The HEPAP University Grants Program Subpanel has just issued a report, concerning the “University Grants Program” in US HEP, that part of the DOE and NSF high energy physics budget which supports research based mainly at universities (as opposed to government laboratories such as Fermilab). Obviously this is the part of the HEP budget that is of most direct concern to university researchers, especially theorists, who receive most of their government funding this way (a small number of theorists are supported by national labs, not universities).
On the experimental side much of the report is concerned with how to manage what will happen over the next few years as many researchers move from working on experiments in the US to the LHC, in particular how to deal with the higher travel and living expenses this will require. I’ll concentrate here on some comments on the extensive parts of the report that deal with theoretical particle physics.
The report is surprisingly light on actual budget data, with few specific numbers about past budget trends, current budget levels or future budget plans. 2006 NSF university grant funding is given as $19 million for experimental particle physics, and $11.8 million for particle theory, astrophysics and cosmology. DOE university grant funding is described as about $110 million per year, with no breakdown between experiment and theory. The only historical data given is that this kind of DOE funding peaked in 1992, at a level of $150 million in current dollars, supporting a total of 1685 people back then, as opposed to 1495 in 2005. The main budgetary recommendation of the report is that 1 % of the total US HEP budget (about $8 million) be redirected to the university grants program as the SLAC and Fermilab collider programs wind down over the next few years.
The recommendations for theoretical particle physics mostly concern funding for graduate students, calling for increasing the number of graduate students in particle theory, especially students working on calculations directly relevant to LHC experiments:
Funding directed at university-based theoretical particle physics for the purpose of increasing the number of HEP-grant-supported graduate students should be given a higher priority in the overall HEP program. Support for students and postdocs doing calculations related to upcoming experiments is particularly urgent.
Though the universities are strong in formal theory, there has been a decline over the years in conventional particle theory (phenomenology), for a variety of reasons. Phenomenology embraces a number of different areas, including data analysis, collider physics, computational physics, perturbative QCD, lattice field theory, model building, flavor physics, and neutrinos; it overlaps with such areas as strings, astrophysics, and cosmology. All these areas are important; but those directly connected with the LHC are increasingly critical. The entire LHC experimental program requires a strong theoretical component involving calculating Standard Model backgrounds and new physics processes, together with interpreting the experimental results and teasing out their implications. However, the number of theorists working on such topics in the United States, especially at the universities, is inadequate. For example, there are only a handful of people in the U.S. working on computational physics, such as event generators. Many more will be needed to fully utilize the physics potential of the LHC. It is important that much of this effort be centered at universities because (a) much of the experimental analysis will be done at
universities, and (b) a university presence is needed to attract graduate students. A general concern is the overall decline in the agencies’ support of graduate students in theory, both formal theory and phenomenology. This decline makes it difficult to train a sufficient number of students. The problem is aggravated by increasing competition for the limited number of available teaching assistantships (TAs) from students in other subfields of physics.
A key component of a strong Terascale physics program (at the LHC and the ILC) is a strong theoretical program involving the calculation of Standard Model backgrounds and new physics processes, together with interpretation of the experimental results. However, as pointed out in this report, the number of theorists working on such topics in the United States, especially at the universities, is inadequate. Addressing this vital need requires an additional level of effort.
Overall, the field of high energy physics faces several critical manpower and infrastructure problems. Declining graduate student support affects the intake of new physicists and therefore the future of particle physics overall.
The report gives no actual numbers (which, presumably are available, since the DOE and NSF should have counts of how many students they support each year), but, based on responses from a survey of university grant PIs, it says that the number of grant-funded RAs for HEP grad students has been decreasing, with, especially for theorists, student support having to come from TAs:
The overwhelming response stressed that the level of (NSF and DOE) grant support for RAs for graduate students is insufficient and has indeed been declining over the last decade. At the same time, respondents noted that the cost of supporting a graduate student on a grant has increased, especially because of stricter university requirements regarding tuition remission and fringe benefits.
As a result, particle physics groups routinely rely on other sources of funds for all or part of their graduate student support. One major resource is TA positions for particle physics students, addressed in more detail below. Some respondents noted that their universities have limited fellowship support for some students. A handful also mentioned seeking outside support from other federal agencies. Many said that they had been forced to turn away qualified students due to a lack of grant support. Some also indicated that students had turned down the chance to join a particle physics research group because other departmental areas could promise steadier RA support, rather than a mixture of TA and RA support…
Survey respondents brought up several difficulties caused by this reliance on TA support. First, spending time as a TA slows senior students’ research progress (increasing their time to graduation) and hampers experimental students’ ability to travel to particle physics labs. Second, TA support is also currently a declining resource for particle physics at many institutions, because university administrations are providing less overall TA money to physics departments, departments are reducing the number of semesters any student may spend as a TA, or other physics subfields are requesting more TA slots. Third, if a particular research group (usually particle physics theory) makes unusually large demands on the available TA slots, this creates friction and resentment within the department as a whole.
The report has no mention at all of what the desirable level of particle theory Ph.D.s might be from a larger perspective. There is zero discussion of the relationship between how many such Ph.Ds are produced, and how many jobs doing particle theory research are likely to be available in the future. This is presumably because the authors are well aware that there remains a huge imbalance between the number of smart people getting Ph.D.s in this subject, and the number of opportunities for them to make a career in the subject. The reference to “friction and resentment within the department as a whole” over TAs makes clear what one of the main concerns driving this recommendation is. The amount of power and influence one’s group has in an academic department, including prospects for being allowed to hire more people, is heavily influenced by how much grant money one brings in, especially how much funding for the graduate program one can provide. This is made even more explicit at another point in the report:
Because of eroding support, more and more theoretical graduate students are being required to teach more and more of the time. This is unfortunate for at least two reasons: first, it lengthens the time to degree; and second, it signals to physics departments a hint of declining support for HEP research, exacerbating hiring worries.
While the report mentions the need to find some more money for postdocs (since those working on experiments will need to travel to Europe more often), it emphasizes support for grad students, not postdocs. It does this even though it is postdocs who are doing much of the most original work in the field, and the small number of postdocs (and junior faculty positions) is what makes career prospects for particle theory students extremely problematic. I can’t even really make sense of this one paragraph from the report that deals with this:
Given the choice between hiring more graduate students and taking on a postdoc, many faculty members will opt for the latter when faced with limited available funding. However, while this may seem be the best solution in terms of immediate research workload, the long-term negative effects of this choice on the field as a whole are clear.
One of the most interesting things in the report is the set of numbers from survey responses about how many grant-supported theory researcher are working in which areas, and what hiring plans by area are for the next 5 years. Figure 3 on page 43 divides theory researchers into six categories, and gives counts for how many are working in each category now, how many expected in 2012. The number of string theorists is supposed to drop from 103 to 84, “field theorists” from 91 to 77, “model builders” from 88 to 70, and “QCD/Lattice QCD” from 50 to 41. “Particle phenomenologists” are supposed to increase from 188 to 194, and “astrophysicists and cosmologists” from 136 to 176. Obviously boundaries of these fields are unclear, especially since string theory in recent years has to some extent moved away from formal theory, with more people describing themselves as “string cosmologists”, “string phenomenologists”, “string-inspired model-builders”, and much of the attention of the field devoted to trying to do QCD calculations with string theory.
If you take these numbers seriously, a grad student would be nuts to work on anything except cosmology or phenomenology, since all other subfields show about as many people leaving them as would be accounted for by retirements, so essentially no new hiring. My suspicion though is that these numbers reflect what departments say they would like to do, not what they will do. Most departments now say they want to hire in the areas of cosmology and phenomenology. But faced with the fact that competition for the best people in those areas is tough, and finding it much easier to get good people in other subfields, I suspect there will continue to be quite a lot of hiring in these other subfields, in string theory especially, which seems to be what looking at the latest data from the Rumor Mill shows.
Given the huge and supposedly increasing dominance of these numbers by the “particle phenomenology” category, the report’s call for an urgent increase in funding to produce more phenomenologists is not so easy to understand. However, the authors make clear that what they want to see is more of a very specific sort of phenomenology: people working on things like event generators to precisely calculate standard model backgrounds for the LHC experiments, something which has always been more popular in the Europe than in the US. The recent NSF-funded “LHC Theory Initiative” is specifically designed to address this, and is being promoted with nationalistic calls to fight the “outsourcing” of these calculations to European theorists. The authors of the report are calling for more Ph.D.s in conventional, experiment-based phenomenology, not more Ph.D.s in string theory or “string phenomenology”. They choose to do so, for obvious political reasons, not by calling for a reallocation of resources within the field of particle theory, but for new resources for this purpose, even if it will lead to a smaller fraction of particle theory Ph.D.s being able to find jobs doing the research they have been trained to do.
Can the creation of event generators really be considered “theory”? You get a Lagrangian from a model builder, plug it into an automated Feynman diagram package, obtain matrix elements ready to plug into a standard procedure skeleton, do a little debugging and testing, and presto, there’s your new event generator. Is it “theory” just because you don’t get to pull cables?
For example, there are only a handful of people in the U.S. working on computational physics, such as event generators.
I’m not sure I understand this sentence. “Computational physics” means computer simulations of physics processes? What is an “event generator”?
Coin,
An “event generator” uses Monte-Carlo methods to, given a model (e.g., the Standard Model), generate “events” with statistically the expected properties, where the “events” correspond to, say, events at a collider. The way you test your model is to run these events through a further simulation of what they will do in your detector, then compare the Monte-Carlo results to your data.
For perhaps the most well-known of these, see
http://www.thep.lu.se/~torbjorn/Pythia.html
AGeek seems to think that the development of this kind of software is not “theory”, but I think many people would disagree, including some who would claim that AGeek’s attitude is what’s wrong with particle theory these days…
So what justifies the huge increase in cosmology positions over the next 5 years? Is there a perception it has been neglected recently?
I confess: I don’t think that the development of event generators is theory. It’s programming, and pretty elementary programming too, to the point that I don’t see why it could not be completely automated, given an existing simulation framework.
Surely an unwillingness to label trivial technical work “theory” is not the worst ailment afflicting particle physics these days? Some might say that the opposite is true…
More fundings for graduate students? Is this really gonna happen?
“The Role of Computation in Physics” By Bradley A. Shadwick in the January 2007 issue of Computing in Science & Engineering clearly and intelligently explains how computation fits into theory and what its limitations are.
I would never consider computation or simulation to be physics theory per se (though Stephen Wolfram it would appear has a different, yet fundamentally misguided view), but instead it is a potentially potent tool, which when wielded correctly, can do what all tools do – amplify our own feeble abilities, which in this case means help us solve equations.
Peter said:
For all of your insight regarding the health of the field, you sometimes seem naive with regards to politics. Money talks and you can’t seem to see that you have won…..or maybe it was Smolin that won….either way, be gracious in victory.
IMHO,
I’d be curious to hear what you actually think about the politics. Whether or not my views on it are naive, I’m always interested in maybe learning something.
I definitely noticed that at least this panel is not calling for more funding for string theory, but more funding for phenomenology and a redirection of most hiring to cosmology don’t seem to me to be the answers to what ails particle physics. Seeing that already fewer theorists describe themselves as “field theorists” than “string theorists”, and physics departments plan on reducing the number of field theorists further is rather discouraging…
But faced with the fact that competition for the best people in those areas is tough, and finding it much easier to get good people in other subfields…
Not necessarily. For instance, our department is (almost religiously) trying to hire a biophysicist. For several years. No success so far — and it is understandable — we don’t have any yet and we are not Harvard (although we do have a decent Medical School). So it really depends on the way a particular department is run and on the vision of its administration/administration of the College, etc. Usually, however, it does not work to the advantage of HEP…
And even people who used to be particle physicists, but now are astrophysicists (and there are many) do not help — their move from HEP to astro is viewed by other faculty as a sign that HEP is dying out and there is no point in investing in it by hiring more faculty.
this is a reply to AGeek, who thinks that people who work on event generators are not really particle theorists.
Of course this claim depends on one’s definition of theorist. But let me point out that constructing event generators is much more than lagrangian + programming. There’s a whole jungle of physics between the raw experimental data and inferring terms in a lagrangian, and experience shows that you need bona fide particle theorists to do the job. Experimentalists can’t do it, string theorists can’t do it. And hired programmers certainly can’t do it. I appreciate this isn’t obvious, but that’s the way it is.
Dear phenomenologist, it is certainly true that “There’s a whole jungle of physics between the raw experimental data and inferring terms in a lagrangian”, but that is not what building event generators is about. As I wrote, “You get a Lagrangian from a model builder”. What you are describing is the job done by the model builder. Event generators are derivative work used to extract predictions from a Lagrangian provided by a model builder in a form directly comparable to experimental data: the information flow is the reverse of what you describe. (You may be thinking of Marmoset; that’s something much more ambitious than I have in mind.)
In principle, it is of course possible that somebody building an event generator also constructed the underlying model. Such a person would be model builder with a handy technical skill, comparable to being unusually good at doing integrals.
AGeek,
You’re misinterpreting “phenomenologist”, who was just pointing out to you that the relation between terms in a Lagrangian and experimental data is quite complicated, involves a lot of physics, and this is what event generators are all about. You seem to think that writing down terms in a lagrangian is non-trivial physics, whereas designing an event generator that allows you to compare these to experiment is “a handy technical skill, comparable to being unusually good at doing integrals”. An equally good case could be made that writing down terms in a Lagrangian is “a handy technical skill”, with the non-trivial physics requiring real knowledge and sophistication being the connection to experiment. Something leads me to suspect that you’ve written down terms in a Lagrangian, but have never written an event generator…
Peter Woit,
you are misinterpreting me, who am simply pointing out that the trivial technical task of building event generators does not involve sorting out the relation between terms in a Lagrangian and experimental data.
Writing down any terms in a Lagrangian is indeed fairly trivial. The generation of all terms compatible with a given set of symmetries and other conditions can also be automated. Writing down terms relevant to our world without performing an exhaustive search of the entire theory space, now that’s something which sets a Weinberg apart from a marmoset.
BTW, your suspicion is incorrect. I have done both. It may interest you that I did the event generator thing before having ever laid eyes on a QFT textbook, simply by being a good geek and having a superficial, undergrad level understanding of particle physics. Unfortunately, it quickly became clear to me that the people doing it for a living did not have a knowledge of physics significantly beyond mine either. They had stopped learning and growing once they had found this little niche where they could provide a service in demand.
AGeek,
Apologies for my incorrect suspicions.
My own experience though is that the phenomenon of people who have “stopped learning and growing once they had found this little niche where they could provide a service in demand” applies just as well in other much more abstract parts of the field.
>It may interest you that I did the event generator thing before >having ever laid eyes on a QFT textbook, simply by being a good >geek and having a superficial, undergrad level understanding of >particle physics. Unfortunately, it quickly became clear to me that >the people doing it for a living did not have a knowledge of >physics significantly beyond mine either.
This is a shockingly arrogant (and narrow-minded) statement, if by `event generator’ you mean what is normally meant by an `event generator’, that is pythia, herwig, etc. Do a SPIRES search on people who do write event generators. events radiate in the initial and final states, jets are produced, partons shower to create more jets, hadrons decay. how many jets should you expect? how many hard? how many soft? how much missing energy how often? none of this is at all obvious from the Lagrangian.
That said, the kind of event generator someone could write without having done any QFT was indeed probably not a great achievement in particle theory.
Anyway, what next? those people who do multi-loop computations don’t do particle theory either, they just have a handy technical skill in doing integrals? and the cosmologists who do N-body simulations, maybe they should be reclassed as computer support officers, after all the real physics was all done by Einstein.
on the other hand: writing down a half-arsed model whose only virtue is that it hasn’t been ruled out yet; that, now that takes a *real* physicist.
Dear “string theorist”, I think we are talking past each other.
First of all, how to handle parton showers, jets, hadron decays: it is perfectly true that “none of this is at all obvious from the Lagrangian”. Because, as you surely know, this all happens at strong coupling. So it’s handled by a phenomenological model motivated by, but not derived from, first principles. Essentially it’s a messy piece of code which parametrizes experimental data.
If you want to call this an “event generator”, then your statement that “the kind of event generator someone could write without having done any QFT was indeed probably not a great achievement in particle theory” is rather questionable. Because, as you point out, “none of this is at all obvious from the Lagrangian”, and little knowledge of QCD beyond an intuitive understanding of gluon strings stretching and breaking up goes into the code handling it.
Second, this hadronization stage is not something which you have to worry about every time you create an event generator. It’s encapsulated in the simulation framework which handles all the output from your event generators, by which I mean procedures working with elementary particles according to matrix elements derived from a Lagrangian. The typical job of creating an event generator consists in writing such a procedure, which takes a list of elementary particles as input and produces a list of elementary particles as output, and linking it into the framework.
I heard this same debate some years ago, when a young physicist expert of event generators sent a public letter to his influential collaborators writing something like: “my advisor warned me that developing event generators is not considered theory, so in order to get a position I need to stop our collaboration and move to something else.”
I’ve heard a number of people argue that studying high energy physics is a more reasonable career choice now because of the opportunities for employment created by the LHC. It seems to me, after reading the report and this post, that this isn’t true. The number of new permanent jobs created isn’t going to be significant compared to the number of graduates that are going to be produced.
In other words, rather than making the profession more healthy, the LHC may just give graduate students a more comfortable ride to nowhere.
In what way am I being too cynical?