Torben Ægidius Mogensen

Summary of speech at A.P. Ershov memorial session in Novosibirsk 2006

I first heard of Andrei Petrovich Ershov as a student at DIKU in 1984 while following a course about partial evaluation and mixed computation. One of the central papers presented at this course was "Mixed Computation: Potential Applications and Problems for Study" by A.P. Ershov (TCS 18 (1982) 41-67). The paper shows the now standard example of program specialization: Specializing the power function. It also talked about the concept of self-applying a specializer in order to obtain compilers and compiler generators. While very promising, this process had not been achieved in any practical way, and this intrigued me and other students at the course, so subsequently I and a fellow student (Mads Rosendahl, who is now at Roskilde University) embarked on a student project that would attempt to achieve self-application of a specializer. We failed doing this, but a year later it was achieved by other people at DIKU (Neil Jones, Peter Sestoft and Harald Sondergaard), sparking a world-wide interest in the subject.

I was still very much interested in the potential of partial evaluation, so I wrote my Master's Thesis about applying partial evaluation to ray-tracing, and shortly afterwards, in 1986, began a Ph.D. study on partial evaluation. At this time Neil Jones from DIKU and Dines Bjorner from DTU in Lyngby were planning a workshop about partial evaluation and mixed computaion, attempting to bring together both old and new researchers in the field.

In October of 1987, this came to be as the 1987 workshop on Partial Evaluation and Mixed Computation (PEMC) at "Gammel Avern?s" on the island of Fyn in Denmark. This workshop was unusual for the time by having a large number of participants from the then Soviet Union. From Novosibirsk, obviously, came A.P. Ershov and also the recently departed Alexandre Zamulin. The other representatives of the USSR were Sergei Romanenko from Moscow, Sviatoslav Lavrov from St. Petersburg (then Leningrad), Boris Ostrovski from Barnaul, Nikolai Nepejvoda from Izhevsk and Jan Barzdin from Riga. Also attending was a Russian expatriate, Valentin Turchin from New York University.

A.P. Ershov was a central figure at the workshop, giving a keynote speech that provided a personal view of the early history of partial evaluation and mixed computation. He also co-edited the proceedings of the workshop that came out partly as a book from North-Holland and as a special issue of the New Generation Computing journal.

Unuaually for a scientific workshop, the PEMC workshop was mentioned in several national Danish newspapers, with special mention of the illustrious international guests, Academian Ershov from Novosibirsk and John McCarthy from Stanford University, as well as the high number of soviet participants.

The workshop had very much a pioneer spirit to it, and it was very influential on the future of research in partial evaluation and mixe computation.

After the workshop, A.P. Ershov made a brief stop at the University of Copenhagen before returning to Novosibirsk. This was the last time I saw him, as he died much too early the year after, in December 1988.

When the first Ershov memorial conference was held in Novosibirsk in 1991, I attended it as a representative from the partial evaluation research group at DIKU. I found the spirit and the place so much to my liking that I also attended all of the later memorial conferences.

While I only met A.P. Ershov briefly, I find that he has been an inspiration in my work. In particular, his work on partial evaluation and mixed computaton has been very influential on my own work in this area, but also in the idea of bringing together theoretical and practical research is something where I follow his lead, trying to be more than "just" a theoretician or "just" a practitioner.

The way Ershov inspired his students is someting I also aspire to, but I fear I can never reach his level in this.

Finally, I thank you for inviting me to speak at this memorial session, which I feel is an honour.

Torben Ægidius Mogensen

© 2008-2024 IIS SB RAS