My Japan speaking tour

31st January 2020

As my EPSRC Fellowship was coming to an end, I decided that I needed to advertise my research on Benders' decomposition to an international audience. The best way I thought to achieve this was by visiting research institutes and giving seminars. This lead to a tour of Japan with my friend Dr Yuji Shinano.

I used research visits in Japan as an opportunity to advertise the Benders' decomposition framework that I developed during my fellowship. I felt that this was quite successful. A little unexpectedly, I was able to discuss the fundamentals of Benders' decomposition and really explain the key ideas that drive implementation decisions. During the trip there were many times where I found that this level of detail was really appreciated.

The aim of this blog is to talk about a use of an interesting use of research funding that is valuable to you, your research and your academic career.

How did it come about?

As I said previously, I wanted to use my EPSRC Fellowship funding to advertise the results from my research. My friend, Dr Yuji Shinano, is originally from Japan and has a large number of contacts at different universities. He also travels to Japan fairly frequently to visit these colleagues. So, I simply asked Yuji whether it would be possible for me to travel to Japan with him and visit some universities.

The timing was perfect, Yuji was due to be in Japan to give a keynote talk at CANDAR conference in Nagasaki at the end of November. Given this as the anchor for the trip, we started to put together a chain of visits to institutes working our way from Nagasaki to Tokyo.

The tour

During the trip, we visited 4 research institutes and 2 companies. We visited, in order, the Institute of Mathematics for Industry (IMI) at Kyushu University, the Research Institute for Mathematical Sciences (RIMS) at Kyoto University, the Energy Systems Analysis Laboratory at Osaka Prefecture University, and the Institute of Statistical Mathematics (ISM) in Tachikawa. We also visited the companies: NTT Data Mathematical Systems Inc (MSI) and Hitachi.

An important part of this tour was that I had Yuji to guide me around. Not only did he know the professors at the research institutes that we visited, he was also able to direct me to interesting places and found great restaurants. My job in many cases was to find good coffee and beer, something that I have a bit of experience in. Because of my interest in beer, however, Yuji also informed our hosts that we would welcome recommendations to get good beer.

There was a great diversity of groups that I presented my work to. At IMI and RIMS, we visited there primarily to give a seminar. The groups were interested in our work, but there were not any planned meetings. At these two institutes, I had the challenge of explaining Benders' decomposition to an audience with limited knowledge of mathematical programming. I didn't get much feedback, but I think that I did ok.

IMI and RIMS are located in two very different cities, Fukuoka and Kyoto. They we both really interesting to visit. Fukuoka had the better beer and coffee, but I did mange to run into one American and two Canadians at a brewery with an Australian head brewer in Kyoto. In Fukuoka I was introduced to the cuisine Kushiage. This is well worth checking out.

Our visit to Osaka Prefecture University was much more involved. Here Yuji has worked quite extensively with Professor Ryohei Yokoyama on a decomposition approach used for solving optimisation problems in energy systems. We had two days of meetings and seminars at the university. This was really interesting, since the researchers and students are using Benders' decomposition and Dantzig-Wolfe reformulation. So there was a lot of work to discuss.

The stop at did not involve as much beer and coffee drinking. This was mainly because Osaka Prefecture University is located to the south of Osaka and we stayed in Sakai, which is more of a transport location. We did get taken to a nice craft beer bar near the university. Also, I experienced another great cuisine, Okonomiyaki. It was initially described to me as Japanese pizza, but I think it is more like a Japanese savoury pancake.

Our final research institute stop was ISM in Tachikawa. Similar to IMI and RIMS, this visit was about giving a seminar to the institute. By this stage we had our routine well prepared. So it didn't matter too much that the audience didn't have extensive knowledge with mathematical programming. We worked out a good strategy for explaining the key details to really sell our research.

Our host at ISM really enjoys beer. Also, there is a thing in Japan where you can pay a flat fee for all-you-can-drink alcohol for a set time. We went to a craft beer bar, which was really nice but also really dangerous, when we decided to try this flat fee thing. The result was a really good fun night, with lots of laughs.

The two company visits were very different and very interesting. First, I was scheduled to give a talk at MSI to their company, but also to any other interested people from outside the company. Our colleague, Koichi Fujii at MSI, organised the seminar and also advertised it to the Japanese OR Society. So we had a good range of people coming to the seminar. I also seemed to get a lot of interest in my work, which was a very good outcome. The final visit was to Hitachi, which was more for a project meeting than to give a seminar. It was a valuable stop on the tour.

The talk

I was worried before this trip that I would end up having to write a number of different presentations. Luckily, during the preparations I realised that I could get away with just one presentation, but tweaked for different audiences. This had two very positive outcomes. The first was that I didn't need to put in as much work in preparing presentations for the tour. The second was that during the trip I had a lot of practice in giving the seminar, so I became very good. Each time my seminar was much better than the previous. At our last visit, Yuji even told the audience that they got the best version of the talk. It is great when you are able to confidently give a well rehearsed presentation. It just makes it much better for you and your audience.

The outcomes

It is hard to know from any research trip what will be achieved. This goes for conferences and research meetings. In this case, it was a little harder since we visited such a diverse range of institutes with different levels of expertise in mathematical programming. In the end, I felt that there were many positive outcomes from this trip.

The first, from a personal perspective, I got the opportunity to visit many different places in Japan. This was a wonderful experience, made much better from the fact that I was travelling with Yuji. He is a great person to travel with. The second, from a professional perspective, I was able to make new contacts at a number of research institutes. You never know when you will need contacts, so it is good to make connections when you have the opportunity. Third, I was given the opportunity to share my knowledge on Benders' decomposition, from fundamentals through to the implementation. This is something that I have not done before, so I now have quite a lot of experience doing this. Finally, from the EPSRC Fellowship perspective, I have increased the reach of my fellowship research by advertising the work to research institutes and companies in Japan. The hope is that this tour will increase the number of people using my software and promote more research into the algorithmic aspects of Benders' decomposition.

Overall, it was a great visit and a really good opportunity. I would suggest that anyone preparing a research proposal to consider the opportunity to take your work on tour. It is a great way to promote your research and you can have a lot of fun in the process.