Interview with Eric Poisson

0:00 Okay, so it's the 17th of August at 5.30 and I'm speaking with Eric Poisson. So, let's see, as I said earlier today that I'm interested in data analysis issues and that. Is that something that you're working on at the moment? I'm not sure. I guess I'm working a little bit on that, but I don't feel I'm where the action really is. I feel a little bit in the fridge of that particular field. Why is that just... Well, because I never really took the plunge and started playing with the real data. I was not letting myself, or I didn't involve myself in the ACES project, and I didn't, I guess I was preferring to do other things. I tried to worry about issues that maybe are not big issues, but things that happened to interest me. Like, you know, this recent stuff about the validity of the stationary phase approximation to calculate for your land spawns and things like that. But I guess I'm keeping track of what's going on and I should pay some attention to it. I don't feel that I'm really in the middle of it. Yeah. And was there any particular reason why I didn't want to be involved with Asus? Well, it's not that I didn't really want to be involved, but I guess I have a fairly... maybe a lack of ambition or something like that. I felt more I wanted to stay in my corner and do what I really wanted to do without being bothered with what other people wanted to get done. And I never felt the great impulse to play with the real data and do code and do, you know, participate in the grasp and all of that, it never really interested me that much. I would prefer to worry about the sort of fundamental issues like the convergence of and things like that. I have more fun with that than the real data. So when Bruce really got started with his group and all that, I wasn't really tempted to join. So partly it was the nature of the problem, as you say, you prefer to work on more fundamental

2:30 and also a little bit that working in something that's a, well that's something like a big collaboration. Well that wouldn't bother me but I guess I felt a little bit like I was isolated and so I'm pretty sure if I had been at Caltech at that time I would have probably got involved and I would have probably enjoyed it and so on. But feeling that I was not there and feeling that I was not, you know, I was isolated from that group, I felt that it was easier to try. That it would be hard to be working in parallel to them but, you know, sort of not being physically near anything. Although my postdoc was involved in that, so he actually did specifically work for the 40 meter beta project. One issue was to try to understand how well the 2PM templates were being used for searching in that data, how well they would do. And that's something that I would consider more fundamental issues and in fact it's something that Serge, my post-doc and I have been doing. But I was thinking in terms of initial LIGO and advanced LIGO, but they said, well, you could do it for the 40 meter and then tell us what the answer is and Serge did that on And because of that, his contribution was in the paper, so he's one of the authors. So in a sense I was indirectly involved in that, but basically it was a repetition on the previous calculation that we had done for the advanced lego and the initial lego, and I didn't feel that I needed to do that again. For me it wasn't anything new, it was just a repeat of a previous calculation. I didn't get involved even though I guess I could have. And is there any significance to you, you know, to you being in Canada as opposed to being in the States? Does that make any difference for you? I think that's why I refer to my lack of ambition. I think if I really wanted to be part of that effort, I think I could. I think there's no reason why I couldn't. I could be part of the phone conversations, the phone conferences. to Pasadena or Milwaukee on a regular basis and I could do all of those things. But I guess I'm a bit lazy and I'd rather stay home and work on what really interests me

5:00 because again that's not something that I would consider. I mean it's something extremely useful and something that needs to be done and I guess it needs to be part of the effort that will develop the tools to search for real signals when the detectors go online. I mean all of that is exciting in a way, but in terms of the actual work and in terms of the actual calculations that you would do or other projects that you would be working on, I think I can find things that are more interesting to me than those things. So what are the things that are interesting right now? Well, to me, I mean, I like to worry about gravitational waves and sources and things like that. So one recent thing that I did with a student was to look at the possibility that you would have eccentric binary systems emitting waves and the possibility that you would not go search for them directly because of practical limitations having to do with the number of templates and so on. But suppose that, you know, we find from astrophysicists sources out there, you have to make a choice. You go search for them explicitly, like everything else, which means that you have to consider an additional parameter to search for in the templates and that increases the number of templates and the computational aspect, the computational burden, and so on. Or you say that, well, we're going to just not search for them explicitly and hope that the current bank of templates, which are based on circular signals will be good enough to pick them up too, and what I wanted to do is to just estimate the ability of the circular templates to search for eccentric binaries. So you're not talking about optimal processing anymore, but you're talking about filters that are approximate at what could still do fairly adequately for eccentric pilots. So that's one thing that I did. But also right now I'm having fun with things like waves interacting with black holes and with propagation in terms of space time. And this issue of, and now it's really academic, you know, science in a way.

7:30 I mean, it's asking, we know that a black hole in asymptotically flat space-time will, how should I say this? So if you have a pulse of radiation interacting with a black hole in flat space-time, that the late-time behavior of the radiation is something very characteristic that corresponds to inverse power law decay, so the field will approach zero at late times as an inverse power of time. And the question would be, and we know now that because there's been quite a lot of calculations and we have quite a lot of insight into that result, that the strong field aspects of the spacetime is irrelevant to that power law decay. It's structure of the space-time far away from the central object that dictates what happens to the field at very late times. So when you hear something like that, when I heard something like that, I thought, well, okay, let's change the conditions far away and see if you can see the signature of the alternative conditions there. And the idea was to now let the black hole not be in complete isolation, but just put the black hole in a cosmology of some kind and see what the results are. So one simple modification that you can do is to look at a black hole in the Sitter universe, in the Sitter space that corresponds to an exponentially inflating universe. Or you can put the black hole in some other type of cosmology and look at how fields will behave. And I have a student who's been looking at that in the context where you have a black hole in a FRW, spatially flat, dust-filled cosmology and try to see where the field becomes aware of the different conditions and start deviating from the expected behavior based on the assumption that a black hole is in isolation. So I find those issues interesting, but they're issues of principle that have nothing to do with detection of anything, it's really a what-if kind of question. So I've been doing work along those lines. Well, we were discussing how one knows what it is that one does in science.

10:00 So this is kind of an interesting question which maybe is ancillary to that. How does one know in science, where does the dividing line fall between the issues of principle, as you say are fundamental issues and those that are related to detection because obviously that's certainly something that does vary a lot over time and do you find that there's a clear distinction when you're thinking about working on a problem that there's a clear distinction in your mind between well I'm motivated by this because it's relevant to problem detection or I'm motivated by it for more fundamental reasons or is it more organic? I mean, I don't think there's really a clear distinction between the fundamental and something that would have a direct application. I think, no, I guess in my mind I would have to, yeah, I might classify a project as being more on the fundamental side or more on the useful for detection side, but I can certainly conceive of a project which I couldn't really separate, I mean that would fall in a crack so to speak. And I guess I don't really operate that way that I am looking for a project on one or the other side. I tend to look for interesting things to do. I try to come up myself and for students. Anything that's really interesting and hopefully doable, I pick because it's hard to find good interesting things to do that also is doable. So the aspect that you want something interesting that can be done and when you're thinking in terms of a student, you want the thing to be done within a time frame of two years, and when you put all of those constraints into the selection of the topic, well, it's, you know, it's a little bit of that. Right. Right. So do you, you spent time consciously looking for problems that are suitable for students or is it? Yeah, especially when I know that I have a student coming in in a few months, I'm trying to think of what could be done, what could be do, and what could be interesting.

12:30 Sometimes it seems obvious, well, I should do that, and sometimes, depending on when I think about these things, I might have lots of ideas or no ideas. Of course, I tend to remember the ideas that I had a long time ago or recently, but of course. And then you start ranking the ideas, so this might be more interesting, but it's loosely and it might take too long or things like that. Sometimes you want a totally straightforward problem for a first problem for a student that you know how we'll be able to predict where it goes. But that doesn't tend to give you something very interesting because if you know where it's going to go, chances are it's not going to be very interesting in the end or you won't write that much in the end. Do you find that students respond more to one type of problem than another for instance they respond more to fundamental problems or more to more application problems? I don't know, I think I haven't had too many students so it's not too, you know, I would guess it depends a lot on the student and it would depend also on the problem. So where does radiation reaction force problems fall from your point of view? That's the sort of problem that I like because I would consider that to be at a fundamental level. because it forces you to address all sorts of conceptual issues and it's a really fundamental subject type of problem. But you know that down the line, if you can resolve all of the conceptual aspects and solve the problem, it's also going to be a very practical, you know, practical tool because, you know, we know that, you know, if we want to understand sources, for Lisa, for example, where you're talking about a very strong field, that strong field radiation reaction is going to be essential to the construction of signals and it's something that we need to be able to do eventually. So I like this sort of thing that a very fundamental problem will hopefully very soon have real practical aspects to it.

15:00 And are the practical aspects, do you think, maybe you have an opinion here too, are the practical aspects responsible for motivating, for instance, this meeting? So I guess last year at Camper Ranch you guys got together. Why decide at that point to really muster forces to attack this problem? Well, I think the fact that there's a need for an understanding of radiation reaction now because of the future experiments or the possibility of future experiments, I think that's always good to motivate a lot of work. But I'm sure for a lot of us, was it really the prime motivation or just further motivation, but we really want to do this because it really sounds interesting from the fundamental point of view. What do you think? I entered this subject mainly because Kip said that his grants being received for gravitational wave research, I will have to devote the case 70% of my time to gravitational waves, because that's what the money is for. But of course, I don't have a subject and I'm going into it more and more the more I work about on that. But for that restriction from Kate, I might have continued doing research on Black Homes and the structure of Black Homes. But I think it's a good thing that I had this restriction I am enthusiastic about it, I think it is, as Eric was saying, it is both fundamental physics and things which might be applied in the near future. But you had exposure to the radiation reaction force concepts of course, working with that Yes, of course. Being in Israel in the neighborhood of Amos, I did get to hear about his approach to that.

17:30 So I was aware through his 1995 physics letters paper and then the expanded versions and the physical review. But I wasn't doing anything myself until I arrived to protect. So in that sense, from your point of view, there was, presumably there was a direct connection to the problem of detection since Kip's grant said. So that's a very, maybe it doesn't sound very nice, but the prime motivation was just that that's what the money is for. And did Kip emphasize to you the application to Lisa or to...? No, he was saying that I would have to devote 70% of my time to gravitational waves, but in the broadest sense I could do anything I'd like with anything I'd like, but with some relevance to gravitational waves, so a scalar field has some relevance, so I'm working with scalar fields. Well, it's interesting, being at this conference, because I just heard quite a bit about Kaplan Ranch, and it's also interesting to see people planning further meetings. So it's like the beginning of a little series. So I guess the question is how many meetings will be necessary and how long will it take? I think it's going to be a long journey, because it's such a difficult problem in a way. I mean, it might prove to be less difficult than we thought, and then progress is going to be faster, but just the technical aspects are difficult. Yeah, it does seem to be. Well, when I started in Kip's group, he set Harris and I the task of looking at the Gauss of formalism for doing just this problem, and after a few months we decided, well, this formalism isn't very good and we don't know how to do this problem, so we went to Eric and suggested that he teach us how to use the Tkalski formalism to do the simpler case in Schwarzschild, so we decided we'd simplify, yeah, so, and here we are almost a decade

20:00 Even last year, a few people were working on this, but not very vigorously. What I notice now is that quite a lot of people are working on this more vigorously. When you have that sort of thing going, it would be surprising if next year there wouldn't be a lot of progress. I think so. over the course of this decade, it wasn't like people were working that much, it's only in the past few years that a few proposals for formulas were made, and as you say, in the last year people have ever been seriously working on that. And so I remember being at at the last meeting of the Backhole Grand Challenge Alliance and KIP got all of the numerical people there to sign up to a bet about whether they would have waveforms from binary back holes before LIGO had seen some questions. But anyway, so I was wondering if it would be a good bet for people working on the radiation reaction force to say, in the perturbation case there, in the extreme mass case, to say that it would be done before at least there you have more time. Everybody would bet. You would have to take that bet. Yeah, I figured Kip wouldn't take the bet. So, the, well, listening to Sathya's talk about the Pade Approximate, which is something I wanted to do recently, trying to, you know, be asking people about, I guess it strikes me, and I might wonder if this is a fair impression that in the realm of data analysis, people moved away more from the idea of optimal filtering and we're going to have a bank of near-perfect templates, which I knew were just going to try to hit one with whatever signal comes

22:30 through, and decided that actually the templates aren't going to be quite as good as might have been hoped that 3 or 3 postentonian templates might not be quite as good as have been hoped and that there aren't going to be very close templates available and so it's better to go for more robust approximation methods and that kind of thing. Is that an accurate impression? It's also a vague impression but anyway. Yeah, well I think it's true that the postentonian series has proven to be disappointing. really doesn't seem to converge to the right answer very quickly. By the way, every time you said a word converge, everybody each time said he didn't. The word convergence is such a word that you should never use either. But I think the pade version of that really shows a lot in front of us. So I think it could well be that a Padre version of a three piano waveform would just be as good as it needed to be. As far as finding other ways of doing it, more robust methods of approximation, I haven't heard a good idea about that since we've been talking about the need for other ideas. Yeah, but it doesn't seem like we're going to be able to do without the post-material waveforms. It's just that they're... So at some level they're... So would you say that people are looking for better methods? Well, I guess that's fair to say, it's just no one's thought of the method to do the competition. I guess there had been, there were some proposals like in the last three minutes paper about some kind of expansion scheme in the radiation reaction timescale.

25:00 but I don't think anyone has really pushed after me. I tried to implement that idea. I think a big problem, and what worries me, I guess, with the post-Antonian stuff is that all of the methods, like Pade, where you hope you're going to be doing better by being clever, those ideas can't be tested very well. And the only reason why they can be tested is the test mass limit, where you can rely on perturbation theory. And I'm certainly willing to believe that the conclusions you would draw from the test mass limit would apply to the equal mass case or the comparable mass case. I would be willing to believe that if it works there, it's going to work here too, but I'm not willing to put a lot of money on that. And if the successful detection of the waves depends a lot on those tricks, and you haven't been able to test that very well, I'm a bit worried. Right. So, well, you can certainly imagine that, um, my impression is, and here again, I'd be interested to know what your view is, my impression is that, I've been told by some people, I guess, that, you know, that there are experimentalists who knowledgeable, particularly knowledgeable about data analysis issues at this time, who aren't too keen on the, just say, just the idea of, optimal filtering against the bank of postentonian templates, so we can certainly imagine that they're not going to be... Yeah, but what else do they want to do? Sure. Well, yeah, I agree, it's not like they have any, but I'm saying in the context that you're talking about where the question of detection actually rests on sophisticated signal processing to get the signal out, then we can certainly imagine that there would then be a vigorous debate within the LIGO community as to whether you can reliably claim the attention on that. Yeah, I think it's going to be very interesting to see what happens when LIGO starts picking

27:30 because chances are that the only signal you're going to get are going to be a threshold, a detection threshold. And if to get to that level you have to play all sorts of tricks and so on, of course it's going to be difficult to sell the detection. I think if we're very very lucky you would get something with hundreds of, you know, signal to noise ratio of hundreds, then presumably you could do data analysis using different techniques and you would get something comparable. Yeah, I think it's going to be a problem, but I think it's probably not a, it's not a unique problem. I'm sure that every single field of physics that we're working with detection threshold had to face exactly that problem. It would be interesting to look for cases in comparison to see what would happen. And of course this field has had a particularly unhappy history with phenomena of special detection. Yeah. So there may be certain sensitivity in that part. Happy voice? Yeah. So does that kind of detailed notion of what's going to happen down the line inform what you're thinking about data analysis of, or is it more, you're just looking for the best ways that you can make use of what you know theoretically as it needs to? Yeah, I mean, that's the sort of thing that I've been thinking about, I guess, the most. I mean, what can you do with a post-entonant series, and what can you say about it? And that's why I felt that perturbation theory was really, really useful, because it's the only place where you have an exact answer you can compare everything with. And I think that this business of comparing the post-entonant series to the numerically generated answer, which is exact, because the numerical errors really are not that important here. I think we find a lot from that. And still, I'm convinced that we've done everything that we could at trying to, you know, re-express the motion

30:00 to the series in the best possible way. I think Paddy was a great, you know, great thing to try is really the best thing, you know, because there are other things that you can do, other games that you can play, using information that you have about the Post-20 series to try to produce something that would be even better than what I did. Maybe pilot is the best, maybe there are other things, I don't know much about these things. But then, and I think you can do all of that by comparing going always back to perturbation theory, but then the ultimate leap that you have to make then and to say that whatever worked for perturbation theory will work also in a comparable mass case, that's a leap that, well, that cannot be justified. I mean, it's really a leap of faith to import the conclusions that you gathered from perturbation theory into the comparable mass regime and maybe it's possible that the extrapolation will work but i think there's a leap of faith there and whether you want to to put a lot of money or a lot of confidence into that extrapolation or whether you'd rather do without i think that's a question that maybe someone will have to face at some point I don't know what the answer is going to be. Yeah, sure, so the problem being that because you have approximations where you don't know what the answer is, you have to look at a limited, you know, special case where you have some answer. It's looking at their lamppost. We have one lamppost. And that's all we can say. Right. But then you can't reliably, you know, the fact that you know the answer in one case, you can't reliably... It doesn't guarantee that it's always going to work, although you would hope that nature is kind and things will work out too. So is that something that you're planning to work on, looking at better ways of expressing the pleasant times of expansion? Yeah, I mean, I constantly think that I don't know what to do about it, and it's true.

32:30 I mean, sometimes I try something and it just doesn't do any better, or working. There was one time, I think it was last year, I was trying something else, and there are all sorts of techniques to accelerate convergence of series, and if you know that a series is convergent, like an original Taylor series is convergent, which we don't, but if you do know that it's convergent, then there's a transformation that allows you to accelerate the degree of convergence, and I think it's called the Shanks transformation or something like that which is basically group terms differently and generate a new series and that new series will converge better. So for the same degree of accuracy while previously you needed 10 terms of the original series by doing discrete expression you're only going to need 3 terms or something like that. So I just tried that as a game. And if I looked at the new series containing, I don't know, two terms, it didn't work at all. It didn't work better. Three terms, it didn't work better. But then at four terms, it gave the exact answer. Five terms, it didn't work any better. It was considerably worse than that. So by accident, n equals four seemed to give the exact answer, but any place else gave disastrous answers. But But for this one particular case, I did better than Pade, better than anything else. But of course, it's not useful because you have no confidence that this is something that's going to work in all cases. But maybe there's some other technique out there that will work. I'm not convinced, I don't want to find it, but I'm not convinced that all has been said. I'm curious, yeah, obviously there's a great, as you say, there's a great deal of pragmatism to the kind of approaches that we've been discussing, as you say, looking at the lamppost of the perturbation case or, you know, trying a convergence technique, even though you don't necessarily know that the pre-passion is convergent in the first place. So I was curious, I mentioned that, you know, I've heard people say that some experiment analysts are unhappy about the theoretical approach, but of course it used to be in the gravitational wave problem

35:00 that you'd have lots of more mathematical relativists who would be always attacking the approach of, as it was in the way I'd say, more pragmatic physicists on the radiation reaction problems and so on. So do you find that with the people interested in diagnosis and so on that you have differences in schools of thought? Do you have people who are saying, well, we ought to be much more rigorous about this? Or is there a debate on that level? Well, the feeling I have now, and it's sort of remarkable right there that you have about currently three people, two. It used to be five. But now there's fewer people now working on Wave Generation than ever before in the history of the few. What I'm saying is that it used to be that Sigoura Moore and Le Blasche were working on Wave Generation with Ba Ayer and that was the first three of the five and then it was Cliff Will and Adam Wiseman. And they, those files produced waveforms to 2.5p in order and we all agree that it's not enough, they have to push higher. Who does that? Lerg can put two people. And it seems that the entire template generation business is realized on those two individuals. it's only two people. Now I think that what's nice about that given that it's only two people is that they're working independently with different formalisms and that was true before too when those five got together as you know they did the first three and the last two were doing the calculation independently and they all agreed in the end which was very good. I think it's likely to happen again because I think both teams, both teams, Lipp and Cliff, are both convinced that they have a reliable formalism and that things are rigorous enough and things are under control that they can actually calculate to high order. I think that's probably true. And I think it's likely that they will generate answers that will agree with each other and I think that's that's good yeah but whether much but is it is it

37:30 sufficient that only two people should do these calculations or should there be like 10 or 20 people doing these calculations it's not clear to me is it sufficient to have two formalisms There should be maybe more formalisms. Why do you think there are only the two at this point? Because it's hard. Because it's horrendous. So you wouldn't do it? No, I wouldn't do it. I don't think I would want to do it. It's just too long and too hard. and I've never done it, so I would have to write it all, and by the time I get up to speak to where it worked five years ago, I'd say, you've done it now. I was wondering, in fact, if that ties into what I was mentioning, because I gather that, say, Bob Beiner and Jim Anderson, who are people who do have the work of all these kind calculations and have independent calculations to do, could do independent calculations, I think they're reluctant to do so because they don't necessarily believe that the series is reliable, pushed to be kind of more reliable. Well things have changed. I think part of the problem also is that it's all very opaque and it's hard to understand what the formalism is. I think the great thing about the Will and Wiseman approach is that I think you can actually read their paper and understand what the basic approach is, whereas the whole series of papers by Des Moores and Blancheres is really, really opaque. Are they going? Oh yeah, do you want it? Look up at these guys, we can... Oh yeah, I think they're... Let me just, I guess we can continue later. Sure. I'll just check. Yeah, pop that and see. So you were saying that it's worked out well, working on gravitational wave stuff.

40:00 You're enjoying it? Yes, I enjoyed it. What I was saying about the Ness Angelus being the kinds of gravitational waves that was just it was half jokingly but it was still true. I just had to do something related to gravitational waves. Of course I chose that particular subfield not a different subfield. That's what I was thinking was the a most interesting for me at that particular time and the second very needed work to be done so yeah I do find it very interesting and hope that we'll be able to do some more work about it you seem to be making good progress at your talk today there is some progress, yes It's interesting to see real calculations based on a radiation reaction, force calculation. It seems like a big advantage really because, as I mentioned, I started as a graduate student looking at the Gauss of a formalism. It struck me immediately about this paper before I decided later that I didn't think it was a very good formalism for calculating. Well, I think answer performance seems problematic because it is not causal. Sure. Yeah. But what struck me immediately was that he had, you know, done a couple of toy calculations at the end of the paper on trivial cases. So, it was apparent that he wasn't too keen to do any actual calculations of performance himself. And up to only very recently, what you had was a number of formalisms that had been put forward. And in the 17 years, I think, that since then, I don't think he has done much more than that. Yeah, and there have been formalisms before. So, are people heading off? Well, they are, but Pat really wants to keep the group small, so Alan is waiting for me. Okay. Yeah. Okay. What were we saying? Oh, well, Lior and I were just talking about, I was saying it was interesting what Lior's

42:30 talk, that it was, he was doing some real calculations, as it were, for the radiation reaction calculations. So it was interesting because for a few years there you had a lot of activity but people were putting forward various schemes. Yeah, that's right. Yeah, it's also probably as good that some people are actually doing real calculations. Well, I think people are also doing calculations. It's just that the one I was presenting today is not limited to slow motion or strong field. So, in that sense it is... Well, that's what I mean by real calculations in a strong field regime. I know I've been thinking about calculations. I haven't even done anything here. Yeah, it's interesting that... I think that's when things start to happen, when you just do it, and along the way you recognize that perhaps it's more difficult than you thought, but at least you're doing something. Yeah, that's why it seems to be… What you were thinking about Bob Wagener and Jim Anderson. Yeah, so, well I remember Kip telling me that he was very keen for Jim Anderson to to push his approach, which is also, I think, you were saying, of course, that, yeah, you were saying that the Wise and Will approach was, that their papers were a little clearer as a little bit. Yeah, much clearer. I think you can gather from their paper that the approach is sound, and the only thing you gather really from the Warren Blanchet is that they've done a lot of work, but it's, I mean, unless you're willing to spend all the time required to go through there, there's stuff that's not as immediate. But I agree with you, I agree with you too, that if Jim Henderson decided to push his compilation, that would be very useful because, again, it would be a totally independent way of doing it. But the problem is that he's one person, and he's probably not ready to put in the effort required to actually do that calculation. Yeah, I think so. And the way he put it to me was that he didn't feel that he wasn't happy. He didn't trust. Yeah, that's what bugs me about him is that he was very cranky about everybody else doing the calculation. And he says, if I did it, it would be better, and because I didn't do it, everybody else was wrong. I mean, unless you're willing to substantiate your doubts, I think you should

45:00 give everybody else the benefit of the doubt, but the answer to come up with at the end is fine. You can still not be happy with the way they did it, but if you can't do it better of making a lot of fuss about it. Right, yeah, so there shouldn't be people pronouncing from the ditch. Well, it's one thing to be concerned and thinking that perhaps the form is a mass limitation. It's another to say that it's going to be wrong. Yeah, so I guess that's an example of what I had in mind about, you know, how about, as I said, scepticism about the rigor of the whole interperson of using Presbyterian I guess it's obviously not something that particularly bothers most of the community who are working on, who are more directly working on the technical problems. I mean, I don't suppose people are spending too much time hearing Jim Anderson's candles. Yeah, well, I think Jim Anderson's... Well, I'm not sure I've only spoke briefly to him once or twice, but it seems to be directed toward very specific points, like using distributions as sources and things like that, and I think he's right to say that it's not proper to do that, but it really shows whether it really makes a difference. And it certainly will be, I mean it's certainly true that up to a certain level, up to a certain post-Newtonian order you don't have to worry too much about what the objects are, what the masses are. But I also know that it will make a difference at some post-Newtonian order whether you're talking about a black hole or a neutron star or a dumbbell or something else. So the internal structure of the object will certainly start to matter at some point. Yeah. Sure.

47:30 So, you were mentioning the other day the fact that it was difficult to get good postdocs. Well, I wouldn't want to be quoted on the record because a lot of people would like checked. See, I don't know if I know how to put it, for the purposes of recording. I just mean, how do you put it? Well, the situation is something like this. Because I work in a field, my interests are limited to a fairly narrow area in relativity. And I would like to have, if I'm going to hire a postdoc, I would like to have someone that has similar interests. Which means that the pool of people I really would want to go for is very small. Now the pool of people who hire those kind of people is also fairly small, but the numbers are comparable. which means that there's a number of people looking for a number of postdocs and those numbers are pretty much the same which means that you may be left over. So what I'm saying is that of the people, of the six people that I really would want because they work in directly related areas, chances are that they will be attracted elsewhere because there are better places than Guelph to do the post-doc. And my comment that I had a hard time finding a good post-doc is solely because I want to hire a certain type of person. I'm sure if I want to hire in all fields and group together, if I didn't care about whether it's a quantum gravity type or a street theorist or whatever, We'll find a number of excellent people, but because I'm talking about a rather select group of possible post-docs. Yeah, sure. No, that was the kind of thing I was getting, because you did say that obviously there were plenty of post-docs out there. and obviously this thing of the idea of getting somebody who has sympathetic interests is important to groups because one can see the way that

50:00 people often move back and forth between similar groups for instance, there's loads of people swap back and forth between Caltech and Cornell, one example or now there's a whole of Caltech people in Milwaukee so you can really see how postdocs move around historical work that I did for a thesis, you could really see how postdocs sort of seem to pollinate, cross-pollinate between different groups and create linkages between different groups as they moved back and forth. So is that still an important thing for developing I guess it's a good thing to maintain some kind of cohesion between groups, but I was I was thinking that it, well, I'm sure of what I'm trying to say here. I was wondering, for instance, do you still, have you maintained collaborations with people that you've worked with before now that you're in Guelph or do you find that you work on sort of different problems with different people or depending on, you know, who's nearby to there? geographic separation is a big obstacle. It doesn't have to be because you have all sorts of ways of communicating now, email, phone and all that, teleconferencing if you need it. So that if you really were intent on maintaining collaboration with somebody, I think you would be able to do it. But it would be slightly harder, but you should be able to do it if But what tends to happen though is that you collaborate with someone when they're in the same corridor, in the same location. And then when you don't see them on a daily basis, well you don't collaborate with them anymore because you don't see them. You don't want to talk to them anymore and so on. So, but, you know, just being said, I have collaborated with Pat while we were both grad students in Edmonton and we've collaborated again last summer.

52:30 I'm sure we'll collaborate again and I can certainly see collaborating with someone like Alan and other people at this conference. I think we have so, we're working in a very tightly different problem, we have similar ideas, I think it would be very surprising if collaborations didn't come up from that. But as being said, I think that it's much easier to collaborate with someone who's just down the hall that you can talk to and argue with every day. Yeah, I suppose it's necessary to hammer out the problems. And as you say, you can perhaps compensate for the lack of physical contact when you collaborate at a distance if you both share such considerable familiarity with common techniques and a defined problem that you can, you don't need such constant contact to keep on the same. Because I'm interested in this because it speaks to sort of issues in sociology of science about the transmission of knowledge and one thing that you notice in numerical relativity is where they have these big multi-group collaborations is the problems they've encountered and it's interesting to see the various ways that they're trying to get around it. So, for instance, I was talking to Ballard at Cardiff last week, and there... Ballard was at Cardiff? Yeah. Why didn't he come here? I don't know, but I think primarily because Thiebaud is visiting Cardiff this week. Oh, okay, so... Yeah, so I guess he had to be there this week. Yeah, as a matter of fact, Thiebaud wrote to me and said, oh, I'm going to be in Cardiff, you know, and I thought, oh great, and then I looked at the date and said, oh, sorry, I'm going to go. So, but anyway, and I think that's why I sat the volunteer earlier. So anyway, so he was talking about collaborating with, say, Luke between India and France, but there they, you know, they had worked together before, as you say, working on a that you mentioned. And then in addition, they would usually have to, you know, start

55:00 working a new problem together for a few months, and then they could go back to their respective places. Whereas in the numerical relativity case, you actually have none of those advantages. The problem is well-defined in one sense, but it's not clear how to actually do it. Right. And you have all these different groups and the different ideas, and clearly it wasn't working necessarily that well. But then you do see them trying to come up with different I've been trying to see if there are any similar issues on the analytics side, but on the whole it doesn't seem as if the size of the collaborations has grown necessarily from what it was before. If it has, it hasn't been a big jump, yeah, from, you know, of the order of what it seems to have been in America. Yeah. Yeah, so it probably doesn't represent any new issues. And also, it would seem as if, although, although the people like yourself have been moving, have been looking at issues that really, I guess, weren't part of traditional relativity like data analysis, signal processing and that kind of thing. It doesn't seem as if you've actually had people come in whose background was in that field. It's been more like people who are a world that have simply taught themselves. Yeah, that's right. So for instance, have you, where have you derived most of your ideas on data analysis? From reading or have you taught much to people that it's just been within the relativity group? Within the relativity community. I don't think I've ever talked on data analysis with someone who would actually know about it in the first hand. It's a little interesting, it's probably a failure of our community to have a lot more expertise from elsewhere maybe. It was done to some extent, I guess. It was done to some extent. I mean, Tom Prince was asked. Yeah, while it's right, yeah. But I'm talking about me, yeah.

57:30 But I, yeah, I agree. I mean, maybe, you know, who knows whether it's a failure or not, but it's interesting to me, right, that people didn't do that. So I guess it would be interesting to see as we go along, because my impression is that the number of the people on the experimental side of LIGO have their own experience in So it'll be interesting, I suppose, to see when they get a chance to turn their attention towards data analysis issues, how much the theorists and they will be on the same page. That's why it's important for someone like Tom Prince to be involved in a big way right now because he's someone who's got first-hand experience with gathering data and so on and so on. Yeah, well, it would be interesting to see that. I think I've covered all the points I jotted down earlier, so maybe we should just leave it there. Okay. Yeah, that's great. Thank you very much. Yeah, it was very interesting for me. Thanks.