Brahmagupta

Brahmagupta

Brahmagupta was a seventh-century mathematician and astronomer. His Brāhmasphuṭasiddhānta is dated to 628 CE and associated with Bhillamāla, now Bhinmal. His Khaṇḍakhādyaka is dated to 665. About 598 is the usual birth estimate; a precise death date and a detailed account of his education are less secure. A place of composition should not automatically become a documented birthplace. MacTutor's account is useful while also showing why the source of each personal claim matters.

The mathematical importance of his work does not depend on saying that he invented every use of zero. His treatment of signed quantities and zero belongs to a larger history of calculation. Binary zero, an additive identity, a Boolean value, and a missing-data marker have different meanings, even when a programming system represents them with similar bits.

Read the rules in their actual setting

The Brāhmasphuṭasiddhānta is a work of mathematical astronomy. Colebrooke's 1817 translation places Brahmagupta's arithmetic in chapter XII and algebra in chapter XVIII; the rules for signed quantities and zero belong to the latter chapter's algorithm section. Translation contents and Cambridge's chapter XVIII record identify the edition and chapter.

Brahmagupta describes positive and negative quantities using assets and debts, and states operations involving zero. Some division rules do not agree with modern field arithmetic, including the assignment of zero to zero divided by zero. Recognizing that limitation is part of understanding the contribution, rather than an excuse to replace history with a story of a finished number system. MacTutor reproduces and discusses the rules.

Worked mechanism: derive a sign rule instead of memorizing it

This is a modern algebraic argument. Assume addition has an identity 0, each a has an additive inverse -a, and multiplication distributes over addition.

First, 0*b = (0+0)*b = 0*b + 0*b. Adding the inverse of 0*b to both sides gives 0*b = 0.

Next, a*b + (-a)*b = (a+(-a))*b = 0. Therefore (-a)*b is the additive inverse of a*b, so (-a)*b = -(a*b). Applying that fact twice yields (-a)*(-b) = a*b.

The point is compatibility: if we want additive inverses and distributivity to continue working, the familiar sign rules follow. We are not claiming that Brahmagupta wrote this axiomatic proof or used this notation.

Take a balance initially equal to -7. Adding an asset of 12 produces 5. Removing a debt of 4 means subtracting -4, giving 9. In contrast, subtracting one negative number from another is not always negative: -3 - (-8) = 5. The result depends on the magnitudes as well as the signs.

Division has a different obligation

In rational or real arithmetic, division a/b seeks the unique x satisfying b*x = a. For b = 0 and a != 0, there is no solution because 0*x = 0. For a = b = 0, every x is a solution, so there is no unique quotient. Ordinary division is consequently undefined in both cases.

An indeterminate form in a limit is another issue. The expressions t/t and t²/t both have numerator and denominator tending to zero as t tends to zero, but their limits are 1 and 0. That observation does not define a numerical value for 0/0.

A software interface can return an error, an optional result, or a specified special value when division is invalid. It must specify the behavior. Choosing a representation for failure does not change the underlying field operation.

Formula conditions matter as much as formulas

For a cyclic quadrilateral with sides a,b,c,d and semiperimeter s, the area formula associated with Brahmagupta is sqrt((s-a)(s-b)(s-c)(s-d)). A cyclic quadrilateral has all four vertices on one circle. The condition cannot be dropped: equal side lengths can form rhombi with different angles and different areas. MacTutor notes scholarly disagreement about whether that restriction was implicit in the historical text.

For an original example, sides 4,5,6,7 give s = 11 and area sqrt(840). That value is warranted only with the geometric condition. Naming a formula is not a substitute for checking its domain.

Finite computer arithmetic introduces still another domain. In a four-bit unsigned model, arithmetic is modulo 16; 9+8 produces residue 1. A signed two's-complement interpretation assigns the bit pattern 1001 the value -7, while the same pattern is unsigned 9. These are modern encoding choices, not properties of zero itself or historical mechanisms attributed to Brahmagupta.

Exercises

1. Prove a*0 = 0 using distributivity and additive inverses. Explain why the proof does not permit cancellation of a zero factor from 0*x = 0*y.
2. Compare the limit of 3t/t with that of t²/t as t approaches zero through nonzero real values. Explain why neither calculation makes ordinary 0/0 a number.
3. A square and a rhombus both have four sides of length 2. The rhombus has an angle of 30 degrees. Compute both areas and show why side lengths alone do not justify the cyclic-quadrilateral formula.

Compare your reasoning with the worked solutions

1. Multiplication by zero

a*0 = a*(0+0) = a*0 + a*0. Add the additive inverse of a*0 to obtain a*0 = 0. Cancellation of a common nonzero factor in a field uses its multiplicative inverse. Zero has no such inverse: if 0*z = 1, the just-proved rule would imply 0 = 1. Thus 0*x = 0*y provides no information about whether x = y.

2. Limits are not quotients at the excluded point

For every nonzero t, 3t/t = 3, so its limit is 3. Also t²/t = t, whose limit is 0. The two limiting processes yield different answers despite both being described by the form 0/0. At t = 0 neither quotient is defined by ordinary division. A limit describes nearby values; it does not silently fill a missing domain point.

3. A counterexample exposes the missing condition

The square has area 4. The rhombus has base 2 and height 2*sin(30°) = 1, so its area is 2. Inserting the four side lengths into the cyclic formula gives sqrt(2*2*2*2) = 4 for both, which is wrong for the rhombus. The non-square rhombus is not cyclic: its opposite angles sum to either 60 or 300 degrees, rather than 180. The formula's hypothesis distinguishes the cases.

Sources and reading limits

• Brahmagupta biography, University of St Andrews: work dates, signed rules, zero, and the cyclic condition. Broad priority claims are not repeated here.
• Algebra, with Arithmetic and Mensuration, H. T. Colebrooke, 1817: verified title and contents of a historical translation containing works by both Brahmagupta and Bhāskara. This landing page is a contents transcription, not the complete translated chapters.
• Chapter XVIII: Algebra, Cambridge University Press reprint record: verifies chapter identity and attribution; the linked catalog record does not supply the full chapter text. The proofs, encoding example, and exercises above are original modern pedagogy.