Visit To APL Centres in Russia November 1993

Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 44

I joined the BAA in 1992, and soon discovered through Vector that APL was also being used in Russia, and that I had just missed a conference in St. Petersburg. By emailing Erkki Juvonen, I was able to get email addresses for contacts in Russia. I was interested in this for two reasons: firstly I had a long-standing interest in what was then the USSR, having written various books and articles about it; secondly my colleague at Bradford, Gavan Duffy, who had introduced me to APL, was working for 6 months of each year in South Africa, contributing to the construction of a family of models of the South African economy using Dyalog APL. In Bradford we were also working in J to build an elementary demographic model of South Africa, displaying the results in a Geographical Information System. I was very keen to find out what the Russians were doing in the field of economic modelling, as well as in other areas.

I received a grant from the British Council for Gavan and me to visit Moscow and Obninsk to explore the possibilities for collaboration between Bradford and the APL communities in these cities. However, at the last minute, Gavan was recalled to South Africa and was unable to accompany me.

Moscow¶

The Computer Centre of the Russian Academy of Sciences (RAS) is well known as the place where the original Sputnik calculations were done. It is also where the game Tetris was invented. I was received extremely warmly, and was delighted to discover that the approach to economic modelling was complementary to that being used in South Africa. There is a very real basis for cooperation, in my opinion. We drew up a draft agreement for funders. I was told that Professor Igor Pospelov and Dr. Andrei Buzin had been prevented from attending Toronto because of delays in receiving visas, and that they expected to be at Antwerp.

The experience of modelling goes back about 20 years, and APL seems to have been used since the late 1970s. The approach to modelling is quite original, compared to British techniques which tend to rely on a lot of data. In Moscow, the technique is to do thousands of iterations before deciding on the parameters of the model, which can then be run with a fair degree of confidence on a relatively small amount of data. They have a similar philosophy with regard to the transparency of the model to that held in South Africa, namely that the assumptions should be as clear as possible and that the model should be used as a basis for debate on economic policy, rather than function as a black box from which unchallengeable conclusions are drawn.

They have been modelling structural changes in the economy since long before Perestroika became a term known in the West. The behaviour of economic agents is described explicitly, and the macro-economic model is built up from the "microdescription" of the behaviour of economic agents. The development of models was done in three stages, beginning rather surprisingly with three models of a market economy. Then a model of a centrally planned economy was developed. The features of this model were extremely interesting to me since they indicated problems which Western economists were beginning to detect in the Soviet economy in the late 1970s. Clearly these problems were being addressed at that time, at least by these modellers. Was anybody listening? A model of an economy undergoing radical reform was then constructed, apparently late in the period of Perestroika. This predicted the hyperinflation which occurred two years later under the Gaidar reforms. Such a prediction was also being made in May 1990 on analytical grounds, without the benefit of a model, by Professor Hanson of Birmingham, surely one of the best British economists in the field of the former Soviet Union. I understand that now the RAS Computer Centre team has the chance to obtain more recent data from the Ministry of the Economy in Russia.

In addition to the models, the other most interesting work consisted of GRAN, a GRaphical ANalyser, which is an interactive graphical manipulation package, unlike anything that I had encountered. It means that many statistical methods and well known economic models can be graphically interpreted in an interactive mode. I see it as a useful complement to the APL Statistics Library work based at Swansea. Most interesting of all to most APL users is the work being done on a compiler. I know that they have had what they call a pre-compiler for some years, but they have been working on a full compiler. Realising the significance of this I did my best to make sure that we were not losing something in translation. We may get a paper on this at Antwerp. I have since received an email communication saying that further progress has been made on the compiler.

I hope to return to Russia to explore these matters in more detail in future. I also sincerely hope that they give a paper in Antwerp, since the Vector discussion of their work at the St. Petersburg conference does not really show how interesting it is. A final point is that a commercial company is using Dyalog APL to construct data bases. A discussion of the relative merits of such data bases and Paradox for Windows was held on the train to Obninsk.

Obninsk¶

Obninsk was a closed city, with a closed Zone inside the city, until December 1991. The reason is that it was and remains the heart of the Russian nuclear industry. The world’s first nuclear power station started generating on 27th June 1954. I was taken right into the reactor core by Sasha Skomorokhov, who will be well known to participants at the St. Petersburg and Toronto conferences. I had read his Toronto paper on adaptive learning networks, as well as those of other Russian Toronto participants, most of whom I met in Obninsk. Apart from such work which can be found in Vector and APL Quote Quad, there is also very interesting work going on in APL in the fields of Computer Assisted Learning (CAL), statistics and personality profiling. I found this out when I met staff from the Obninsk Training Centre.

This is the main training centre for the Russian nuclear industry, with its own hotel (where I stayed) and dedicated training building. The Centre is suffering from the fact that outside agencies now have to purchase their training, and many are now unable to do so in a situation of hyperinflation. Nevertheless, they are continuing to work on, for example, CAL material for training nuclear engineers. I had never seen APL used in this manner before, with moving diagrams of the flows within a power station. Having just received a detailed lecture on the design of the first such power station, I was able to follow the logic of the lessons. The statistics work is unlikely to be discussed at Antwerp, because the person concerned does not feel sufficiently confident in English. We discussed the relative merits of SPSS and APL for exploratory data analysis.

The personality profiling for nuclear power station engineers was very interesting, using a panel method for assessment by colleagues and a member of the management, which is then analysed using APL statistics programmes and discussed with the subject. Again, unfortunately, the language barrier makes it very unlikely that a paper will be presented at Antwerp, despite my attempts to encourage those concerned to do so. I feel that the work would be of interest to management in the West, being of quite general applicability, but with interesting differences from some Western staff appraisal methods, because of its emphasis on safety. One might see this work and that of Sasha Skomorokhov as part of a post-Chernobyl attempt to improve safety in the nuclear industry. While I discussed Chernobyl with Sasha, I did not at the time think to link these aspects of the work in Obninsk.

I also gave a lecture in English at the Atomic Energy Institute (with Sasha translating, except when I forgot at one point and spoke in Russian, which he translated into English!) on current problems in the British economy. I spoke to various economists and received papers on monetary theory. I also met two Vice Rectors of the Atomic Energy Institute and signed an agreement on research collaboration, not only in the area of APL, between it and Bradford.

Conclusion¶

I am now seeking funds for further research collaboration with both the RAS Computer Centre and the Obninsk Atomic Energy Institute. Apart from the extremely interesting work which I encountered, I had an unforgettable time with a warmth of hospitality that is hard to describe. For that reason alone, I want to return, but I also feel that the work going on there is very important for the APL world community. The hyperinflation is having very adverse effects on the APL work there, and financial support is needed. I hope that a way can be found to augment the existing fund.