At lunch time, he would emerge briefly from the classroom to eat his meal of rice with yogurt and feed the bullock hay. Occasionally, he found that the bullock would have come unmoored, and run all the way back home! An early childhood photo shows him with the traditional shaved head and topknot of the South Indian brahmin, and wearing diamond studs in his ears. However, after his father passed away when he was twelve, years of financial hardship followed. He found his solace in books, reading everything that he could lay his hands on. A favourite family story is about the time he was sent to the town with money to buy some sugar, and discovered to his delight that a circulating library had just opened there. He spent all the money on a library membership instead, returning home much later with a bag of books (having forgotten all about the sugar). He also found joy in solving 'riders' or trigonometric puzzles, and declared that he would devote his life to mathematical research. His proud family responded by painting the walls of a room in their house black, so that he could scribble equations on the walls using chalk.

This student looks clever, but he should know that this is neither the kind of mathematics we do nowadays, nor the kind he should be doing.

Father Racine was a mentor not only for me, but also for almost all of that whole generation of mathematicians. ... the amazing thing was that he was in touch with many of the modern developments in mathematics. He was also a close friend of many top French mathematicians of the day like Jean Leray. ... Racine introduced algebra - the so-called modern algebra of that time - and it was taught very well. ... He taught only the honours class, and never used to teach in the first year. However, he still wanted to get in touch with his students before teaching them.

... my father did not have enough money to pay the train fare from Madras to Mumbai to attend the Tata Institute of Fundamental Research (TIFR) admissions interview; a teacher found out and lent him the money. He and his classmate C S Seshadri travelled to Bombay, and found free accommodation, for a few days, in a temple. Many people have remarked how dapper my father looked in his elegant suits and ties, however the first time he wore European trousers (as opposed to the Tamil veshti) was for his TIFR interview. He felt intimidated during the interview, especially by K G Ramanathan whose dark glasses made him appear particularly inscrutable. My father and Seshadri were both admitted to TIFR, their 'hostel' was the former servant quarters of the Old Yacht Club. My father was to wonder later whether the dismal living conditions there were responsible for him subsequently catching tuberculosis.

In the early fifties, Komaravolu Chandrasekharan initiated a programme for promoting mathematical research at the highest level at the Tata Institute of Fundamental Research (TIFR) in Bombay. At his invitation, Laurent Schwartz visited TIFR in 1955 for the first time and this was a landmark in the history of mathematical research in India. He gave a course on complex analytic manifolds. Starting from the definition of a differentiable manifold, the course covered - among other topics - currents, de Rham's theorem, elliptic partial differential equations, Hodge theory, Kähler manifolds and the Riemann-Roch theorem for compact Riemann surfaces. His inspiring lectures not only introduced young mathematicians at TIFR to important topics and techniques which were hardly studied in India at that time, but also introduced them to a whole new way of thinking about, and of doing, mathematics. l was a research student at TIFR at that time. I wrote the notes of his lectures and this gave me an opportunity to get to know him. Conjeeveram S Seshadri also attended and our common interest in the field of algebraic geometry can be traced to the impetus received from these lectures.

I myself worked in Paris with Schwartz, and Seshadri with Claude Chevalley. We used our stay in Paris to familiarise ourselves with the tremendous development that was then taking place in algebraic geometry, and naturally this played an important role in our future research. ... I am personally grateful to Laurent Schwartz for my mathematical development and the inspiration he provided. ... During my stay in Paris I had some serious health problems and Schwartz made special efforts to see that l received the best medical attention. During my recuperation, he gave me, as "light reading", the huge manuscript of Kodaira and Spencer on deformations of complex structures - in retrospect, this turned out to be a very perceptive choice of gift. The work of Kodaira and Spencer played a crucial role in my future research.

Schwartz used to come once a week to the university, give two lectures and then see his research students. The students used to sit on a bench outside, awaiting their turn to go and see him. Then one day a Japanese guy was sitting next to me, and we started talking. He asked me, "Do you know a mathematician called Narasimhan?" It turned out that he was referring to me. Then we talked and that's how we started collaborating. It's a kind of an accident that I met him. ... Probably my first best work was with him.

Upon his return to TIFR in 1960, Narasimhan embarked on an intense collaboration with Seshadri that resulted in the famous Narasimhan-Seshadri theorem, published in 1965. This theorem captures the interconnection between various branches of geometry, topology and theoretical physics, and was the basis for later fundamental works by some of the greatest mathematicians of our time such as Michael Atiyah, Raoul Bott, Simon Donaldson, Karen Uhlenbeck, Shing-Tung Yau, Nigel Hitchin and Carlos Simpson, among others.

D Mumford has defined the notion of a stable vector bundle on a compact Riemann surface X and proved that the set of equivalence classes of stable bundles (of fixed rank and degree) has a natural structure of a non-singular, quasi-projective, algebraic variety. We prove in this paper that, if X has genus ≥ 2, the stable vector bundles are precisely the holomorphic vector bundles on X which arise from certain irreducible unitary representations of suitably defined Fuchsian groups acting on the unit disc and having X as quotient.

When he returned to India in 1960, I was a graduate student looking for a problem to work on. Our first work was on universal connections, which proves that the bundle universal for principal unitary bundles has a connection that is universal for unitary bundles with connections. As soon as I told him that I felt this could be true, he got excited and began asking me questions, such as "did you check it for U(1)" and "did you check it for trivial bundles?" This helped me understand how one should go about solving problems: try and solve simpler cases and look for analogies. We solved this question within a couple of weeks, and it appeared in a prestigious journal.

Narasimhan was a great teacher as any of his students would attest. He was good, not spectacular, in the classroom, but he had few peers in informal one-on-one communication with students or fellow mathematicians. He had the gift of getting to the heart of the most abstruse problems and then conveying it to the student, side-stepping technicalities that could cloud the issue. I benefitted from this ability of his early on (towards the end of a seminar on Differential Geometry he conducted with Sundararaman Ramanan): He explained to me all of the Kodaira-Spencer deformation theory (incidentally, one of his favourites, and a recently developed theory at that time) in about 3 hours during a few walks on the seashore at TIFR.

My father read widely and voraciously, in English, Tamil, French and occasionally Italian. He also enjoyed reading the Times Literary Supplement, the New York Review of Books, and the weekend arts and literary supplements in The Hindu, Il Sole 24 and the Financial Times. We shared a love for detective fiction, and would wait eagerly for the latest instalment in mystery series by our favourite authors. ... In addition to mathematics and books, he enjoyed good food and wine, cricket and the company of friends. Well into his eighties, he was open to new experiences and new adventures. ... My father was also interested in art, with his favourites being the French Impressionist painters.

I acquired from him a feeling of empathy for the disadvantaged and disenfranchised in our society, a non-jingoistic pride in being Indian, and a respect for rationalism and the power of science.

First of all get a broad based knowledge when you are a student, by reading good textbooks and seminar notes, classics and masters and by associating with good mathematicians. One often learns faster many branches of mathematics by discussions with teachers and fellow students. At the time of launching into research, one should be in an environment where good mathematics is cultivated, otherwise there is a danger of pursuing trivial research. When you wish to learn a new subject or wish to pursue a new field of research, try to approach the field from as high and as sophisticated point of view that you are capable of.

In the wake of independence, impelled by Nehru's vision, India saw a great flowering of scientific institutions. Among these was the Tata Institute of Fundamental Research, and this is where M S Narasimhan grew from being a graduate student to a towering figure in Indian science. He distinguished himself as a mathematician, teacher and administrator. After his retirement from TIFR, he went on to a second innings as director of the Mathematics Department at the International Centre for Theoretical Physics in Trieste, where he built up a strong school in algebraic geometry. Narasimhan was an extraordinarily versatile researcher, contributing significantly to analysis, representation theory, differential geometry and algebraic geometry. Two related discoveries have in particular proved to be of great importance. The first is the application of stability in classifying a family of algebrogeometric objects, identifying the dominant part that parameterises semistable objects, and then adding other components built out of extensions of smaller semistable objects. The second insight is that stable objects are characterised as those that satisfy non-linear partial differential equations, a discovery that foretold major developments 20 years later.