Advice for (Prospective) Graduate Students

A topic that is being increasingly covered in TCS blogs is that of giving advice to (prospective) graduate students and beginning researchers. (See, for instance, here, here or here.) This is a welcome development, and a very good way of using the medium for the benefit of an important component of our research community. (After all, young researchers are the future of research, aren't they?) In fact, I have no problem in admitting that I enjoy reading those blog posts or anything similar myself. I feel that I am still learning on the job every day, and that those pieces of advice remind me of things that I should keep in mind, but that I tend (consciously or unconsciously) to "forget". After all,

Advice is what we ask for when we already know the answer but wish we didn’t. (Erica Jong)

The latest few words of advice on research for graduate students I read have been penned down by Fan Chung. She addresses mostly combinatorialists, but what she says applies equally well to theoretical computer scientists at large. I like the fact that she stresses the collaborative nature of the research enterprise, and that she embraces one of my favourite hobby horses, viz. the Hardy-Littlewood rule: authors are alphabetically ordered and everyone gets an equal share of credit. She adds:

The one who has worked the most has learned the most and is therefore in the best position to write more papers on the topic.

(I had never thought in these terms myself, but yes that's true.) She also writes:

If you have any bad feeling about sharing the work or the credit, don't collaborate. In mathematics, it is quite okay to do your research independently. (Unlike other areas, you are not obliged to include the person who fund your research.) If the collaboration already has started, the Hardy-Littlewood rule says that it stays a joint work even if the contribution is not of the same proportion. You have a choice of not to collaborate the next time. (If you have many ideas, one paper doesn't matter. If you don't have many ideas, then it really doesn't matter.) You might miss the opportunity for collaboration which can enhance your research and enrich your life. Such opportunity is actually not so easy to cultivate but worth all the efforts involved.

I could not agree more. I will add Fan Chung's advice to the list of links I suggest to all my students and colleagues. Maybe you'd like to do so too.

Wolf Foundation Prizes for 2008

The Wolf Foundation has released the list of awardees of the Wolf Prizes 2008. Pierre Deligne and Phillip Griffiths, both of the Institute for Advanced Study, and David Mumford of Brown University, will share the 2008 Wolf Foundation Prize in Mathematics. (I note that the third of these mathematicians seem to be doing research with a high computational content these days!)

Claudio Abbado is one of the two recipients of an award in the arts. I did a quick-and-dirty Google search and it seems that none of the main Italian newspapers has devoted an article to this prize. Sadly, I am not surprised.

Concurrency Column for the February Issue of the BEATCS

I have just posted the concurrency column that will appear in the February 2008 issue of the Bulletin of the EATCS. This installment of the concurrency column is devoted to a double bill, as it offers the following two contributions:

• Part 1: Dynamic Spatial Logics: A Tutorial Survey by Luis Caires. [PDF]
• Part 2: Applying Concurrency Research in Industry: Report on a Strategic Workshop by Luca Aceto, Jos Baeten, Wan Fokkink, Anna Ingolfsdottir and Uwe Nestmann. [PDF]
  

The first piece is a survey devoted to spatial logics contributed by Luis Caires, one of the prime movers behind the development of this exciting kind of specification logics. Since the original work of Pnueli, (temporal) logics have been a prime formalism for the description of the behaviour of concurrent systems. Spatial logics are specification logics for describing the behaviour as well as the spatial structure of concurrent systems. Despite being only a fairly recent addition to the family of specification formalisms for concurrent computation, spatial logics are already the subject of a large literature reporting on a substantial body of non-trivial results. Luis Caires’s survey gives us a highly readable and welcome bird’s-eye view of this fast-moving subject.

The second contribution is a strategic report on applying concurrency research in industry. This non-technical article is one of the outcomes of the Workshop on Applying Concurrency Research in Industry, colocated with CONCUR 2007 in Lisbon, which I co-organized on behalf of IFIP WG1.8 “Concurrency Theory”. The essay tries to distil the contents of the presentations delivered at the workshop, and of the ensuing discussion, for the benefit of the concurrency community as a whole. I thank all the participants in the workshop for their contribution to the event. Special thanks go to the invited speakers (Vincent Danos, Hubert Garavel, Jan Friso Groote and Kim G. Larsen) for making the trip to Lisbon to share their considerable experience on this topic with us, and to Rance Cleaveland, Joost-Pieter Katoen, Moshe Vardi and Frits Vaandrager for providing their answers to the questions that the audience asked the invited speakers at the workshop.

I hope that you will enjoy these contributions, and that you will feel enticed to contribute to the ongoing discussion within the concurrency theory community.

Last, but not least, do contact me if you'd like to contribute a piece to the column!

Using Model Checkers in "Intro to OS" Courses

The advent of mature model-checking tools has made algorithmic model-based verification much more accessible to the average computer scientist/engineer. So much so that we can now teach first-year students to use model-checking tools by sweeping essentially all of the theory behind them under the carpet. See, for example, the very recent experience report:

Roelof Hamberg and Frits Vaandrager. Using Model Checkers in an Introductory Course on Operating Systems. Technical Report ICIS-R07031,
ICIS, Radboud University Nijmegen, December 2007.

I strongly recommend reading this report to anybody with even a passing interest in formal methods. It is well written, content rich, and presents material that can inspire many of us in the lecture room. I myself wish that the paper had been posted last August, when I was planning (at the last minute) a second-year course on Operating Systems. I would have used the authors' tips and teaching material then. Not to mention great quotes like:

“Programs are not released without being tested; why should algorithms be published without being model checked?” (Leslie Lamport)

Last November, on the spur of a sudden moment of inspiration, I did use the Uppaal model checker in a one-week intensive course on operating systems for engineering students, and the results were very encouraging.

Frits and his coauthor have done all of us a great service by making their material available on the web, and by sharing their experience with the rest of us. Let's expose our students to easy-to-use model checkers like Uppaal from the first year of the studies in CS and Engineering. This will also a increase the impact of research in formal methods. Several of the students taking my one-week course wrote in their course evaluation that they were glad to have been exposed to Uppaal, and that they think they will use the tool again in their future studies on, e.g., control systems. This indicates that, once students have seen how useful model checkers are, they will be enticed to use them later on when facing similar problems. Last, but not least, the mathematically inclined students may be motivated to carry out independent studies and (under)graduate research in formal methods and other areas of theoretical computer science.

Quoting from the concluding section of the paper:

“Why should algorithms be explained without the use of a model checker?”

Indeed, why not? It will be a good day for the construction of reliable software systems when our students will routinely simulate and analyze concurrent algorithms using model checkers. For the moment, let's share our teaching experiences following the example set by Frits and his coworker, and let's make model checking and logic permeate our undergraduate education as much as possible.

Author Contributions, Redux

Some time ago, I pointed out an article with a very detailed account of the authors' contributions. Here is another similar case I have just seen, via Galileo. (Look at the acknowledgements.) This makes me wonder whether it is standard practice in the medical literature to list what each of the authors contributed to a paper. Is it?

Knuth and Me (Guest Post by Sergey Kitaev)

Guest post by Sergey Kitaev, a very good colleague of mine from the combinatorics group at Reykjavík University.

To show the significance of Donald Knuth in my life, it would probably be enough to indicate that he introduced in 1969 the area of “permutation patterns” which is the field of my main research interest in combinatorics that I’m dealing with almost every day. However, while thinking on the subject, it comes to mind the personal communications with Knuth at Mittag-Leffler Institute at the beginning of 2005. In particular, after attending my talk on partially ordered generalized patterns, Knuth decided to include a result of mine in volume 4 of “Art of computer programming.” It is remarkable that Knuth was collecting information for this volume for over 40 years! However, this was not the thing I was offered Knuth’s famous $1.28 reward for. Unlike most other rewards, this one was not directly related to mathematics – I let Knuth know the middle name of Alexandr Kostochka that he recorded in his name database both in English and Russian (Knuth is able of writing things in Russian which he learned in college, and I find this to be impressive). In any case, stupidly enough, I refused taking the check from Knuth, which would be a nice souvenir as I realized later on; I simply told Knuth that it was a great pleasure for me to be helpful for him …

Addendum. A (permutation) pattern is a permutation of a totally ordered set. An occurrence of a pattern P in a permutation p is a subsequence of letters of p whose relative order is the same as that of the letters in P. As an example, the permutation 461352 has three occurrences of the pattern 321, namely the subsequences 432, 632 and 652.

The initial motivation for studying pattern avoiding permutations came from its connections with container data types in computer science. In 1969 Don Knuth pioneered this work by showing that the stack sortable permutations are exactly the 231-avoiding permutations.

A Free Journal-Ranking Tool

The latest issue of Nature feutures a news item reporting on a freely-available tool that can be used to generate citation statistics for papers, journals and countries. The SCImago Journal & Country Rank is a portal that includes the journals and country scientific indicators developed from the information contained in the Scopus® database. This platform takes its name from the SCImago Journal Rank (SJR) indicator, developed by SCImago, a Spanish data-mining and visualization group. This indicator is based on Google PageRank. This tool is a competitor to Thomson's Web of Science, and covers more journals (15,000 in lieu of 9,000) and 20-45% more records than the Web of Science.

The availability of this tool, as well as of Google Scholar of course, puts Thomson under some pressure. I think that this is welcome pressure. To see why, you might wish to read this editorial. Basically, the "impact factor" is one of the Gods of modern-day academia, together with "leadership" and a few other criteria not necessarily related to scholarship. It has "a strong influence on the scientific community, affecting decisions on where to publish, whom to promote or hire, the success of grant applications, and salary bonuses. " However, as claimed in the editorial, "members of the community seem to have little understanding of how impact factors are determined, and, to our knowledge, no one has independently audited the underlying data to validate their reliability." This is obviously undesirable.

I think that, for good or for worse, impact-factor-based evaluation of our research output is here to stay. However, when making decisions based on impact factor, citations and what not, I hope that deans, employers, funding agencies and rectors will consult several different sources and compare the results that they get. Moreover, I do hope that good, old-fashioned evaluation of the quality of one's work will not disappear altogether to be replaced by purely quantitative indicators.

For the moment, let's play with our new toy. In case you are interested here are the rankings of countries in computer science: all subjects, computational theory and mathematics, TCS (but as a subcategory of mathematics), logic (as a subcategory of mathematics) and mathematics as a whole.
Draw your own conclusions.

The Theoretic Centre of Computer Science

The latest issue of SIGACT News features a very entertaining piece entitled The Theoretic Center of Computer Science by Michael Kuhn and Roger Wattenhofer. This article is a "printed version of a frivolous PODC 2007 business meeting talk, held by the second author". It speculates on the central conferences and researchers in computer science, with emphasis on theory, and makes for excellent after-lunch reading. I recommend it to the readers of this blog.

In this post, I'd like to offer a couple of comments on the list of most central authors in computer science. First of all, how do Michael Kuhn and Roger Wattenhofer determine how central an author is? Here is the relevant excerpt from the paper.

Analogously to the construction of the Erdos number, we base our method on the co-author-graph. We then create the induced subgraph for each region of interest (PODC, STOC/FOCS/SODA, and computer science). For PODC, for example, this graph would only contain authors that have at least one PODC paper, and so on.

Other than in the construction of the Erdos number, we do not rely on shortest paths, but rather on the PageRank idea: We start several short random walks at different nodes of the graph, and count how often each author gets visited. This idea is then extended to time dependent centrality, by starting the random walks only at authors that have published in the last five years.

What are the results of this approach? Table 1 on page 62 of the paper gives the most central authors for computer science as a whole. Alberto Sangiovanni-Vincentelli (another Italian expat) tops the all-time list, and Noga Alon is the runner-up. Without doubt, the list contains only major players, and concurrency theorists will be happy to see Moshe Vardi at #7 on the list of most "central" authors in computer science. Some analysis of this table is provided in the paper. Here are a couple of quick thoughts and questions from yours truly.

• Computer scientists working in the area of databases feature prominently in the all-time most central authors in CS. Indeed, a look at the list of researchers with the largest number of entries in DBLP, the data set used by the authors, shows that some of the most prolific authors in CS work in that area. Could it be that people publish more and have more coauthors in the field of databases, broadly construed, than in other areas of CS?
  
• The list of most central authors does not include any Turing award winner, as far as I can tell at first sight. Does this mean that Turing award winners are not central? Of course not! To my mind, this just means that an analysis of the collaboration graph favours prolific authors with lots of coauthors. Consider, by way of example, two giants of CS research like Tony Hoare and Robin Milner, who are not even in the list of top 1000 CS authors. (Neither are Stephen Cook, Don Knuth, Gordon Plotkin or Leslie Valiant to name but a few giants, by the way :-)) By modern standards, their "number of papers" is not outstanding. However, their ideas and writings have had, and still have, enormous impact on computer science research. All of what I have done myself, for instance, is built on their original work, which has kept many computer scientists busy for about thirty years. (Of course, Robin and Tony are not responsible for the noise I have generated myself :-)) These tables are a fun read, and they do tell us something worthwhile about the players in our subject. However, as the authors point out themselves, "Without doubt the future will teach our evaluations a lesson, ultimately revealing in which direction computer science evolves, and maybe even discover the most influential computer scientist. After all, research is not about how many papers we write, or how many citations they get, but rather, what the best contributions are."
• I think that the right-hand side of Table 1 is strongly influenced by the phenomenon of name ambiguity. Consider, for instance, Wei Wang, who is ranked as #2. When I saw that Wei Wang has 86 DBLP entries for 2006, I asked myself the question: "Who is Wei Wang"? This Google search return over 1.8 million entries! It seems to me that Wei Wang is a (very large) disciple of Bourbaki or Lothaire :-)
  
• Table 2 on page 62 is a veritable who's who in the FOCS/STOC/SODA branch of TCS. It has a truly amazing number of Israeli computer scientists (six out of ten on the "all-time" list, and four out of ten on the "last-five-years" list, I believe).

Enjoy!

AMS Prizes 2008

Via the Geomblog, I see that the list of AMS prizes for 2008 is out.

First of all, let me join the chorus of congratulations for Shlomo Hoory, Nati Linial and Avi Wigderson, who have been awarded the Levi L. Conant Prize for the best expository article in the Notices or the Bulletin of the AMS in the last 5 years. They received this prize, which is a great recognition for TCS research within the mathematical community, for their article Expander graphs and their applications in the Bulletin of the AMS.

As Nati Linial says in his response for the prize

I believe that the full potential impact of combinatorics on the rest of mathematics is only starting to reveal itself and the study of expander graphs can give us some idea of the true power of these connections.

Combinatorial research is also honoured with the Leroy Steele Prize for seminal contribution to research going to Endre Szemeredi for the paper On sets of integers containing no k elements in arithmetic progression, Acta Arithmetica XXVII (1975), 199–245. I love these words in Szemeredi's response:

This award could not have occurred were it not for the fundamental work of
other mathematicians who developed the field of additive combinatorics and
established its relations with many other areas. Without them my theorem is only
a fairly strong result, but no “seminal contribution to research”.

Calling his theorem a "fairly strong result" is really faint praise, but I like the fact that Szemeredi points out that a result becomes a seminal contribution to research when it used by other researchers to obtain fruits that were considered beforehand too high on the tree of knowledge to be picked.

The prize booklet makes for some interesting reading. Wearing my Italian expatriate's glasses, I note in particular the two awards to Italian mathematicians (Alberto Bressan and Enrico Bombieri) , both of whom work in the US.

I look forward to seeing who will be the recipient of the EATCS award for 2008.

The Dangers of Blogging

In this entertaining TED talk, Yossi Vardi issues a word of warning for the male bloggers out there, and addresses the "local warming" problem.

I know that this post is somewhat different in nature from my typical ones, but it's the first day of term and a little after-lunch entertainment was called for :-)

Enjoy!

Science Magazine's Breakthroughs of the Year 2007

Anders Claesson just alerted me to the fact that Solving Checkers has been listed by Science magazine in tenth position in the list of breakthroughs of the year 2007. See here.

My colleague Yngvi Björnsson from the School of Computer Science at Reykjavík University was a member of the team behind this breakthrough. Congratulations to Yngvi and his colleagues at Alberta.

A Good Example from Canada

I recently became aware of the Canada Research Chairs programme. That programme has been running since the year 2000, and aims at establishing 2000 research professorships—the so-called Canada Research Chairs—in universities across Canada by 2008. The Canada Research Chairs programme invests $300 million a year to attract and retain some of the world's most accomplished and promising minds.

I encourage the readers of this blog to have a look at the web site for the programme. There is a lot of interesting material there, and I cannot help but think that many countries would be well served by setting up a similar programme to attract the best possible scientists in all disciplines. Now, this is something well worth lobbying for in the coming year, isn't it?

If you do not have time to look at the web site I linked to above, here is my executive summary of the programme.

• Each eligible degree-granting institution in Canada receives an allocation of Chairs. For each Chair, a university nominates a researcher whose work complements its strategic research plan and who meets the program's high standards.

  Three members of a college of reviewers, composed of experts from around the world, assess each nomination and recommend whether to support it.

• Universities are allocated Chairs in proportion to the amount of research grant funding they have received from the three federal granting agencies: NSERC, CIHR, and SSHRC in the three years prior to the year of the allocation.

• There are two types of Canada Research Chair:

  Tier 1 Chairs, tenable for seven years and renewable, are for outstanding researchers acknowledged by their peers as world leaders in their fields. For each Tier 1 Chair, the university receives $200,000 annually for seven years.

  Tier 2 Chairs, tenable for five years and renewable once, are for exceptional emerging researchers, acknowledged by their peers as having the potential to lead in their field. For each Tier 2 Chair, the university receives $100,000 annually for five years.
  

As you can see, there is a strong financial incentive to attract people to these endowed chairs!


• Chairholders are also eligible for infrastructure support from the Canada Foundation for Innovation (CFI) to help acquire state-of-the-art equipment essential to their work.

• If an institution's performance decreases relative to other institutions to the extent that the next recalculation of Chair allocations results in that institution's allocation being reduced, the Chairs Secretariat will reclaim, as appropriate, one or more of its unoccupied Chairs. Should all of the institution's Chairs be occupied, the secretariat will negotiate with the university on how best to reclaim the lost Chair(s).

Of course, the success of a programme like this one should be measured by the quality of the people who take up the chairs. (Italy has a similar programme already in place. You can read about it in a short article in Nature, with commentaries in the blog posts "The Runaway Brains" and "Brain Drain and Brain Gain".) You can look up the chairholders in all disciplines here. A quick browse through the names of the Canada Research chairholders in Information Technology and Mathematics makes me pretty sure that you'll find outstanding people in your area of interest.

Wouldn't it be great if we could convince our own ministries for education, university and research to set up a Research Chairs programme along the Canadian lines? Let's see what the new year will bring, but I do not hold my breath. I am already doing so waiting for the result of the pending research grant applications .

I wish a happy and productive 2008 to all readers of this blog.

Workshop on Women in TCS

The department of CS at Princeton University is hosting a "Women in Theory" student workshop in Princeton on June 14-18, 2008. See http://www.cs.princeton.edu/theory/index.php/Main/WIT08 for more details and list of confirmed speakers. (Via in theory.)

I am happy to see an initiative like this. In fact, I believe that we should have more events that highlight the achievements of women in TCS and that offer prospective students role models that they can look up to.

During my student days in Pisa, I thought that it was very natural for CS classes to be attended by roughly an equal number of men and women. I also followed a good number of classes where lecturers or TAs were female members of staff. Back then, I never thought that computer science was a male-dominated subject, and I had no reason to think so. Only much later, did I realize that what I thought was the norm was, in fact, an exception and, by the time I taught a class in Aalborg that was being attended by only one female student out of about 50 registered students, the lack of women in CS was not a surprise to me any more.

It is still my impression that Italy has a fairly substantial number of female (theoretical) computer scientists---at least compared to countries in Northern Europe. People often ask me for the reasons behind this phenomenon, and I am always at a loss to try and explain it.

Do any of you have a good explanation why countries like France and Italy seem to suffer less than others from the lack of women in subjects like CS? Could it be that things are getting worse there too?

Computer Scientist: 21st Century Renaissance Man

The period from January till May each year is when faculty at the School of Computer Science at Reykjavík University try to make a determined effort to entice students to study our lovely discipline. (I know, we should do so all the year round, but somehow our good intentions do not become good deeds on a regular basis by themselves :-()

As part of our 2008 campaign, I have coauthored a short essay entitled Computer Scientist: 21st Century Renaissance Man. There is nothing particularly new in it, but I hope that it is readable and that it carries the message that CS is much more than most laypeople believe it is. Maybe some of you will find it useful for your own PR campaigns. Feel free to use it, if you think it may help.

You can draw some more inspiration from the items in our suggested-reading list. More essays of general interest may be found here. See also the excellent survey collection at Theory Matters.

Positions in Computer Science and Applied Maths at Reykjavík University

Some readers of this blog might be interested in the following job announcement. We are particularly interested in applicants in Computer Security, System Dependability, and related areas within the field of computer science. Moreover, Software Engineering is intended in a broad sense and we welcome applications from people working in, e.g., formal development techniques, model-based software development, and testing and verification.

In order to implement its ambitious strategy in research and teaching, the School of Computer Science at Reykjavik University seeks to hire faculty members for new academic positions. The following links point to pages with more detailed information about the vacant positions.

Applied Mathematics: http://hr.is/?PageID=6595
Computer Science: http://hr.is/?PageID=6596
Software Engineering: http://hr.is/?PageID=6608

In all cases, position levels can range from assistant professor to full professor, depending on the qualifications of the applicant. Salary level is negotiable and relocation assistance is offered. The position is available immediately, but later starting dates can be negotiated.

Informal communication and discussions are encouraged, and interested candidates are welcome to contact the Dean of the School of Computer Science, Dr. Ari K. Jónsson (ari@ru.is), for further information.

Registration for ICALP 2008 is Open

The registration page for ICALP 2008 and affiliated events went live yesterday evening at http://www.ru.is/icalp08/registration.html.

We have done our best to keep the registration fees as competitive as we possibly could. The prices on the registration form are in ISK, but, by way of example, the regular early fee for a non-ICALP participant to a one-day workshop is around 77 euros, which become roughly 56 euros for somebody who registers also for ICALP. The early registration fee for ICALP is a little below 380 euros (including the excursion).

If you know that you will attend ICALP 2008, as you should :-), I strongly encourage you to book your flights and accommodation early. July is prime holiday time in Iceland, and you are more likely to get a good deal on your flights if you book as early as you possibly can.

Let me end, in the style of Numb3rs, by noting that ICALP 2008 is

13 workshops
8 days
5 invited speakers
3 tracks
2 prize awards
1 conference.

Thanks to my co-organizer Magnús M. Halldórsson for pointing out this "ICALP for fun" countdown and that this is going to be a Fibonacci-like ICALP!

Ranking of Excellent European Graduate Programmes in Natural Sciences

This report of the Centre for Higher Education Development (CHE), which was released about a week ago, may be of interest to readers of this blog. The CHE is a think tank for higher education. Based on international comparisons, they develop models for the modernization of higher education systems and institutions.

Their report develops a Ranking of Excellent European Graduate Programmes in Natural Sciences (viz. biology, chemistry, mathematics and physics), which is intended as an orientation guide for undergraduates, helping them find their way around European Higher Education while at the same time helping them to choose a suitable university for their graduate studies: Master’s and PhD.

At first sight, the report looks very well done, and for an Italian expatriate like me it is good to see that Italian institutions are doing rather well. I'd like to see a similar analysis carried out for programmes in computer science.

Let me try and provide a few remarks on the CHE report. In passing, I'll also offer some personal conclusions related to what the findings of this report may mean for a country like Iceland, which has great ambitions despite its tiny population.

Let me start by focusing on one message from the report that I find most important here (quoted from page 14 of the report). While reading the following text, bear in mind that "subject areas" refers to biology, chemistry, mathematics and physics, and not to a huge array of different disciplines.

"Another interesting finding is the fact that most institutions (33) are selected in only one subject area, 15 in two subject areas, 4 in three and also only 4 in all subject areas. If, even in the relatively closely connected academic fields of the natural sciences and mathematics, only 14% of the very top institutions in one geographic region are featuring three or all four subject areas, this can indeed be taken as an argument against institutionwide rankings."

Even though I enjoy reading the results of university-wide rankings, I believe that what should concern students choosing where to pursue their studies and funding agencies determining where to invest their research funds in specific disciplines is not the overall ranking of a university, but rather its excellence in the specific topic of interest. For this reason, I agree with the finding that subject-specific rankings are much more informative than institution-wide ones.

Of course, in general, a university that scores highly institution-wide won't have any very weak department. However, there may be, and indeed there are, universities that have peaks of true excellence in specific areas, even though they may not be world-beaters in many areas. If I were a prospective PhD student, I would prefer going to study in a department which is known to be top-class in the specific area of my interest rather than going to university X just because it has a globally good reputation. Quoting from the report:

"Prospective doctoral students are possibly less interested in the general performance of a faculty or department than in a specific research group. They usually have very clear ideas about the specialised topic on which they are focusing. Thus, it might be of some value for a student searching for a biology doctoral programme specialising in insects to know that the faculty at University A is excellent in its research output in this domain. However, it might be much more interesting for this individual to
learn that he could delve into honeybee studies in Würzburg's bee group. Or, a student in astrophysics might be attracted less by the overall performance of the Physics Department at the University of Copenhagen than by its research group focusing on dark matter and cosmology."

So my first conclusion is:

Conclusion 1. Our business as academic institutions is reputation. It is better to be known in a few selected areas than to be unknown in many. Icelandic universities should prioritize and place more resources in those areas where they can maintain or build a strong reputation internationally. The competition is growing stronger by the day; nobody stands still and we will need many more resources in the future just to maintain our present standing where we have one.

The second point that I'd like to pick out from the report is the minimum entry requirement for even entering the evaluation. The 3000 ISI publications from a institution over the evaluation period are indeed a very tall order for any Icelandic institution at this moment in time. Sometimes we pat ourselves on the shoulders and tell each other how well we are doing, and for very good reasons. However, we should never lose sight of the "big picture". A very good practice for any scientist is to remain humble, to know that there is a lot one does not know, and to keep in mind that there are very many strong scientists and departments out there.

As Socrates famously put it, "A wise man is one who knows he does not know." In this setting, I would translate this statement into something like this:

Conclusion 2. A wise rector/dean/head of department is one who knows that her university/faculty/department will need to improve its research quality and output considerably just to maintain its present status, no matter what its present strength is. The only way to do so is to hire the best possible researchers, to give them the best possible working environment and the freedom to follow their research interests. Research output will need to be considered when distributing research money to ensure that the most funding goes where the highest "interests" (read "quality publications in internationally recognized outlets") will be generated.

Two of the indicators considered by the CHE Ranking are

• the percentage of international and female staff within the group of staff with a doctorate and
• the percentage of female and international doctoral and master's students.

I am afraid that, despite our snow queens, we score badly on both of these fronts. Ranking measurements aside, it is of paramount importance for science in Iceland to nurture female talent and to seek actively to hire the best available female applicants. Mind you, I am against hiring female applicants just because of their gender. What I am saying is that our departments should have search committees who actively nurture connections with the best possible female applicants for positions and that outstanding female applicants should be given precedence when they are at least as good as the competition. Here the ministry could also chip in with some financial incentives to universities to hire outstanding female applicants. (I won't turn this into a conclusion though )

One may wonder whether some research groups from Icelandic universities can make it into the big league. The answer to this natural question that emerges from the CHE report is, I believe, positive. Look at the bottom of page 12 in the report. There you will read:

"Looking at table 2, the United Kingdom not only attains the largest number of gold medals but also the largest number of medals in total within the excellence group. Switzerland, with only three universities in this group, is in third place concerning gold medals and holds the largest relative percentage of gold medals: 16 out of 22 medals in the whole."

This is an outstanding, and not unexpected, performance of Swiss institutions. In fact, ETH Zurich is one of only four universities with gold medals in all of the subjects in the excellence group (the others being Imperial College, the University of Cambridge and the University of Utrecht)! How can Switzerland achieve this outstanding level of academic achievement? Rather than trying to answer this question myself, I will rely on higher authority and freely quote a few excerpts from an interview to the Italian mathematician Alfio Quarteroni (professor at the Ecole Federale Polytechnique de Lausanne and at the Politecnico di Milano) published in this book.

• Switzerland has only two federal universities (ETH and EFPL).
• These are two truly international institutions. To wit, about 70% of their professors are foreigners, and so are about 65% of the PhD students and about 33% of their undergraduates.
• Each of the few and carefully chosen full professors in those institutions has the financial resources to build her own research team. For instance, Quarteroni's team has about 20 members. (As a curiousity, they helped build Alinghi, the boat that has won the last two installments of the America's cup.)
• Quarteroni roughly says: "EFPL offers outstanding environmental and quality conditions that I have not found elsewhere. I have worked at the University of Minnesota, at Paris VI and, for shorter periods, in about 50 universities and research centres throughout the world, including NASA at Langley; well, on the basis of my personal experience, Lausanne is the place where I have been able to realize my goals in the simplest, fastest and most efficient way."

Conclusion 3. I let you draw your own conclusions as to what we need to do here in Iceland in order to approach the lofty heights that those Swiss institutions as well as several universities in Finland, The Netherlands, and Sweden have managed to attain. The above opinions of Quarteroni's raise many questions which I hope Icelandic university administrators will be willing to answer.

RU in the Guardian Education

Colin Stirling pointed out this article in the Guardian Education to me yesterday. This article is going to generate a little publicity for my current institution (Reykjavík University) and we can certainly do with that! It is also certainly true that Icelandic (academic) institutions feature more women in leading positions than elsewhere, and that their wages are comparable to those of equally qualified male colleagues. However, I am not so sure that Iceland is a particularly good example of a country that attracts good numbers of female students and members of staff in science and technology. We still have far too few women enrolling in computer science degrees, for instance, and my department employs one female professor and two female assistant professors. (Yes, we now have English web pages!) We will need to work very hard to try and change this situation, and this is one of our tasks for the future as far as recruiting is concerned both at student and staff level.

But enough grumping, let's enjoy our five minutes of fame in the British media, even though, as my rector Svafa Groenfeldt wrote to me,

"The interview was ok but as always the journalists take a bit of an artistic license when they quote what we said :-)"

A Cancellation Theorem for BCCSP

Wan Fokkink, Anna Ingolfsdottir and I have recently completed the paper A Cancellation Theorem for BCCSP, which is now available from the web page where Anna and I collect our papers.

The aim of this paper is to prove a cancellation result for the language BCCSP modulo all the classic semantics in van Glabbeek's linear time-branching time spectrum. The statement of this result is as follows.

Theorem. Let t and u be BCCSP terms that do not contain the variable x as a summand. Let <= be a preorder in van Glabbeek's spectrum. If t+x <= u+x then t <= u.

Apart from having some intrinsic interest, this cancellation result plays a crucial role in the study of the cover equations, in the sense of Fokkink and Nain, that characterize the studied semantics.

Fokkink and Nain proved the instance of the above theorem for failures semantics, with the aim to obtain an ω-completeness result for this semantics; their proof is rather delicate. To the best of our knowledge, failures semantics has so far been the only semantics in the spectrum for which the above result has been published. In our paper, we provide a proof of the above-mentioned property for all of the other semantics in the linear time-branching time spectrum. Despite the naturalness of the statement, which appears obvious, these proofs are far from trivial (at least for yours truly), and quite technical. I myself was really surprised by the amount of work we needed to do to prove such an "obvious" statement. (In case you wonder, the "obvious" statement is false in general. Consider, for instance, an algebra with carrier {0,1} and where the sum of any two elements is 1. Then the inequation y+x <= z+x holds, but y <= z obviously does not.)

I hope that some of the readers of this blog will find the paper worth reading. The techniques used in the proof of the cancellation theorem may also have some independent interest.

Accepted Papers at FOSSACS 2008

The list of accepted papers for FOSSACS 2008 is out. I was in the PC for the conference, and I have to say that this year the number of very good submissions was substantially higher than the number of slots. (And this without counting the papers with which I had to declare a conflict of interests.)

Looking at the accepted papers, it is striking how many of them have French authors. In fact, France was the country with the largest number of authors of submitted papers this year (67 to be precise) , and the acceptance ratio for papers co-authored by French authors was high. By way of comparison, the country that had the second largest number of authors was the US with 29. Germany, Italy and the UK were roughly on a par with the US.

French TCS is hot, judging by these figures. LSV alone contributes at least five papers to FOSSACS 2008. This is quite a way of celebrating their tenth anniversary.