Interview with Aszimo Tinto & Albert Lazzerini

0:00 What do you want me to tell you, but I can tell you. Maybe I should give you. Well, I'll say quickly, I'll say that it's the 16th of July at nearly 11 o'clock and I'm talking with Massimo Tintin. Okay. So we are starting. Yeah. We are on. We are on. Yeah. Okay, I can tell you what I'm doing specifically now, maybe where I come from. Yes. And now I got where I am right now. Why I'm doing it. I'm doing it right now. But the why I'm not so sure I can tell you. Okay. Well, to make the long story short, I can tell you, while I came from Europe, you know that I did my PhD in Cardiff at the time when gravitational wave detectors exist on paper and it was even, I mean there were some proposals around of doing one kilometer but at the time this was purely theoretical work. And because of my thesis, I came here to Caltech, and I joined KIP's group for working on problems related to LIGO research and data analysis. Well, that sort of led to two years of research, and then I met this group at JPL, and that's how I got involved with space research. Well, as you know, there has been a research group at JPL since the late 60s. They've been performing experiments using Doppler tracking. Yeah. And so I got involved with that. It was in 1989 or so, maybe 1990. and so first I sort of extended some research work that was done that I did for ground-based interferometers to doctor tracking experiments. That was sort of fairly easy. It was a natural way to open the door to get into this new field. And then from there I learned all sorts of aspects including how all these microwave components work, how you're going to optimize them to improve the sensitivity of your experiments, and so on and so forth. In fact, I became also an engineer, an electronic engineer.

2:30 Well, since 1991, I sort of moved into the engineering field. I think that brought me lots of understanding about how all the devices work and ultimately our experiment should be run in order to try to detect gravitational waves. So were you working on the audio electronics of the spacecraft and stuff? Yeah, on the spacecraft first. In fact, I worked on the development of radio components that are actually flying right now on a Cassini spacecraft. And that was work I went on from 1991, approximately, And then I moved, well, actually further than 94, on and off, but almost until a year before the launch, which was in 97. So I was involved in that from the spacecraft point of view until almost 97. And then I moved from the spacecraft development to the ground station. we are building a new finishing to put up a new beam waveguide antenna 34 meter dish and the cost of upgrading that antenna to meet the requirements for the Cassini gravitation experiments was like $34 million and so I was involved in identifying all the components, all the requirements for the ground and the hardware that is needed for this experiment but I think the exciting part of being involved in this this engineering task was sort of to prepare the background of the understanding for moving to a more outrageous and exciting mission like Lisa. And so that's really what I'm sort of interested about right now. That's what I'm working on. So it was a sort of long path starting from being a theoretical physicist, working on theoretical problems and moving to engineering and now from engineering coming again with an engineering background and now I'm complementing with a theoretical background to this new experiment. Yeah, it sounds like an interesting Yeah, so diversifying your views and trying to put everything together

5:00 Yeah, it's impressive to help develop skills in so many different areas. How different did you find actually more satisfying in a sense because you know where you're starting and you know where you have to end and so your end points are given and you know you have to do that within that time scale within that resource allocations and you have certain requirements to meet and so you sort of work something towards some milestones and reaching those milestones gives you an immediate satisfaction If I did that, then I know there's a next place to reach. Well, in research, sometimes you start, there is maybe the problem is sort of vaguely defined. And you go through so much frustration trying to just identify and formalize the problem in a way, to define the problem. And then there is a next stage of tackling it, to have an idea somehow. So it's a little more demanding on yourself, I think. I find it, you know. Yeah, sure. And it's interesting the way you put it I suppose maybe that suggests a way of thinking about the theory side, what some of the theorists have to do right now. It's a little bit made more like engineering in that LIGO has certain requirements. We'd like to have certain templates. We'd like to have them by such a time so the theorists are forced to think, okay, how can I do that? How can I get there? which as you say it's relatively unusual so that's sort of an interesting an interesting change of pace for them and that's sort of from the point of view of the work itself but then so many other factors are related to moving partially from pure academic work to the engineering work the personalities of people you are dealing with very very different and somehow simpler you know the ego is not so strong like in academia so it makes it easier to interact with them everybody is sort of working towards a particular goal which is a practical one there is no success in any sense

7:30 from the personal point of view trying to achieve trying to do your job right to produce hardware that is supposed to work, and there is not going to be a name attached to it, it's just a project. So, egos are not involved. Interesting. It's more of a collective effort. Exactly. And it makes it sort of easier, at least to me. I mean, previous experience, it was, you know, I was... So now that you're back, now that you're working more and an experiment, how does that compare to the other experiences? Well, they're still going on. I mean, I still, now I'm 50-50. I'm 50% a scientist, in a sense, an authorized scientist. In a sense, I get my living. 50% of my living comes from some research ground. And the 50% is engineering. And that's sort of a job, in a sense, a real 9 to 5 job in some respect. So I'm living with these double personalities and somehow I can complement the two and so far it seems to work very well. And I don't really miss the academic environment. and in a sense I come to meetings you can see that the ego part is gone even higher now because now big money is involved results are expected people that have done certain kind of work they want to become territorial this at least that's my perception so what's what's the nature of your work on least well the way i sort of now i i've been i became only last month i became a member of the admission definition team the reason was essentially because i discovered a way of synthesize an interferometer that has unequal arms. You see, Lisa intrinsically will not have equal arms.

10:00 So the distance between the spacecraft, you know, you have a triangle, essentially, is not exactly equilateral. Arm lines are different by a few percent or so. And so from the point of view of the titling gravitational waves, that means is if you have, let's say, the Korn spacecraft and you send light to the foreign spacecraft and then the light comes back. If you recombine that light at the same photodetector, then you don't cancel out the frequency fluctuations of the laser. That's because those fluctuations experience different travel times. So when they come back to you, they're sort of delayed to each other. And so they don't cancel out exactly. And the laser fluctuations or so. So you couldn't ever reach 10 to minus 23 if you wouldn't compensate for this problem. And so over the past 20 years, they thought that they had a scheme that was, you know, everybody thought it was working fine and so on. And then it turned out it wasn't really as effective as originally thought. There was, you know, the devil is in the details. sort of proved to be crucial in a sense it was not really working and so almost by chance because i was working on some prominent to not to be related i sort of realized that but i was purely lucky i mean it sort of happened to stumble on something related and i remember to study the technique and so i sort of made a link there and i said oh but this cannot work and it cannot happen so I was lacking and simultaneously I found a way to solve the problem and so I found a way out to you can remove these fluctuations from the laser even if the arm length are different implementing a proper trick there is an algorithm which allows you to do that and so and then there are generalization of this idea and how you know imply certain things then from what happens to the signal pattern looks like. It's not anymore the standard response of a long wavelength limit interferometer as you move from 10 to minus 3 hertz and higher, let's say, which is most

12:30 of the band, at least it will be sensitive. Then you have to take into account of delays. When the wavelength is shorter than the armlines, then a signal propagates through and one spacecraft gets hit but the other one doesn't feel yet. effect of the wave. And so there is a delay there. And so when you send photons from one spacecraft and back to the other, the signal will appear three different times. So there is a delay in which the signal enters into the response from each arm. And so when you synthesize this unequal arm interferometer, you get a response that shows eight pulses if you have a gravitational wave burst, let's say. It will appear in the output at eight different times, separated in time, you know, in a data stream, depending on where the signal came from and what frequency it has and so on and so forth. And so the question is, you know, you have a very complicated antenna pattern, which is not the usual quadruple or long wavelength expression like what Kip has in his 300 years of hermitation article, but it's something a little bit more complicated. And like it turns out that if you have a signal are propagating exactly orthogonal to the plane of the interferometer, and the frequency is exactly equal to the inverse of the round triplied time, then you find that the response is zero, contrary to what happens in a long wavelength limit when the response is maximum. So there are certain nuances there that are sort of counterintuitive. Yeah. Yeah, that's interesting. I remember in fact you were interested in unequal arm detectors you even were looking at a single arm detector yeah correct and that was what led me to this yeah in fact I was in 95 or something that I published this paper yeah you were still a Coltach yeah yeah yeah so yeah so you happened to have been studying the right things yeah in a sense and I was probably the only one who was thinking those directions I think in terms of one arm interferometer when everybody is already putting money into regular it was a little controversial not very well we see it but it turned out to be useful interesting so I'm curious I was at the session yesterday morning

15:00 I'm always amazed listening to people working on these space projects how, maybe I mentioned this last night, but I'm amazed at how people have spent such a long time working on the details of a project and not knowing for sure whether it's going to fly. So what's your impression at the moment, based on your experience? Well, it was exactly the same thing in 1984 when I joined Werner Schutt's group. Actually, it was really the beginning the Cádiz groups got involved into these issues regarding detection of gravitational waves. I mean, Bernard was doing purely theoretical work on relativity and post-internation approximations in general. he gave me a problem to actually give me a choice of two problems to work on. One was radiation reaction and the other one was finding the antenna pattern of a laser interferometer. And I decided to go to something more practical. at the time it looked more practical to work on those and that's how we got involved in TLCS but as I said at the time it looked still you know a far and long almost impossible shot to get funding for building a large and long baseline a ground base and now it's a reality in a sense it's becoming a reality but without that vision without that effort and then along the years until the refining came across yeah, what an amazing story that's come so far so I can see that probably similarly, maybe actually much faster than you expect could happen for space based interferometers because now all the background work done by the ground based has been done, you know, and somehow we have already a case. Ground base has been funded. They are going to be built pretty soon, a year or two. Maybe we still get some data. And so many countries in the world, the United States, first of all, committed themselves to projects. And in a sense, I mean, it's not really guaranteed. The event rates we heard are not very high, and lots of luck has to play, in fact.

17:30 Space-based, we know if we get that sensitivity, we will see it for sure, if the theory is correct. So you would expect a follow-on on these facts, and maybe funding to come sooner than we expect. At least that's my anticipation. Yeah, there's good grounds for optimism. As you say, there seem to be probably a lot more sources out there for this. Exactly. Yeah, and also in a frequency band, it's complementary to a LIGO, ground-based inter-ground detectors we will observe, and so it's not really hurting anybody or any other project. It's not competing, in a sense. So do you see yourself working a lot unleason? Yeah, I think there are so many other problems now that can be looked at or re-looked at. Yeah, I can see at least three years at least. That's the timescale of the next funding I'm applying for. So I have to say three years at least. No feeling. So, what do you think are the main issues that need to be addressed for this? The design issues, I think, stay open. that probably they haven't realized yet what they thought about but not in a way and that's something else they will follow on with the JPL group Frank Castlebrook, he met him and you know John Armstrong is the only man who was involved in this I can see that but in principle there are no showstoppers fix little things here and there but that's natural I mean that's how things evolve I'm sure there would be many other things that might not be considered before we fly Just to quickly jump on to another topic, what's happening with dopper tracking these days? Doppler tracking, I think, after Cassini, which is in the experiments that would be in 2001 and then 2002 and 2003,

20:00 so there would be three 40 days campaigns of data collection, 40 days of continuous tracking. And that's sort of considered to be the ultimate spacecraft Doppler tracking experiment, because it's using Ka band. It's using a microwave link that is the shortest frequency ever used, the shortest wavelength ever used so far. And that means you minimize going to 30 or so gigahertz. scintillation, so the frequency fluctuations introduced by the interplanetary plasma in the ionosphere. And so, and plus it's relying on, there is another hardware which is probably unique in the world so far, and it's been developed by JPL, it's called a water vapor radiometer. It's an instrument that sort of looks at parts of the sky where the antenna also is looking at, and monitors the brightest temperature of atmosphere where the beam goes through. By doing that, there is then an algorithm by which you can, from the temperature, infer the variations of the index of refraction of the atmosphere, which is the main noise source after you remove the plasma noise. And so by doing that, you really bring the sensitivity down to whatever your reference clock, hydrogen measure, it sort of keeps track of the frequency that you're using, and that's part in term of 16 or so. So Cassini is going to be the ultimate Doppler tracking. It's the ultimate one-arm interferometer, so to say. And I think since LISA will come up, I don't think we'll make, I mean, if there will be an opportunity, why not to try? And they are very cheap experiments. You don't have to build anything that is provided in on the ground, you just use it, you just track the spacecraft and analyze the data. At some point I think the experience you have built copper tracking spacecraft using microwaves

22:30 can be translated or brought into the analysis of data that Lisa will produce. So I think John Armstrong in particular has been working on analyzing the real data from so he has decades of experience in analysis and they will be very valid and will be very valid because a lot of the people who have been looking at data analysis have never really had any real data probably the only ones that will be interesting well is there any chance you think that Cassini will see in principle I mean, there's some huge black hole, a huge pair of black hole spiral together. Because it works in the same band as Lisa, essentially. Actually, at a slightly lower frequency than it's sensitive. But the levels for sinusoids, we're talking maybe over a period of 40 days of integration, maybe parts in 10 or 17 or so. So we have five or more orders of magnitude away from what Lisa is expected to do. So the chances to see something, to me, I mean, it sounds, there is always a possibility, you know, if you look at what kind of source you could have, in principle you could, you know, put together, you know, certain masses and certain orbit parameters in such a way you would see something. But then, you know, you can go up to maybe a beautiful cluster of galaxies or so. I mean, there are a thousand galaxies out there, but, you know what I mean? You have to be lucky. So, to go back to what you were saying about your unusual background, is that when you now come back to working on something like Lisa and an experimental project, is it a big advantage to have both the theory background and the... I can tell you, you know, it's hard to... One has to try to remember what you knew before you do certain things and how you know now. always impossible yeah but i think in a sense you feel more comfortable uh when you talk about spacecraft configurations and payloads and what's really the structure of the spacecraft how is oriented how you sort of monitor the status of the spacecraft i mean things that

25:00 are sort of routine in on any spacecraft you know then star trackers and objects of that sort I think whether you're talking ESA or any other kind of spacecraft developments if you have gone through some engineering work you sort of become familiar I mean it's unavoidable whenever you do something you learn and so I'm sure it helped me to understand a little bit more if you go read the proposals that they have concepts, in a sense. I'm sure, I mean, the same thing that if I would have read before becoming an engineer would have been a little bit more obscure, I'm sure. Yeah, sure. How many people are there who are interested in gravitational wave detection who are familiar with space projects and, you know, have the background? Well, Faulkner, Bill Faulkner clearly has a strong background. and he comes from, again, spacecraft missions. He has analyzed data from spacecraft. I think he has done the LBI experiments. So Bill Faulkner, I think, is really a very knowledgeable man, and I think he has proved that with what he has done, and he's doing right now for the project. Then I think Frank Estabrook and John Armstrong, derived this generalized response the three-pulse response in 1975 has been P.I. and Galileo and I think John Armstrong from the point of view of the data analysis is really the man as far as So there'd probably be relatively few people maybe some in the States and some in Europe who Yeah, but I think from Europe I don't know all of them, but I know most of them. Most of them, they come from the ground. They have a ground experience, like Jim Hoff and Albert Rudiger. So, Ronald Schilling, you know, they have a strong background. I mean, they are very well-known experimentalists in interferometers development field.

27:30 Right. And so, in that sense. but I'm not sure whether they have ever dealt with the spacecraft development. Right. So they probably don't have that kind of mechanism. Yeah, I'm not in their brains, but they are so smart and so very well as some of the experimenters. After all, it doesn't take very much if you put yourself into it. And they've been working on this project for many years, so they could very well have developed. Sure. But I was impressed listening to Faulkner's talk yesterday and the level of background detail that seemed to be required. Just, it seemed to me, in the sense of knowing what's there. Well, for instance, he was speaking about efforts to get the test experiment into space at a certain cost and looking at other satellites that were going to fly and trying to see, well, could I get on that one or so on? That seemed to be, well, the background information that one really needed to know. Yeah, at that level, managerial experience. It might sound trivial, but it's not really. And also being a good diplomat, a good political beast, in a sense, it's not trivial. I don't know whether one can really develop that or it has to be sort of natural. certain people cannot deal with other human beings so they cannot deal with trying to compromise and sometimes strength is in compromising the way things go, maybe it doesn't give you 100% of what you're looking for but make things happen and I think Bill has got capability obviously that's an important skill it was interesting to see Ray was No, just go ahead. Do it. Well, you know, it's easy to say that when you're A-wise. You can just do things. Otherwise, but... Well, it'll be interesting to follow these. I've been interested in it since I started off as a graduate student because some of the work I was doing I knew would be... Well, actually, I remember the first project I was doing

30:00 detectability that Christa did in memory. Oh, yes, actually, I remember your presentation. It quickly became apparent that LIGO wasn't going to be able to see it. But on the other hand, that Lisa might actually see it from certain types of sources. So right away I was sort of made aware that, you know, Lisa has some interesting things that LIGO can't probably do. So I've always been interested in it. It's interesting to see how it's... There was a proposal, I gather. So ESA at the moment, yeah, ESA at the moment is an ESA project. Yeah, but it was a dance one, so I gave it. Mm-hmm. And the point is ESA, the way of counting and making cost estimates of what a mission like that would cost. It's always gold-plated, so to say. that's a word that has been this phrase that is used I don't think it would happen soon it would happen in 2000 15 or 17 if it would happen I mean it's it's been approved but for a long shot which means you know go back and come back maybe in 2017 so gold plated means that it's that is is done using you see when when you design a spacecraft you rely most of the time on the data bank of information regarding the components you're going to use and how much redundancies you want to have and the reliability of the components you're going to use and so on and so forth and only I mean at JPL we They start looking at things that are produced commercially and available now. And so we are sort of updating all the data bank of information about components that we put on board the spacecraft. Now, at least the ESA approach is a little bit more conservative. And they put in there that the bank stuff has been flown maybe 20 years ago and it's clearly reliable, it's been working and so on. But the cost and arm and the lag is obsolete and it's very heavy and so on and so forth. if you use those other building blocks putting together a spacecraft then you come up with a design which I'm sure it will be 100% full crew

32:30 but you're going to pay it maybe 10 times or 20 times more than what it would cost otherwise if you rely on commercial it's probably as reliable and so you see they have a completely they have a different approach to design missions and do a spacecraft and that's what makes a difference in cost and therefore in time scales So during which you can expect a mission to fly. So people are still hoping to get a mission that would fly a lot earlier than NASA's people? Yeah, I think it could happen if, for instance, now what the main goal now is to make ESA, NASA, to talk and find a way. And most importantly, to have each of the two sides commit themselves. You know, now it's like to say, okay, I'll do this, but you ought to do something like that. And vice versa, you think exactly the same. Nobody's sort of taking a first step and sort of postponing. But if eventually we converge to a table and say, okay, we'll sign an agreement. I'll put half, and you put half, and we'll do it. Then the chance it will fly. It's pretty much higher. Because meanwhile, you know, ESA might agree that this way of costing spacecraft components obsolete and converge to a more modern realistic approach of designing spacecraft. So it could happen. I mean, maybe before the year 2010, in fact. And if we would, I mean, I hope LIGO would see, LIGO or any other interferometer would see gravitational waves by then, but, you know, I remember in the early 90s, you know, first of all, they were thinking to be aligned by 1995, 1999, and at the time, if I remember correctly, LIGO, the first LIGO 1, so-called, the one that would come up online, was already supposed to see something. At least on paper, that's what they claim. Now, it's not quite clear. They're already starting LIGO 2, advanced LIGO, whatever it's called, and then maybe even third generation. So they're already sort of putting their hands in front. And so, you know, maybe timescales now might be comparable. At least from the first detection point of

35:00 you maybe could have it in space. That's true, it would be interesting. And that would help, I mean, not only whoever does it would get a Nobel Prize, but then the whole community would benefit. It would be a uniquely identified detection, and it would be either in one band or in the other, and so if LIGO detects it first, then obviously Lisa would probably go up much sooner. Yeah, so the main goal is to try and get AESA and NASA to do a join? I think that's really what is behind the scenes. I know that there had been an effort maybe by Ron Hellens was the leader to have the Amiga project flying NASA, but that didn't get anywhere. Yeah, at the end, it has been seen as a diversification of energies. Why do you have a competing mission when you can just focus on one? One probably is endorsed by everybody and put in bets on that. I think recently, maybe it was a few months ago, So it was decided at JPL to stop this Omega mission and put all the efforts into RISA. Yeah, but until a few months ago, until the Morion meeting was generally the same. Yeah, Omega was still around them, it was a smaller Lisa version. Right. So at JPL, there's sort of, at some level, an official decision to work on leasing-related stuff. Yeah, now JPL has endorsed, let's put some money already. There is a project office for leasing, in fact. Faulkner is sort of the manager, and I don't know exactly what the figure has been put in. It's not much, but I mean there is an official reference office for Visa for Visa project. So JPL is behind. And JPL has done quite a bit for Visa in general, even when, you know, for the European. as far as the latest cost estimate was done at JPL there's a group

37:30 called Team X and what they do is they sort of design missions for whoever comes in forever and you can have a PI from some university that's planning to do an experiment in Spanish and he comes to JPL and he describes his mission and then there is a group of people maybe 15 or 20 people In one room, there is a sort of oval table, and behind it is a large oval table where all the workstations are. And each person is sort of an expert in spacecraft design. There could be a person expert in trajectory, another one expert in telecommunications, another one in attitude control, and so on and so forth. And so given the kind of scientific goal that the PI wants to achieve, let's say, want to perform, the kind of experiment he wants to perform, in one week or so can design a mission, can design a spacecraft that will fly and will do what the PI is asking for and gives you and provide a cost estimate of what it will take financially to do it. And so LISA was one of the studies that was done at JPL eventually to converge to a more realistic cost. So the figure was presented the other day of 400 or so millions as a result. it's going to be interesting to follow the story yeah it will be also in a sense what we have seen from Faulkner and Dance all the politics of this playing behind the scenes it's sort of fascinating if you leave it from outside that's true so like a chess game Yeah, it must be very difficult to follow the internal politics. Yeah, I sort of see whatever you have seen. Yeah, yeah. I sort of hear from the folk but I'm not right there. Yeah, it does look fantastically Byzantine. Oh yeah, exactly. It's moving, I think, especially in the last year. The vision of how to make things happen. driving the old project to make it happen soon. Well, I'm really conscious of a lot of optimism about Lisa recently, just from talking to people at every level

40:00 from the people who are relatively closely involved and then to astrophysicists who tended to be very down on LIGO or seem to be much more interested in Lisa, which is interesting as well. so it does look like it has good prospects well thanks very much well thanks to you in a sense talking out loud about what has done people do maybe I mentioned that last night when I go and talk to people I say well I'm looking at the for instance say the numerical and say, well, I'm interested in kind of the sociology of the field. Well, let me tell you about the sociology. I've got a lot to say about that. So how many interviews have you done during this meeting? In this meeting, let's see, just a handful in this meeting, and a lot of informed conversations. Did you interview Ron Dreeber? No, I haven't, although not this meeting anyway. I have talked to him previously one time, and I'd like to do so again. It's one of the problems. It's so hard to get to talk to everybody. For instance, at 1 o'clock, I'm going to talk to Albert Lazzarini. Oh, sure. Yeah, that analysis stuff. It keeps me busy when I'm on these trips. But I enjoy it. It is interesting to talk to everybody and hear what people are doing, hearing them talk about their work. Oh, that would be nice, too. And it seems to be working okay, and it's the 16th of July at a quarter to one. I'm talking with Albert Nazarene. So, well, as I mentioned, I'm interested in data analysis issues, and as I understand it, right now you're working on getting a standard software package that would be used by different people interested in LIGO data analysis issues. So I was interested to kind of try and find out something about the background of that. Sure. When we started looking several years ago at putting together the data analysis environment for LIGO, we realized that there are two conflicting needs that we need to fulfill. One is the archival need, which is storing massive amounts of data

42:30 and delivering massive amounts of data to users. And the other is the part which has to do with allowing people look at small snapshots of data, and in that sense, basically, we call it a sandbox, which is basically you do rapid prototyping of things. We determined from our perspective that the more important and more difficult problem in the long run was the archival and data delivery part. So what's evolved is that the quick look analysis, prototyping stuff, will be done with a host of either public domain tools that we've inherited through collaborations with other gravitational-related programs. The GEO people have put together a Java-based front end that's basically a web browser environment where you can look for and visualize small data sets for rapid prototyping. The Virgo people and some of our own researchers have identified than the root package coming out of CERN, which has been developed almost exclusively for high-energy physics analysis. But it turns out that if one is willing to build signal processing front end to that, it's a perfectly adequate environment in which you can do similar types of analyses to the ones that this package from Geo can do. And then if you want to go to commercial packages, things like MATLAB are perfectly viable for this quick-look analysis since there's a whole self-consistent set of libraries that have been bought. So what we've focused on is working with Virgo, actually, is a standard for how the data are written to disk or to tape so that by a relatively small investment building what are called application programming interfaces, APIs, that allow these other tools, these four or five other tools that are out there, to understand and read LIGO or Virgo data, data also because they've committed to it and the Japanese have also. So we have a standard structured data format that grew out of originally some prototyping work that was done at Virgo at the ONSC group which is ONSC friends and we worked with them and we went from a prototype to a firmly defined specification that now we both control that details exactly what is on a data frame and so we now have C libraries and C++ libraries that have been independently written

45:00 that have the frame format as a medium of exchange. You can write it, you can read it, you can manipulate it. So that's one form of standardization. Then what we're working to on the LIGO side, and we're focusing only on LIGO data analysis for the time being, is coming up with an efficient set of numerical algorithm libraries. Now, this has been done in other instances for other applications, so there's a lot of heritage that we can bring to LIGO, but what we want to do is make sure that what we bring in is a self-consistent standard and specification, and we've called it the LIGO-slash-LSC algorithm library, L-L-A-L. And basically, if you're familiar with other numerical techniques, there are matrix manipulation libraries out there, like LIDPAC or LAPAC, Numerical recipes. You can buy the book, get the CD, and you have a set of libraries. And each of these are very good for what they do. And they usually form the kernel or the seed around which you amplify for your own customized need. So the intent we have is to build a lot of LIGO-specific data analysis tools that will allow us to manipulate time series, frequency spectra, frequency time plots, The relatively few types of data types that we have, the data products, which are basically series in time, series in frequency, X, Y pairs of numbers for scatter plots or things like that. And so the idea is to very rigidly define the data types that we will support, and then a standard or style guide by which any piece of software that manipulates these data types should have its input-output behavior. here, if there's an exception that's generated because a divide by zero is an example, that will be specified such that all software behaves the same. So when you see the error, the error has a certain amount of information in it. So during the debugging integration phase, we don't have to spend a lot of time scratching our heads. So where we are now is that this is something that the laboratory, the LIGO project is working closely with the scientific of collaboration that's involved around LIGO. So we've been working with Ray Weiss as a spokesman and several of the scientists coming out of the data characterization for the detector and the analysis and confidence groups.

47:30 And three individuals have volunteered to write a couple of test programs in this environment, see how it works, iterate on the SPAC, and then hopefully use those as templates with which others who haven't been as familiar with it can kind of copy and cut and paste and start to evolve software. That's where we are today. We're basically at the point where we've released the initial draft of the specification and we're receiving comments back in the next month for an iteration on that. At the same time, as I said, there have been two or three toy packages that have focused on several different functions that we want, things like efficient decimation of data. So you collect data at 16 kilohertz, but some of the physics you want to look at is at 100 hertz. You're oversampling, and for efficiency of data transmission and data manipulation, you conceivably want to decimate the data so that you only keep frequency content consistent with 100 hertz. So there's a filter that will be built for that. Some of the pulsar work, you target a point in the sky, you look for a pulsar at a putative frequency, then look at residuals around that frequency. So you want to be able to digitally heterodyne where you multiply sine or cosine of omega t times the data and then baseband it back down to dc. So dc corresponds to the carrier, so to speak, and then you just look for residual frequencies around that. What that also binds you is if you're looking for very narrowband phenomena, once again you can take data that are sampled very rapidly Since you refer 0 hertz back to some high frequency, you can actually reduce the bandwidth and still keep the content. For example, if you do the heterodyning at 500 hertz, then a 20 hertz bandwidth contains all the information from 500 to 520 hertz, but you're only keeping the data at consistent with a 20 hertz sample rate. You give up the dynamic range of frequencies, the volume of data accordingly for certain analyses. So that's where we are right now. So when developing these packages and so on aimed at the biological scientific community, how many people did you, you know, did you, you know, take input from it?

50:00 How many people do you have to sort of... Don't put it in the past because we're in the formative stage and we're really kind of, I'm concerned about how quickly the point that we'll be generating data is coming. So let's say our doing and should be doing. What we did is we formed a working group of approximately eight individuals that correspond to the chairs of the data analysis groups, the spokesmen from the LIGO scientific collaboration, several of us from the laboratory who were involved in data analysis, the chief software design task leader in our group, Kent Blackburn and myself, and other people who've done Pulsar work that are in our group. And amongst the eight or so of us, we hashed out and iterated to an initial definition of the specification I was talking about. Once the comfort level increased to the point where we had consensus, then the next step is you broadcast that or you publish that SPAC and ask for greater feedback from the community. We had, at the beginning, we started with a much bigger group, but it became chaotic because you had so many feedback comments that it was impossible to accommodate everybody. Literally black and white requests. So the idea was to have a small consensus come up with a prototype or a draft, send it out, and then recollect the comments in a coherent fashion and then filter the comments and decide what needs to be incorporated and what, unfortunately, cannot be incorporated. So that's where we are now. So a smaller corporate provider so people prove better? Yeah, and the intent is there's a lot of goodwill in the LSC in terms of people who want to do algorithm development because they have some nuance on how to look at things that they want to exploit. So as soon as we get these tools in place that allow them to build tools as opposed to just building standalone pieces of code that more or less becomes a tower of babel. And a lot of us look back at high-energy physics, and most of that stuff has been much more individualistically driven where analysis packages have been written by one or two groups, and there's not much intercommunication between them. And what that promotes is a lot of orthogonally independent analysis of data, but it also means that there's a lot of manpower that's spent. And what we find is that if we're building code

52:30 and an infrastructure that's going to last for a while, we really want to spend some time up front thinking through how we're going to put it together. What are the main issues, do you think, from the point of view, creating that type of infrastructure that will enable people writing their own packages to really be able to use them to change them? There are two parts. If you look back even a couple of decades ago when there was a lot of code that was being written for physics analysis of custom code, as I was telling you earlier, you didn't have a model to draw on, which is commercial software development, which is now just an exponentially growing field. What you find is that the success in the commercial software has been going towards object-oriented programming. C++ is a standard, more, not dated, but there are newer tools that are coming out like Java, et cetera. We have focused on C++ because it is a standardized and C-standard language that supports onboarding the program. So all our infrastructure, the behind-the-scenes archival code, the data manipulation code, is being written by C++. We've had to hire several professionals, contract programmers. What we're finding is, you were asking what one of the issues is, is that by and large, the science community has a whole spectrum of Fortran people, C people, scripting language people, Mathematica people, and we're working very hard to develop a good impedance match between the need to have a boilerplate industrial hard-proof code that doesn't crash and not to stifle creativity on the part of people who don't quite know that. So the approach we're taking is that the algorithms, even though our base environment is C++, the algorithms are all going to be done in procedural C, which is more commonly known. And then we, as a laboratory, who are responsible for the infrastructure, will strive to build wrappers, as they're called, which allow C subroutines to be incorporated into a C++ analysis environment. And that's how we do it. So there would be sort of work for the central group in trying to help fit these various packages together? Yeah, I mean, we're working very closely together. We interact regularly via teleconferences between the MSC and us.

55:00 We're still forming, it's still very cultural. We're still forming rapport and working relationships. need right now, and Ray Weiss's spokesman can tell you about it, is we need to have an authoritative spokesman from the collaboration at large who represents the LSC in a requirements posture, which says, this is what we need, this is what we want, and then we work with the spokesman. As I mentioned earlier, if you go too far down into a grassroots approach, It becomes chaotic. So that's what we're working is to establish a good rapport with the management side of LSC so that their needs can be represented to us and we can work with them to implement what's needed. So on the other side of the coin that you mentioned when working with the other detector projects, does it make it easier in that case that they're more... We focused early on a unified data format. And to date, we also began tentatively exploring joint software development with Vergo that hasn't gone very far for various political reasons that I'm not going to get into. But the other part is that our decision to go with a straight C++ ANC standard is not necessarily something other groups want to follow. So at some point, once you agree that you have a common currency, it's okay to diverge a little bit because you know that you can always compare apples to apples when it comes to the science. So to date, the Virgo people are building an environment which is mostly based around this packet group that I talked about. It's not quite the implementation LIGO has. And to the best of my knowledge, the Geo people are focusing very strongly on a Java-based implementation for data analysis. So our concerns are longevity. We don't want to rewrite everything every five years if we can avoid it, robustness, and also, just as importantly, is the ability to parallelize analyses over many, many computers. We anticipate that many of these analyses are going to be very compute-intensive. So we've picked an implementation that supports message-passing interface protocol, which is a way you can parallelize many computers to act as one big supercomputer.

57:30 It's a standard, industry standard implementation of C or C++, and we will be using that for tasks that need to be parallelized. So between the detectors, it's the level that has been elected to coordinate to is at the data format. So beyond that, people have gone their own way. I'm sure we will pass algorithmic implementations around it. If we receive a code fragment from Virgo that does something that we particularly want, it's essentially very easy to just immediately cast it in the format that we need. So that's not the concern. In an ideal world, you know, we would be able to kind of allocate tasks. You do A, I'll do B, you do C. We're not quite there yet. We've been talking to the Virgo people about that. And there is mutual interest in that. exploring it, but it has a lot to do with just how the various groups collaboratively interact. Right, right. Yeah. And so it's sort of, are there both pluses and minuses then when you're dealing with a group that presumably has a more centralized structure than, say, the LSC? On the one hand, presumably you don't have to please many different people, but I'm just wondering if there's a difference between... say there is. Most of the interactions, you know, they take place simultaneously at two levels. There's the directorships that take the peer, and then the workers have good grassroots relationships. So it's really the people who are actually... We have, you know, at least once or twice a year, we go to we meet with the Virgo people, mostly the French group, because... Yes, sir? Have you seen our two favorites? I saw one of the two. Which one? Finn. After I talked to him? This morning. Yeah, good. Strike that from the record. Okay. Where was I? I don't know if it'll pick it up. So, let's say you were commenting that when Claverny said with Virgo with the people. We worked very closely exploring much more close software development. We had weekly telecoms, and now it's basically quarterly we get together.

1:00:00 And like I said, we work always to keep the current, the specifications for the format of the data very current. Benoit Moore and company at ONC have actually been implementing certain changes that we've requested from the LIGO site because we have three interferometers. There are certain needs that being able to put multiple data sets together in this frame structure, which wasn't originally envisioned because the sphere goes focusing on one instrument, one interferometer chain. And so since they were the ones who were most conversant with the code structure, they're the ones who agreed to look at how to do that in a self-consistent manner with an eye to retroactively supporting everything we've done in the past. You don't want to break legacy code. And they've successfully implemented that. And we haven't yet started using it, but we will try to test the code make sure that it works as advertised. So we're working collaboratively in code development at the frame libraries. So the experience actually of collaborating with one of the other big detector projects is actually not that different? And as you know, Dan, the Geo people are actively involved within the LSC, so they actually are going to contribute software more collaboratively than do Virgo. And the hope is at some point we'll be at the same level of interaction and all the indications are that that will certainly happen when there's real science to be discussed. Right. So, and I presume especially the fact that you've already got good experience with that. That was the intent, yeah. Well, that sounds very interesting. Thanks very much. Sure, my pleasure. Thank you.