Chapter 14: The Modern Period: Ramanujan and After¶
The Sanskrit mathematical tradition continued into the nineteenth century, but it was gradually displaced by the mathematics taught in the schools and universities established under colonial rule. Out of that new system came, in the early twentieth century, one of the most original mathematicians of any age, Srinivasa Ramanujan, and after independence a strong research community whose members have won the highest international honours. This chapter outlines that history, discusses the modern book of arithmetic shortcuts known as “Vedic Mathematics”, and closes with some reflections on how the Indian mathematical heritage might be studied and taught today.
14.1 Mathematics in Colonial India¶
14.1.1 The traditional scholars¶
Traditional astronomers and mathematicians continued to work in Sanskrit throughout the colonial period. In Kerala, Śaṅkara Varman’s Sadratnamālā (1819 or 1823; sources differ) restated the Kerala series (Chapter 6). In Odisha, Sāmanta Candraśekhara (1835–1904), working with traditional instruments and naked-eye observation, corrected the parameters of the Sanskrit siddhāntas in his Siddhāntadarpaṇa. Scholars such as Bāpūdeva Śāstrī (1821–1900) and Sudhākara Dvivedī (1855–1910) taught at the Sanskrit College in Banaras, edited the classical texts, and wrote on both Indian and European mathematics.
14.1.2 Village schools¶
Outside the Sanskrit tradition, arithmetic was taught in thousands of village schools, which declined in the course of the nineteenth century as the colonial system of education spread. What they taught, how they taught it and whom they admitted, as the colonial surveys of the 1820s and 1830s record, are described in Chapter 13 (Section 13.8).
14.1.3 The new education¶
From the early nineteenth century, colleges teaching European learning in English were founded in Calcutta, Bombay, Madras and elsewhere, and the universities of Calcutta, Bombay and Madras were established in 1857. A remarkable figure of the transition was Yesudas Ramchandra (“Master Ramchandra”, 1821–1880), a teacher at Delhi College, who wrote A Treatise on Problems of Maxima and Minima Solved by Algebra (1850), showing how such problems could be solved without the calculus. Augustus De Morgan arranged for its republication in London in 1859, with a preface praising its author. The Indian Mathematical Society was founded in 1907, and its journal published Ramanujan’s first papers. The Calcutta Mathematical Society followed in 1908, founded on the initiative of Ashutosh Mukherjee, himself a mathematician before he became a judge and vice-chancellor of Calcutta University.
14.1.4 Mathematics in the service of the state¶
Colonial administration needed large-scale calculation, and Indians did much of it. The Great Trigonometrical Survey of India, begun in 1802, measured the subcontinent by triangulation, a vast exercise in geometry and the reduction of observations. Its computers included Radhanath Sikdar (1813–1870), who joined the Survey in 1831, became its chief computer, and is credited with the calculations that in 1852 identified the peak then called “Peak XV”—later named Mount Everest—as the highest in the world. Revenue surveys, railway and irrigation engineering, and the decennial census created further demand for mathematically trained men, and the new colleges supplied them.
14.2 Srinivasa Ramanujan¶
14.2.1 Life¶
Srinivasa Ramanujan (Śrīnivāsa Rāmānujan) was born on 22 December 1887 at Erode in the Madras Presidency and grew up in Kumbakonam, in a Brahmin family of modest means. He showed exceptional mathematical ability at school. At about sixteen he obtained G. S. Carr’s Synopsis of Elementary Results in Pure and Applied Mathematics, a compendium of several thousand results stated with little or no proof, which shaped his style. Concentrating on mathematics to the exclusion of other subjects, he lost his college scholarship and never completed a degree. From 1912 he worked as a clerk in the Madras Port Trust, while recording his discoveries in notebooks (Kanigel 1991; Hardy 1940).
His talent was recognised by a few people in South India before it was recognised abroad. V. Ramaswami Aiyar, founder of the Indian Mathematical Society, and R. Ramachandra Rao, a district collector who supported him financially for a time, encouraged him; S. Narayana Iyer, his superior at the Port Trust, helped him with his work and correspondence. His first paper, on Bernoulli numbers, appeared in the Journal of the Indian Mathematical Society in 1911. In 1909 he had married Janaki, then a girl of about ten, in a marriage arranged by his mother, as was usual at the time; she lived until 1994.
In January 1913 Ramanujan wrote to G. H. Hardy at Cambridge, enclosing a list of theorems. Hardy, after study with J. E. Littlewood, recognised that the writer was a mathematician of the highest originality; some of the results were known, some were wrong, but others were new and extraordinary. Of certain formulas on continued fractions, Hardy later wrote that they “must be true because, if they were not true, no one would have the imagination to invent them”. Ramanujan came to Cambridge in 1914. He was awarded a degree by research in 1916, and in 1918 he became a Fellow of the Royal Society and a Fellow of Trinity College. His health, poor in the English climate and wartime diet, failed; he returned to India in 1919 and died on 26 April 1920, aged thirty-two (Hardy 1940; Kanigel 1991).
Ramanujan was devout, and accounts by his family and friends record that he attributed some of his insights to the family deity, the goddess Nāmagiri of Namakkal. A saying often quoted as his—that an equation has no meaning for him unless it expresses a thought of God—is reported only at second hand and cannot be verified.
14.2.2 The letters to Hardy¶
Ramanujan’s first two letters to Hardy contained many dozens of results. Most were identities for series, integrals and continued fractions; some were already known; a few were wrong, notably a claim about the distribution of prime numbers. Others were unlike anything Hardy had seen. One famous line of the second letter reads, in modern notation, \(1 + 2 + 3 + 4 + \cdots = -\tfrac{1}{12}\), with the warning that this held “under my theory”. The series does not converge, and Ramanujan was assigning to it a “constant” by a summation method of his own, which in this case agrees with the value \(\zeta(-1) = -\tfrac{1}{12}\) of the analytic continuation of the Riemann zeta function. Hardy saw that the writer, though untrained in modern rigour, had found his way independently to deep parts of analysis (Hardy 1940; Kanigel 1991).
14.2.3 Other work in number theory¶
In a long paper of 1915 Ramanujan studied “highly composite numbers”, numbers with more divisors than any smaller number (such as 12, 24, 36, 48, 60, 120, …), and determined their structure in great detail. With Hardy he proved in 1917 that the number of distinct prime factors of a typical integer \(n\) is about \(\log\log n\), a result that founded the probabilistic study of the arithmetic functions. He also discovered a large number of identities and congruences for arithmetic functions, many of which were proved only decades later.
14.2.4 Partitions¶
A partition of a positive integer \(n\) is a way of writing it as a sum of positive integers, order not counting; \(p(n)\) is the number of partitions. For example \(p(4) = 5\), since \(4 = 3+1 = 2+2 = 2+1+1 = 1+1+1+1\). The values grow very rapidly:
| \(n\) | \(p(n)\) | \(n\) | \(p(n)\) |
|---|---|---|---|
| 1 | 1 | 9 | 30 |
| 2 | 2 | 10 | 42 |
| 3 | 3 | 17 | 297 |
| 4 | 5 | 50 | 204,226 |
| 5 | 7 | 100 | 190,569,292 |
| 6 | 11 | 200 | 3,972,999,029,388 |
For \(n = 100\) the first term of the asymptotic formula below gives about \(1.99 \times 10^8\), within about five per cent of the true value; the additional terms found by Hardy and Ramanujan bring the computed value to the exact integer.
Mathematical Result: Hardy and Ramanujan, 1918
Hardy and Ramanujan in fact gave a series of terms of which this is the first, and from which \(p(n)\) can be computed exactly for large \(n\). The method they invented, studying a generating function near the points of the unit circle, became the “circle method”, one of the main tools of analytic number theory. In 1937 Hans Rademacher modified their series into a convergent series that gives \(p(n)\) exactly (Rademacher 1937):
where the \(A_k(n)\) are certain finite sums of roots of unity (Hardy and Ramanujan 1918).
Ramanujan also discovered congruences for the partition function:
For example \(p(4) = 5\), \(p(9) = 30\), \(p(5) = 7\), \(p(6) = 11\) and \(p(17) = 297 = 27 \times 11\). He proved the first two and gave the remarkable identity
which Hardy, agreeing with P. A. MacMahon, said he would choose if he had to select one formula from all of Ramanujan’s work. He conjectured further congruences for powers of 5, 7 and 11. In 2000 Ken Ono proved that for every prime \(\ell \ge 5\) there are infinitely many arithmetic progressions \(An + B\) with \(p(An+B) \equiv 0 \pmod{\ell}\) (Ono 2000).
14.2.5 The tau function¶
In a paper of 1916 Ramanujan studied the coefficients \(\tau(n)\) of
He conjectured that \(\tau\) is multiplicative (\(\tau(mn) = \tau(m)\tau(n)\) when \(m\) and \(n\) have no common factor), that it satisfies a recurrence at powers of primes, and that \(|\tau(p)| \le 2p^{11/2}\) for every prime \(p\). The first two conjectures were proved by L. J. Mordell in 1917. The third, the “Ramanujan conjecture”, was proved by Pierre Deligne in 1974 as a consequence of his proof of the Weil conjectures (Deligne 1974). Its generalisations are central to the modern theory of automorphic forms.
14.2.6 Series for 1/π¶
In 1914 Ramanujan published seventeen series for \(1/\pi\), of which the most famous is
Each term adds about eight correct decimal places; the first term alone gives \(\pi \approx 3.1415927\). The series rests on the theory of modular equations. Proofs were supplied by J. M. and P. B. Borwein in 1987, and in 1985 R. W. Gosper used the series to compute \(\pi\) to more than 17 million digits. Series of the same type are used in modern record computations, above all that of the brothers D. V. and G. V. Chudnovsky (1988),
each term of which adds about fourteen correct digits (Ramanujan 1914).
14.2.7 Continued fractions and q-series¶
Among the results in Ramanujan’s first letter to Hardy were evaluations of the continued fraction now named after L. J. Rogers and Ramanujan,
The closely related Rogers–Ramanujan identities, found by Rogers in 1894 and rediscovered by Ramanujan, express certain infinite sums as infinite products and have interpretations in the theory of partitions. They later appeared in statistical mechanics, in R. J. Baxter’s solution of the hard-hexagon model (1980), and in the representation theory of Lie algebras.
14.2.8 Mock theta functions¶
In his last letter to Hardy, in January 1920, Ramanujan described seventeen functions which he called “mock theta functions”, such as
which behave in some respects like the classical theta functions but are not of the same kind. For eighty years their true nature was unclear. In his doctoral thesis of 2002 Sander Zwegers showed that they are parts of “harmonic Maass forms”, and K. Bringmann and K. Ono used this framework in 2006 to prove long-standing conjectures about them. Mock modular forms now appear in number theory, in the “umbral moonshine” connecting them with finite groups (2012), and in calculations of black-hole entropy in string theory (Zwegers 2002; Bringmann and Ono 2006).
14.2.9 The notebooks¶
Ramanujan left three notebooks from his years in India, containing several thousand results, mostly without proof, and a sheaf of papers from his last year, the “lost notebook”, found by George Andrews in the library of Trinity College, Cambridge, in 1976. Bruce Berndt’s five-volume edition of the notebooks (1985–1998), and the edition of the lost notebook by Andrews and Berndt (from 2005), provide proofs or references for nearly all the results (Berndt 1985; Andrews and Berndt 2005).
14.2.10 Ramanujan and the Indian tradition¶
It is sometimes said that Ramanujan’s style—results stated without proof, in compressed form—reflects the Sanskrit sūtra tradition. There is no evidence that he studied Sanskrit mathematical texts; his mathematical education came from English books, above all Carr’s Synopsis, whose format of unproved results he followed. His mathematics belongs to the European tradition of analysis and number theory of Euler, Jacobi and their successors, to which he made extraordinary and original contributions. What connects him with earlier Indian mathematics is not a method or a doctrine but a shared delight in numbers and formulas.
14.2.11 Recognition¶
Ramanujan’s birthday, 22 December, was declared National Mathematics Day by the Government of India for the 125th anniversary of his birth (2012). Prizes in his name are awarded by the International Centre for Theoretical Physics in Trieste (from 2005, to young mathematicians from developing countries) and by SASTRA University (from 2005). Robert Kanigel’s biography The Man Who Knew Infinity (1991) brought his story to a wide public and was the basis of a film in 2015 (Kanigel 1991).
14.3 Mathematics in Independent India¶
14.3.1 Institutions¶
The modern research community grew around a few institutions: the Indian Institute of Science (Bangalore, 1909); the Indian Statistical Institute (Calcutta, 1931), founded by P. C. Mahalanobis; the Tata Institute of Fundamental Research (Bombay, 1945), whose School of Mathematics was built up by K. Chandrasekharan; the Institute of Mathematical Sciences (Madras, 1962); the Harish-Chandra Research Institute (Allahabad, founded in 1966 as the Mehta Research Institute); and the Chennai Mathematical Institute (1989). The National Board for Higher Mathematics (1983) supports research and the selection of students for the International Mathematical Olympiad, in which India has participated since 1989.
14.3.2 Statistics¶
India’s most distinctive contribution to twentieth-century mathematics was in statistics, and it grew from one institution. P. C. Mahalanobis (1893–1972), a physicist by training, founded the Indian Statistical Institute in 1931 and the journal Sankhyā in 1933. His “generalised distance” (1936) measures how far an observation \(\mathbf{x}\) lies from a population with mean \(\boldsymbol{\mu}\) and covariance matrix \(\mathbf{S}\), taking account of the correlations between the variables:
and it is now used throughout multivariate statistics and pattern recognition (Mahalanobis 1936). Mahalanobis also designed the large-scale sample surveys that became the National Sample Survey (1950), and his statistical models shaped the Second Five-Year Plan. Around him the Institute became one of the world’s leading centres of statistical research. C. R. Rao (1920–2023), in a paper of 1945, proved the lower bound for the variance of an estimator now called the Cramér–Rao inequality and the method of improving estimators now called the Rao–Blackwell theorem (Rao 1945). R. C. Bose (1901–1987), working on the design of experiments, was led to the combinatorics of orthogonal Latin squares; with S. S. Shrikhande and E. T. Parker he showed in 1959 that Euler’s conjecture of 1782 on such squares is false (Bose et al. 1960), and the Bose–Chaudhuri–Hocquenghem error-correcting codes bear his name.
14.3.3 Pure mathematics¶
In number theory the line from Ramanujan was continued by S. S. Pillai (1901–1950), who made important contributions to Waring’s problem. One of the most influential Indian-born mathematicians of the mid-century, however, worked far from number theory: Harish-Chandra (1923–1983), who began as a physicist under Homi Bhabha and Paul Dirac, created much of the theory of infinite-dimensional representations of semisimple Lie groups and the harmonic analysis on them that underlies the modern Langlands programme.
At the Tata Institute of Fundamental Research, the School of Mathematics built up by K. Chandrasekharan produced work of international importance in algebraic geometry and the theory of Lie groups. M. S. Narasimhan (1932–2021) and C. S. Seshadri (1932–2020) proved in 1965 the theorem relating stable vector bundles on a compact Riemann surface to unitary representations of its fundamental group (Narasimhan and Seshadri 1965); M. S. Raghunathan (b. 1941) worked on discrete subgroups of Lie groups; C. P. Ramanujam (1938–1974) gave a topological characterisation of the affine plane; and S. S. Abhyankar (1930–2012), working mostly in the United States, studied algebraic geometry in positive characteristic, including the resolution of singularities. In probability, S. R. S. Varadhan (b. 1940), trained at the Indian Statistical Institute, received the Abel Prize in 2007 for the theory of large deviations.
Algebra and number theory have remained strong. Raman Parimala (b. 1948) and Sujatha Ramdorai (b. 1962), among many others, have made major contributions to the arithmetic of quadratic forms and to Iwasawa theory, and Neena Gupta (b. 1984) showed in 2014 that the Zariski cancellation problem has a negative answer in positive characteristic (Gupta 2014). Manjul Bhargava (b. 1974) and Akshay Venkatesh (b. 1981), both of Indian origin and working abroad, received the Fields Medal in 2014 and 2018. Bhargava’s “higher composition laws” (from 2004) found new composition laws generalising Gauss’s composition of binary quadratic forms, whose simplest case is the identity behind Brahmagupta’s bhāvanā (Chapter 4). The mathematical connection is real; it is a case of a modern mathematician generalising a classical identity, not of an ancient author possessing the modern theory (Bhargava 2004). In the later twentieth century, mathematics in India also broadened to include theoretical computer science, combinatorics and mathematical physics, with strong groups at the Institute of Mathematical Sciences, the Chennai Mathematical Institute and elsewhere.
14.4 “Vedic Mathematics”¶
A book called Vedic Mathematics is widely sold and taught in India and abroad, and is often presented as ancient Indian knowledge. Its history should be clearly stated.
The book was written by Bhāratī Kṛṣṇa Tīrtha (1884–1960), who was Śaṅkarācārya of the Govardhana Maṭha at Puri, and was published posthumously in 1965, edited by V. S. Agrawala (Bhāratī Kṛṣṇa Tīrtha 1965). It presents sixteen short sūtras and thirteen sub-sūtras, such as ekādhikena pūrveṇa (“by one more than the previous one”), nikhilaṃ navataś caramaṃ daśataḥ (“all from nine and the last from ten”) and ūrdhva-tiryagbhyām (“vertically and crosswise”), and uses them as headings for methods of rapid mental calculation. The author said that the sūtras came from an appendix (pariśiṣṭa) of the Atharvaveda. No such passage exists in any known recension of the Atharvaveda; the book’s own editor noted that the sūtras are not found in its published appendices, and the author’s account presents them as reconstructed by him after years of study and contemplation. The sūtras are not in Vedic Sanskrit, and the methods use the decimal place-value system, which, as Chapter 7 has shown, was developed more than a thousand years after the Vedic period (Dani 1993).
An example: the nikhilam method
To multiply \(97 \times 96\), note the deficits from 100: \(3\) and \(4\). Subtract crosswise: \(97 - 4 = 96 - 3 = 93\). Multiply the deficits: \(3 \times 4 = 12\). The answer is \(9312\).
The method rests on the algebraic identity
here with \(a = 3\), \(b = 4\). It is a useful trick for numbers close to a power of ten, of a kind found in many arithmetic textbooks; it is not Vedic.
The methods of the book can make mental arithmetic quicker and can interest pupils in numbers, and some teachers use them for that purpose. But presenting them as the mathematics of the Vedas misrepresents both the Vedas and the history of Indian mathematics, and it obscures the genuine achievements described in this volume. Mathematicians and historians, among them S. G. Dani, have made these points repeatedly (Dani 1993).
14.5 Studying and Teaching the Heritage¶
India’s National Education Policy of 2020 calls for the inclusion of the Indian knowledge systems in education, including the history of Indian mathematics. This creates an opportunity, and a responsibility. The history can enrich the teaching of mathematics: Āryabhaṭa’s sine table, Brahmagupta’s rules for zero and negative numbers, the kuṭṭaka and the cakravāla, Piṅgala’s enumeration of metres and the Kerala series are mathematically substantial and can be taught as mathematics. But the history must be accurate. The experience of the past century suggests some simple principles:
-
Use the primary texts in reliable editions and translations, and the work of specialist historians, rather than popular compilations.
-
Check the mathematics: rules, examples and numerical values should be recomputed, as has been done throughout this volume.
-
Distinguish what a text says from modern interpretations of it, and avoid the language of “anticipation” and “priority”.
-
Present disputed questions—the date of the Bakhshālī manuscript, the transmission of the Kerala series, the origins of the place-value system—as disputed.
-
Recognise the exchanges with other cultures, and the social limits of the tradition, as part of its history.
A large body of reliable scholarship now exists to support such teaching, from the critical editions of K. S. Shukla, K. V. Sarma and their successors to the surveys of Kim Plofker and P. P. Divakaran and the translations by Agathe Keller, Takao Hayashi and K. Ramasubramanian and his colleagues (Plofker 2009; Divakaran 2018; Keller 2006; Hayashi 1995; Sarma et al. 2008).
14.6 Conclusion¶
The history traced in this volume extends over more than three thousand years: from the decimal number words of the Vedas and the altar geometry of the Śulbasūtras, through the place-value numerals and zero, the astronomical mathematics of Āryabhaṭa, Brahmagupta and the two Bhāskaras, the combinatorics of the prosodists and of Nārāyaṇa, and the infinite series of the Kerala school, to the exchanges with the Islamic world and Europe and to Ramanujan and the mathematicians of modern India. It is a history of genuine and lasting achievement. It is also a history of particular people, working in particular social settings, who learned from others as well as teaching them, and who made mistakes as well as discoveries. Told in that way, with its evidence and its uncertainties, it is more interesting—and more useful—than any story of unbroken superiority.
Exercises¶
-
List the 15 partitions of 7.
-
Verify that \(1729 = 1^3 + 12^3 = 9^3 + 10^3\).
-
Check Ramanujan’s congruences for \(p(9) = 30\), \(p(12) = 77\) and \(p(17) = 297\).
-
Use the first term of the Hardy–Ramanujan formula to estimate \(p(50)\), and compare with the true value \(204{,}226\).
-
Compute \(98 \times 97\) by the nikhilam method and explain why it works.
-
Compute the value of \(\pi\) given by the first term of Ramanujan’s series for \(1/\pi\).
Answers. (1) \(7\); \(6{+}1\); \(5{+}2\); \(5{+}1{+}1\); \(4{+}3\); \(4{+}2{+}1\); \(4{+}1{+}1{+}1\); \(3{+}3{+}1\); \(3{+}2{+}2\); \(3{+}2{+}1{+}1\); \(3{+}1{+}1{+}1{+}1\); \(2{+}2{+}2{+}1\); \(2{+}2{+}1{+}1{+}1\); \(2{+}1{+}1{+}1{+}1{+}1\); \(1{+}1{+}1{+}1{+}1{+}1{+}1\). (2) \(1 + 1728 = 1729\) and \(729 + 1000 = 1729\); Ramanujan remarked to Hardy that it is the smallest number expressible as a sum of two cubes in two different ways. (3) \(30 = 5 \times 6\); \(77 = 7 \times 11\); \(297 = 11 \times 27\). (4) About \(217{,}590\), some 6.5% too large. (5) Deficits 2 and 3; \(98 - 3 = 95\); \(2 \times 3 = 06\); answer 9506, from \((100-a)(100-b) = 100(100 - a - b) + ab\). (6) \(9801/(2\sqrt2 \times 1103) \approx 3.1415927\).
Scholarly Note: Further reading
Kanigel (1991) is the standard biography of Ramanujan, and Hardy (1940) a mathematician’s assessment by his collaborator; Berndt’s edition of the notebooks (1985–1998) and Andrews and Berndt’s edition of the lost notebook (from 2005) give the mathematics; the partition formula is in Hardy and Ramanujan (1918) and Rademacher (1937). On “Vedic Mathematics”, see Dani (1993) (Kanigel 1991; Hardy 1940; Berndt 1985; Andrews and Berndt 2005; Hardy and Ramanujan 1918; Rademacher 1937; Dani 1993).