

Why the Paradox Happens

When summing N = 1,000,000 terms, the difference between the partial sum and π/4 is:

error ≈ 1/(4N) = 0.00000025...

This creates predictable strings of zeros (or nines) in the decimal expansion at position 7+ — exactly where Powell saw the correct π digits mysteriously reappearing. The "paradox" is just Madhava's correction term becoming visible in the digit expansion.

The Cultural Suppression

Madhava's results were recorded on palm-leaf manuscripts in verse in the Malayalam language. His original texts are lost; they survive only through commentaries by his students in the Kerala School. Western mathematics ignored this tradition until the 20th century. The history of calculus as "Newton and Leibniz" is a narrative that erases two centuries of prior Indian work.

4. The Insight

The mathematical universe doesn't care who names a discovery. Madhava's correction terms existed in 14th-century Kerala, waiting to explain a 1983 computer paradox. Mathematical truth transcends geography and era — but history is written by those with printing presses.

The deeper mathematical insight: the error of a partial sum of a well-behaved alternating series is not random. It's structured, and that structure is exactly what Madhava's correction terms capture. Understanding the anatomy of error is how you turn a bad approximation into a great one.