Anirban DasGupta shares his memories of Shanti Gupta, a “quiet scholar, architect, and shepherd”:

Shanti Gupta devoted 40 years of his life to the foundation and nurturing of the statistics department at Purdue University, USA. He was a selfless soldier of the profession, an effective and yet gentle fighter. He had the nose of a hound, the energy of an athlete, and a dogged resilience in building a department in the US Midwest that earned the respect of the world. An architect and visionary, uncompromising on quality of work, a soft-spoken leader with incredible people skills: that is the Shanti Gupta I knew for 20 years.

He was born in a remote, small village in the Indian state of Uttar Pradesh in 1925. From modest beginnings, he did his formal education at Delhi University and the University of North Carolina at Chapel Hill, where he was an early and distinguished PhD student of Raj Chandra Bose. In time, he was the President of the IMS, Editor-in-Chief of the JSPI, an advisory member of the National Research Council, a special chaired professor at the Academia Sinica, an Erskine Fellow, and faculty member at Stanford, Berkeley, and the Courant Institute in New York. He was elected a Fellow of the IMS, the AAAS, and the ASA. At the time of his passing in 2002, he had 29 PhD students, including several who are now leaders of the profession.

In 1968 he founded the Purdue University Department of Statistics, and served as department head until 1995. I could write a whole article just about the world-class people he recruited, using his hound-like senses. He never missed a colloquium talk, and could judge within the first 15 or 20 minutes, who had “substance.” Here is a very short list of people he hired in his department: Joop Kemperman, Herman Rubin, Sree Pillai, Prem Puri, Jim Berger, Bill Studden, Jayanta Ghosh, Andrew Rukhin, David Moore, Leon Gleser, Burgess Davis, Steve Lalley, Tom Sellke, Glen Baxter, Mary Ellen Bock, Virge Anderson, George McCabe, Wei-Liem Loh, Tony Cai, Dimitris Politis, Rebecca Doerge, Ker-Chau Li. During his 28-year tenure as Head of the Purdue statistics department, the best shining stars of the profession visited regularly: D. Basu, Susie Bayarri, Peter Bickel, Larry Brown, George Casella, Y. S. Chow, Morris DeGroot, Holger Dette, Persi Diaconis, Malay Ghosh, Len Haff, Iain Johnstone, Jack Kiefer, T.L. Lai, Lucien Le Cam, Carl Morris, Tony O’Hagan, Ingram Olkin, C.R. Rao, Christian Robert, Jerry Sacks, Pranab Sen, Bill Strawderman, Bob Wijsman, Michael Woodroofe, Grace Yang, and numerous others. During this period, the department became (and remained, until the end of his tenure in 1995) ranked seventh nationally; statistics faculty were made Sloane Fellows, Guggenheim scholars, IMS Presidents, COPSS award winners, IMS special invited and Medallion lecturers, ASA Presidents, editors of several of the Annals… Purdue University Honorary DSc degrees were given to some prominent leaders of the profession. Shanti had his focus on one dot on the wall: become and remain a top-10 department in the USA. The Purdue statistics department was his family, and he was about the best “father” possible. Selfless, responsible, societally ambitious, encouraging of the “cream,” not afraid to trim. He maintained wide international networks to know quickly who to recruit.

He launched the Purdue International Symposia, co-chairing it from 1976 to 1997, and it became the most prominent international conference in statistics and probability after the Berkeley symposia folded.

His publications were many, including several well-known books; he was a household name in selecting and estimating largest means, or a group of largest means. These have potential for concrete use in the high dimensional variable selection problems of today. His beginnings were certainly classical, but in the mid-eighties, he ventured into Bayesian and full decision theoretic solutions of some of his problems. Sometimes he did empirical Bayes. He would always do a theoretical study of the operating characteristics of his methodologies. He would sometimes work out convergence rates of probabilities of correct selections, sometimes convergence rates of Bayes risks or regrets, conservative minimax such as Γ-minimax. It is astonishing that 63 years ago, with hardly any computing power available, he doggedly produced tables of the joint CDF of multivariate normal and multivariate t-distributions. He also did a lot of work that was original and practically useful on reliability and statistical quality control. Many people are surprised that Shanti Gupta wrote a number of articles on Mahatma Gandhi’s ideas on labor-capital relations and economic philosophy, whether Gandhi was an economic egalitarian.

Shanti Gupta was a gracious person, with a gentleman’s manners. Hosting faculty or members of external review boards at his home, he would serve the best wine in the world. He enjoyed north Indian food immensely, and his wife Marianne learned to make a good chicken curry from him.

Shanti Gupta was one of those vanishingly rare types: a man of quiet scholarliness, an enthusiast, a self-minimizer, and the epitome of integrity. In short, a rare, delicate, elegant, and in some ways a peerless statistical academic. I miss him. He should be missed, a whole lot.