rick@cs.arizona.edu (Rick Schlichting) (01/14/91)
--------------------------------------------------------------------------
Trip Report -- Electrotechnical Laboratory and (June 28, 1990)
Tsukuba University
Richard D. Schlichting (rick@cs.arizona.edu)
Associate Professor
Department of Computer Science
The University of Arizona
Tucson, AZ 86721, USA
[This report represents the personal opinion of the author, who was on
sabbatical in Japan from Dec. 1989 through July 1990. The sabbatical
was supported in part by grant INT-8910818 from the NSF U.S.-Japan
Cooperative Science Program.]
---------------------------------------------------------------------------
On Thursday was my day-trip to Tsukuba ``Science City'' to visit
Tsukuba University and the Electrotechnical Laboratory. The former is a
relatively new university strong in science, while the latter is the
premier national laboratory for research in computer science. When I
arrived at Tsukuba Center on the bus from Tokyo Station, I was met by
Dr. Yutaka Ishikawa, who works in the Programming Language section of
ETL. He graduated from Prof. Mario Tokoro's laboratory at Keio University
about 4 years ago and has worked at ETL since then, although he spent one
of those years at Carnegie-Mellon University in Pittsburgh working with
Dr. Hideyuki Tokuda on the development of the ARTS real-time operating system
kernel.
Ishikawa took me by car to ETL, which are a series of very modern
buildings in the AIST research park in Tsukuba. There, he introduced
me to Dr. Kokichi Futatsugi, my official host and the head of the
Programming Language section. As it was close to the time set for
my talk, he gave me only a very brief overview of ETL and his section
at this time. The section which is in the Computer Science Division,
has 9 researchers, of which 4 currently have a PhD (one more is currently
working on it.) He indicated that both the size of the section and
the educational level were about average for ETL. As is
also normal for ETL, there are no technical support staff at
all associated with the section. He said this is one of
their biggest problems since it naturally means that the researchers
have to do literally everything. One technique they sometimes
use to get some technical support is to have long-term visitors from
industry. They're there essentially to learn, but are also utilized
to do programming, etc. About 1 or 2 at a time seems normal. He also
said that students are sometimes used in a similar role.
Following my talk, Futatsugi, Ishikawa, Michiharu Tsukamoto of the
Distributed System Section, and I went to lunch at a restaurant in
one of the ETL buildings. We spent the time talking about CS in Japan
vs. the US, and other general topics. For example, we talked about
recruiting people for ETL. Since ETL is viewed as a prestigious place
to work in Japan, they did not seem to be having problems hiring PhDs
despite a nationwide shortage that I heard about almost every place
I visited. The pool they they draw from is the same as the universities --
those inclined basically to an academic-type environment rather than
industry -- a fact that I think would be similar to national
laboratories in the U.S. In this context, we also talked about the
way in which hiring is done at ETL and how it compares with the
situation in the U.S. For example, people with an MS degree and
wanting a job at ETL have to take a rigorous and competitive exam.
I judged this to be somewhat akin to the civil service exam in the U.S.,
but more more technically oriented and difficult. They are, however,
able to bypass this step for candidates with doctorate degrees.
Following lunch, I sat down with Futatsugi for a short one on
one talk. He first told me about some of the work going on in
his section. His general interests personally are in specification,
especially protocol specification similar to the work going on at
Tohoku University in Sendai. In particular, one thing he's been working
on is a LOTOS interpreter written in OBJ, an algebraic specification
language that was done while he was visiting SRI in California.
He gave me several papers on the OBJ work. Other projects in the
section include a software environment designed to work with algebraic
specifications and tailored specifically to the evolution of software,
and the work on real-time operating systems and languages being done
by Ishikawa.
We also chatted about more general things, such as his
experience at SRI and the support programs for foreign visitors
at the labs. He also described a new tentative MITI program on
New Software Structuring in which he is involved. He made it sound
like a very modest program, although he did make a point in saying that
it was the first MITI program to concentrate solely on software.
My next visit was with Tsukamoto down in the lab of the
distributed systems section. The focus of their work has been
primarily on OZ, a distributed object-oriented operating system based
on OSI standards. The OSI aspect is seen to be significant, as is
the emphasis on object migration. The work is funded as part of the
MITI large-scale project on database interoperability (someone made a
comment that anything involved in "information systems" in a general
sense fell under the database moniker as far as MITI was concerned.)
He gave me a fairly long description of the project, plus a paper.
Some of the researchers then gave me a demonstration involving a couple of
robot arms moving disks on a platform to the tune of a Towers of Hanoi
algorithm.
The next item on my agenda was a visit with Toshio Shimada,
who is involved with the SIGMA and EM projects in the Computer Architecture
Section. We were running out of time, so this consisted primarily of
a presentation on his part about the project. Technically, what I
heard corroborated what David Kahaner wrote in his trip report, with
the slight exception that Shimada stated that EM-4 was intended for
general purpose computing rather than only symbolic computing as implied
in DK's report. When I asked about when the various projects started,
he said that SIGMA started in 1982 and the EM project in 1986. An
interesting side note: the presentation was also attended by 3 US graduate
students who were visiting ETL, one from Colorado, one from Stanford, and
the third from New Mexico State.
At this point, I left ETL and was driven by Ishikawa to Tsukuba
University for my visit there. The campus itself was huge -- I was told
several times that it was the largest in Japan (I believe it too.) The
building that was our destination was back towards the back of the
campus, so we had an opportunity to go through the whole thing. Once
there, I met with Dr. Yoshihiko Ebihara, an Associate Professor in the
computer science department who had graduated from Prof. Noguchi's
laboratory at Tohoku University a number of years ago, and whose current
research interest is networks. After introductions, he proceeded to
describe the department, or, more accurately, the Institute of Information
Sciences and Electronics, giving me a copy of an English brochure as well.
It turns out that the department has 51 faculty members, which is huge by
Japanese standards and large even by US measurements. (Some people are,
however, apparently working in areas that we would consider peripheral
to CS.) There are 96 graduate students, with the vast majority apparently
being in the MS program; I was told there were about 4/year in the PhD
course. There are also about 120 undergraduates/yr in the program,
a 50% increase in the normal enrollment of 80/yr due to the
recent increased enrollments (noted consistently at other universities
as well.)
Following my technical presentation, Ebihara and I continued
our previous discussion together with Prof. Ikuo Nakata, the chairman
of the Institute. We talked some about the structure of the university,
which was established in 1973 more or less on the American university
model rather than the koza (chair) model as is common at other Japanese
universities. Nakata professed to like the system better, mainly because
of the extra flexibility it provides (e.g., came hire into
any research area rather than being constrained by the requirements
of the chair.) Among the negative points about the system was
that it required more meetings among the faculty to resolve issues.
He said that the full professors met approximately 2-3 times/month,
while the entire faculty of 51 met about once every 3 months.
We also talked some about Monbusho support for the Institute, which
appeared to be very good. For example, the size of the faculty is
large, yet there are still getting additional faculty positions to cope
with increased enrollment. We continued this discussion until
about 5:30 PM, at which time Ebihara took me to the train station
to catch the next train to Tokyo.
Addresses
----------
Prof. Ikuo Nakata
Institute of Information Sciences and Electronics
University of Tsukuba
Tsukuba, Ibaraki 305, Japan
Prof. Yoshihiko Ebihara
Institute of Information Sciences and Electronics
University of Tsukuba
Tsukuba, Ibaraki 305, Japan
Dr. Koukichi Futatsugi
Chief, Computer Language Section
Computer Science Division
Electro-Technical Laboratory, AIST-MITI
1-1-4 Umezono, Tsukuba
Ibaraki 305, Japan
Dr. Yutaka Ishikawa
Computer Language Section
Computer Science Division
Electro-Technical Laboratory, AIST-MITI
1-1-4 Umezono, Tsukuba
Ibaraki 305, Japan
Michiharu Tsukamoto
Senior Researcher, Distributed Systems Section
Computer Science Division
Electro-Technical Laboratory, AIST-MITI
1-1-4 Umezono, Tsukuba
Ibaraki 305, Japan
Toshio Shimada
Chief Scientist, Computer Architecture Section
Computer Science Division
Electro-Technical Laboratory, AIST-MITI
1-1-4 Umezono, Tsukuba
Ibaraki 305, Japan