So thank you so much everyone for coming. And I'm the chair for Darwin, Darwin Symposium. I'm a faculty at the Department of Physics and Astronomy. And I will first present the fantastic variety of our registrants. So we have faculty and students and postdocs and the researchers. And as you see it, not only from University of Delaware, for also all of our regional partners. So thank you so much for coming. And as you see, we have attendees from many, many different fields, from many different departments. And at some point, all of us probably have finally face the problem which we couldn't solve on a single core, right? So one of the reasons why here doing high-performance computing, it's that somewhere along the line, you or maybe even your advisor have tried to solve a problem on one thread or 11 single node, and it just didn't work. For us. It started about five years ago when we were faced with the problem. We had the code. And you tried to do all possible code optimization. We have tried to do all possible approximations in terms of physics. I'm a physicist and it just wasn't enough. So what our team does, its atomic physics assumption, we can compute properties of atoms in the US, atoms for anything from building quantum computers to actually understanding fundamental laws of universal searching for dark matter. Dark matter, so fantastic thing. When you look at stars, you'll think that's our universe. That's what universe is made of, was a beautiful stars. It's 0.5%. So we know what the rest four per cent is. We don't know what the rest of the universe is made of. And one of us think our team and many other researchers do is trying to understand what is actually is. And we had a very specific high-performance program problem. Our experimental colleagues were trying to build a very high-precision atomic clock. If you ever use the GPS. That's works because we have atomic clocks on space. But this is much more precise, precise clock. This is clock will not lose 1 s in a lifetime of the universe. And we want to design it in order to look for some specific dark matter has a lot of practical applications, but that was our interests. And our experimental colleagues told us that this proposal just doesn't work as they look at this atom and they're not seeing the light, right light coming out of the right color. Nothing which theory predicted could see was seen by experimental. And they essentially that was not our proposals. They ask us, could we solve this problem? Because it can't build a clock if it don't understand how this algorithm works. It wasn't one of those for experts. Highly charged ions have lots of electrons stripped out and there's no experimental data for it. It's all theory prediction. And they said, we looked for it for five years and that doesn't work. So we had a group meeting and then we also had invited our collaborators who works without doubt codes. They came to conclusion. We just have to really run on many, many, many cores. So Charles Chen quiz, Here's an audience. He was a student in this time, so he got allocate and with this project, the MPI version of the core. And it's not for us to just took too long. It's because we need to so much memory that we need a really, really a lot of nodes via ran on. If you know. For Darwin, lush memory partition fits 30 1 tb. We literally discuss how to fit the job into 30, 1 tb literally. So it worked first jobs which took us three weeks, took now 15 min. And I still think it's a miracle. You just look at it in there. Just like stuff which took so long, text a little. But there we were able to actually solve problems, science problems, which we could not solve before. And that's where the Darwin and Calvinist in the fantastic HPC community at UD is important for science because we can solve problems which we can solve any other way. And phosphate bound actually, what the problem was, the problem that the color of the light was correctly predicted, but the intensity was not. So the intensity of the light which were seeing was just three orders of magnitude below which it was predicted. So the instruments just couldn't see it. So we understood what the problem was. And that was one of our first problems, which are sources Darwin. And now today with posters you see many, many other interesting science problems. And I highly encourage everyone to look at posters and talk to students and postdocs, which have made fantastic drops and presenting it. So today we will see posters. You also hear many science talks of what actually became possible because of high-performance computing, because of Darwin, and which important science problems have been answered. And also for us, with the high-performance computing, it's important to be part of the community because I collaborate now obviously with the Eigen man who when he came to UD, met with everyone who actually have done high-performance computing here. And the now also understand how to take our research codes, it makes them into software to really think as a computer scientist and not just as domain scientists to see how is it actually possible to make all of our data we ever produce a valuable for the community. And that is also very important because now it's not only that we can solve science problem, also can be part of your cyberinfrastructure to add to existing knowledge and to make every, all the codes and all the data valuable. But just a little bit of a statistics on Darwin. There is almost 600 user accounts here as you see, so many people have both axis. Of course, in the UD accounts, there have been 209 startup allocation, so initially, and now there is also 24 major projects. That's four groups which actually submitted proposals. And God, large-scale half 1 million or 1 million CPU hours allocations. And today you will hear about fantastic science which have resulted from it. And Darwinian, for those who may be new to the community. It's been purchased as a part of the major, major research instrumentation ground from NSF. And you'll hear more from NSF today. It's 2 million facility where it's 1.4 million was given by NSF and x 600 K came as a cost share from your stove, Delaware. And we're very grateful for newest of Delaware and to make this happen. Now, one thing which I found this high-performance computing is that it doesn't really matter how many computers you have. It's really never enough. Somehow. You always find this problem you could solve if you just has more memories or if you just have more CPUs or GPUs. And yes, we would like to make the algorithms as efficient, as smart as possible. And that's another part, the algorithmic development and also the NPI. And being able to talk to computer scientists and two experts from IT also makes a lot of different first, but eventually we ran to the point that, well, how many CPU hours per year do you need? And generally more than you actually have. So I remember the first time I heard one was a kindness metta scientists saying they're using 10 million CPU hours. I was like What I wouldn't be using it for. Like what kind of problems we can solve the extra thought, thought about it. Well, I have to admit we did use more than 10 million CPU hours last year. And again, for us it's as a memory, whereas the number of CPUs all also algorithmic support, which is necessary. Of course, you may ask what about artificial intelligence? Well, this is really fantastic new opportunity we have. We do have an AI project as well for us for now, the first thing which you are doing because AI is to do a central smart computing. That AI tells us what actually needs to be computed, but eventually we'll feed it into our high-performance computing. So now it's been more than three years since Darwin. So one of the reasons for this workshop is to discuss what's next. We would love to hear input from the community. What are you computing needs in Delaware and beyond? What really you need? And if you have seen our survey, we very much asked you to fill out the survey. It will inform today's panel where we discussed also the future darwin and high-performance computing FUD. And now if regional partners and we really would very much like to hear as the input from you. Thank you so much for your attention. And I would like to invite our next speaker is a Provost, Laurel Carson of University of Delaware. Thank you so much for coming. Hi everybody. Good morning. It's my privilege to join you today and welcome and maybe set the stage for some motivation for you today. So I'm a cognitive scientist and a self-described data geek. We could list Psychological and Brain Sciences on your, on your plot of disciplinary interests that intersect with all the work that you do. My research area is a field called spatial cognition. So I study how we represent the external world internally with a focus on the cognitive mechanisms that we use to build these representations and the language we use to describe space to others. My favorite work is interdisciplinary in nature. And across my career I've worked with computer scientists, linguists, anthropologists, architects, and mechanical engineers. I believe in the power of data be at survey results, GIS analyses, response time distributions, linguistic corpora, simulations, or databases of observations. And that brings me to the important work that you'll talk about today. In addition to all of the high-performance computing enabling you to solve problems that you already know. I believe that there are questions that we don't even yet know how to ask. And those will arise from the tools and approaches and perspectives that you will bring today. As you think about the creation of high-performance computing systems, algorithms, and processes to bring to bear on big data. And that I think is the most exciting part of today. As an interdisciplinary scholar, I believe that the biggest and best ideas come from bringing together partners across disciplines and across arenas and industries, and inviting them to spend timesharing and most importantly, thinking together. In my mind, that's what today is about. That's behind this incredible innovation that we see that's transforming the University of Delaware, the strategic and collaborative partnerships across the region and across the state. And this hosting of the Darwin computing Symposium by the Data Science Institute is one of the best examples of three elements intertwined. Define what is so special about UD for me. And that would be academic excellence, community and mission. For Academic Excellence, UD as a place with vision, a place that sees the firmly believes that science and education contains the world and sees the infinite possibilities of each. I think this belief translates into a concentration on big problems. Require many disciplines, approaches, methods and strategies for true impact. Think about disaster relief, biopharmaceuticals, climate change. Data science is critical for advances in any of these areas and more. With respect to community, I studied buildings with an eye towards the activities that they cultivate and the expectations of what happens within those spaces. Buildings are more than physical structures. It's what happens in them, and it's the synergies that they create through co-location and architectural design and the convening of interested and interesting thinkers. That's where the magic happens. So e.g. when you walked into this room today and you saw the setup and roundtables that setup some expectations, I hope for you that this would not be about you sitting there looking in one direction toward a speaker who's looking back at you, right? Otherwise, it would have been configured spatially very differently. Instead, I hope it set the appropriate tone that you are expected to interact with each other at your tables. The last piece that's really resonated with me since I've been here at UD is its mission. And that's connecting, that connects academic excellence and community. And that's this unifying commitment to impact both to the state, to the nation into the world. So we've gathered you today as a community in pursuit of the mission that we met through academic advances. Thank you for joining us and engaging with us and with each other through the Darwin symposium. The problems in front of us require us to dream big. And our ability to dream big depends upon our ability to bring data to bear, to enable us to see patterns, test theories, and anticipate consequences. Your work is critical and I wish you the most productive day. Thank you. Thank you so much. And next we will have remarks from NSF, director of our own, Shambhala. He's from office for advanced cyberinfrastructure, which is what's responsible for the high-performance computing for Darwin. And also force Artificial Intelligence. Thank you so much for coming. Here. Perhaps the media, there are questions. Perhaps you can actually ask them questions as well after the talk. Good morning, everyone. First of all, let me thank you for inviting me here. Thank you, Mariana, Rudy and Darwin t. And this looks like a great event. I would, I just took some time to look at the posters around here and I was really blown away with the diversity and the research that is going on enabled by Darwin. And that's something that we strive to achieve. As Marianna said, my name is one-inch and Dora, I'm a Program Director at the Office of advanced cyberinfrastructure at the National Science Foundation. And our office. As was said before, funds the Darwin computing system. And the reason we have, and we do have a vast portfolio of computing systems across the country. And the reason we do that is to enable scientists and engineers like yourself to be able to use this infrastructure for your research, right? So that's, that's the vision of our office. And if you're interested in knowing more about what our office does, you're welcome to take a look at our website. Within the NSF website presents, our office vision is to enable or to create this cyberinfrastructure that can enable cutting-edge research, support our researchers, faculty, staff, students to be able to do this research, right? So Darwin is an example of that, but as Marina said, there's never enough computing available for your research. I'm sure that many of you might have other types of needs, computing needs that maybe Darwin is not able to satisfy, or maybe you want to scale bigger or explore other different types of systems. I also encourage you to look at other OAC funded systems around the country. We have a fairly large portfolio of capacity systems. We have the big supercomputer at the Texas advanced computing center, That's our leadership facility. We have a whole array of capacity computing systems. We also have an array of new test beds and new prototype systems if you're interested in that kind of research. So please take a look at that. And I'm very happy to see Darwin within the Access portfolio. I think that's and I was talking to al soirees, who's the Program Officer for Darwin, my colleague. And he asked me to really emphasize this point that we are happy that Darwin is accessible through our access program, which used to be axi exceed, which means that it is serving a much broader community which is great, which is great to see. And we look forward to seeing more of that coming in in the next few years. Other than that, I also wanted to encourage you to take a look at the software portfolio and the data cyberinfrastructure that OAC funds, right? And I know we have some investments that are at Delaware as well at the University of Delaware. So we do find a lot of software development. As part of your research, you might be interested in putting your software and your data out for the entire community to access. And those are some of the activities that we also fund. So please take a look at that too. We also have a new direct rate. Some of you might be aware of that and I was talking to someone from the industry just a little while back. The tip directorate at NSF, which is technology innovation and partnership. This is something that came about last year that was meant to provide a unified umbrella for all the convergent activities and private-public partnerships. And you're working with industry to bring your research into the market or to the society and all of the translational work that you might be interested in. So all of that now is, comes under the NSF nsf tip directorates umbrella. So please take a look at that too. For any possible opportunities. Other than that, I'm looking forward to the rest of the event. I am. I will be talking to a lot of views, so please feel free to come up. And if you have any questions, if you have any suggestions for us as to how we can better serve the community, please let us know. I congratulate you all again for this fantastic group of fantastic event and all of the great research that's going on here. So with that, I'll stop here and I don't know if you want me to take any questions now or as time actually, so but let's think our speaker. Thank you very much. We do have a microphone and back in case people have questions. Okay. I think we can have the microphone. They actually have quite some time for questions. So I was expecting that maybe somebody else would ask that question. So let me, let me be the one asking it. So we have a lot of young researchers here in this room. I wonder if you have any advice just from either from oases point of view or NSF in general, right? Yeah. So for those of you who have just maybe started your career as an academic or are in the industry. Or if you're a student, you might be wondering, how does this fit in your research agenda or your research goals. So I would like to stress that. Anything that needs to be done as part of your research that involves computing. You can look at OAC as a resource for that, right? So we don't directly provide resources, but things like Darwin, we have many such systems and within our portfolio that you can use right at the same time, if, if if you are interested in doing research that directly impacts the cyberinfrastructure. Then we also have funding programs that you can target to get your research funded that way for students. Again, as Marina said, ai is the next, the big thing everyone's talking about. And of course, when you look at these large data-driven machine learning systems that require a large amount of memory intensive systems or GPU enabled devices. Those are the things that you can avail off through, through oases, different programs like access. So that's another thing that you can look at for students. We also have, and this is for faculty as well, is that as part of OBC, we also recognize that the workforce development is an integral part of cyberinfrastructure. So we have several programs and several mechanisms through which we can support training and education of students and cyberinfrastructure professionals. So that is another thing that you can reach out to OAC 4.1 thing is that program directors like myself, we are always happy to answer any questions that you might have. So feel free to email us, call us, reach out to us in any any way you want, and we'll be happy to provide more guidance. Thank you. I think we have a next question. Yeah. I'm wondering whether you can shed shed some light about the center Research Institute, National Research Institute initiative. I guess there's going to be a face for, and certainly many of our colleagues and our collaborators are very much aspired to really look into that opportunity to see how we can further expand this wonderful collaborations and networks that we have built towards AI Institute add in the region. Sure. Thank you. Yeah. So for those of you who might not be aware of this, so the National AI research institutes initiative was one of several federally driven AI programs that came out, out of the White House directive, leading a few years back in building competitiveness and excellence in AI research across the country, right? So the AI Research Institute program at NSF is one of those. And through that we have already funded maybe while they're eating until last year, 18, yeah, institutes across the country that focus on foundational research in AI as well as use user defined research as well. And sometimes both, right? So if you, if you're interested, you can take a look at, you know, we have a sort of a umbrella website. I forgot the URL, but I can share it with later. But that's the place to see what has already been done there. We have made some awards this year as well. And as Kathy mentioned, there are talks of continuing this program. Of course, I'm not at the liberty of saying anything about when this will be. But of course, ai continues to be at a very integral part of NSF's vision. So all I can say is stay tuned. But what I would also say is that if you look at the last solicitation, we outlined several thrust areas, right? So you can expect something like that. But in terms of what you can do right now, is exactly what you mentioned is to, is to continue this collaboration, multi disciplinary collaboration within the university statewide as well as, as well as other partners, right? So what we look for our teams that are not only geographically spread, but also in terms of the disciplines, as well as the types of use cases and foundational topics that you want to address. So I think at this point, I would encourage you to continue this will this collaboration so that when time comes, you have something ready to put forward. Thank you. I think we have another question. So unfortunately, funding is not infinite and like computational resources, you never have enough. All right? Similarly, the University of Delaware is not the only group gaining money. Cabinet, Darwin cabinets is not the only group getting money from NSF. My question is, how would you recommend that we? Your work, applications or communications to best both make sure we get the most out of what we have and, and get as many resources as is constructed. Right? So, great question by the way. So first of all, in terms of the resources that are out there, we have, as I said, a lot of computing and other kinds of resources that are available to you that you can get access to through programs like access. So I would strongly encourage you to see what's there. Make sure you are you are able to use what is out there. At the same time, if you have ideas about how we can expand on this infrastructure, expand on this ecosystem, then we do have ways to support that. So, you know, through, through different funding mechanisms as well as other, other ways of collaborating within this larger network. So please reach out to us and we can help you with that part as well. In terms of how you can make sure that you are reaching out. I think one thing is that what OAC always encourages is this notion of sustainability, right? We want our investments to be sustained beyond the ears of funding that we provide. As you said, funding is not finite, is not in finite. So we do want to make sure that our investment dollars go as far as they can. And so we have programs like the cyberinfrastructure for sustains scientific impact, which is our software and data infrastructure program, right? Wherever focus is for researchers like you to build software and build these data archives and data repositories and then put it out there. And I think that's the important part. We focus on this community engagement. Bringing the community, letting them know what you have, asking them what their needs are and developing things that support them, right. And we have funding programs to do things like that. Mariana will be your last CSI grant is example of that as well. So this notion of engaging the community and making them aware, and that would go a long way to making sure that what you produce a sustained. And certainly at OISE, that's something that we encourage a lot. And as I said, we do have different ways of making sure that you have the support to build such sustainable technologies? Yes. He'll bring you a microphone Just a second. Yeah. So one way of leveraging that initial investment by NSF of the government would be through obviously through SBIR and STTR grants. And that way you're really using that initial investment is a catalyst and it's in everybody's interests to really grow up beyond whatever the government can invest. So what are the initiatives there? So I totally agree. And I think that is where the new directory that I mentioned, tip, which is the technology innovation partnership, that is the, the vision or the mission of that particular direct rate is how to take the research that is done within the NSF umbrella or the larger federal funding or state funding perspective, how to take that and bring it to the community in a way that they can use it and maybe also benefit the people who are developing it, right? So, so we have several programs, like some of the older programs have been folded within the tip umbrella. So e.g. SBIR or I corpse or things like convergent accelerators, right? So, so I would encourage all of you to take a look at what tip has been doing right now In in bringing or making that happen, right then there are several new programs as well. I think there was one called ART that came out a couple of weeks back that is focused on helping universities create programs that will allow students to bring, allow the students to think about bringing their research to the market, right? So how do you encourage startup culture within universities? So those are things that are now possible through tips. So I'd encourage you to look at that as well. Thank you. Other questions? I actually have a question. If you live in each whereas the hardware is rapidly developing, our Farber facility was better than mils Kevin yes, was better than the Farber and Darwin facility are still also different now from Calvinists. Already when we were talking about the next-generation facility is going to be perhaps different types of processors, etc. So what's in the surf vision here where we decided we really have to start building a new facility as opposed to increasingly old facility. And what are your recommendations here? Yeah, that's a great point. I think. From the OSC perspective, we do like to provide a stable set of resources for the community, which means that we do want to continue with what we have and maybe keep reinforcing those. But at the same time, we also recognize that the landscape is changing continuously, right? We have new technology that can enable new signs. And so there's a balance that we tried to go for and that's where we have a portfolio of this capacity systems that we think of as these stable resources that we are always Providing to the community. But at the same time we have this prototype test beds and prototype systems, new processors and machines that have, that are built on these new technologies that allow researchers to also explore these, right? And our vision is that with that, we can add some point transition some of the more successful systems to more production or capacity capability. So we're, we're always trying to straddle this balance between stability as well as innovation. And it's not always easy, but yes, that is certainly in our mind. At the same time, always see an NSF course. The directory that we see is within the science directorate, we're always looking for funding new ideas of new cyberinfrastructure technologies as well. We see itself has a core or participants in the syscall research where we do fund research into the cyberinfrastructure. So if, if that is something that you're interested in, please take a look at that too. But yeah, I totally agree that there's a balance that we need to maintain between stability and new systems. Thank you. Other questions? I guess we still okay. Let's cover this perhaps last question. And I think of here is, here is here. I believe here is there any question? And then we'll go to the next section. Hi, Thank you. So I had a question. You were talking a lot about sustainability in research and innovation and helping out movies and things like that. So my question is, how can we discuss sustainability in research when many of our population isn't being funded. But that research, so e.g. um, of all the money that the NSF grants to all research-based institutions, less than 9% of that funding goes towards HBCUs. So we have an entire community of black and brown and Latino students being underfunded for the research. So how are we expected to compete on this world stage, right? That's a great question. And from NSF side, we totally recognize those issues. And we have several programs that we add, some long-standing programs to encourage but more participation of all the communities and not focusing on a certain geographical areas or certain communities. And so e.g. I. Would like to mention things like our expand AI program, which came out maybe two months back or a month back, which allows smaller Universities, HBCUs, community colleges, to build their AI capabilities so you can get funding from NSF to expand on your capabilities that way. At the same time, even in size, we have different programs where smaller institutions, minority serving institutions can come in and, you know, build more capabilities in the computer science side of things. So we do have programs like that, but I completely agree that we still need to do more and it needs to do a better job. So I look forward to any suggestions that this community has in that respect. But at the same time, I would also ask you to take a look at some of the programs that I mentioned and I'll be happy to provide more information to see where NSF is going in that direction. But NSF has always been. One of the parts of NSF's vision has been to make sure that our investments go through the entire country and all the communities, including any smaller universities and colleges. Thank you so much. Thank our speaker for answering all those questions. Thank you. And also I would like to stay in the Data Science Institute and as director Kathy Wu, who is going to be out next year for the support of the symposium and the high-performance computing and delaware. Thank you so much. Okay. Well, that's again, our conference chair, Dr. Slough ANOVA, I'll provost Carson and Dr. Chen dollar for really inspiring opening remarks and provide a lot of wonderful information for our community. So as the director of the Data Science Institute, I would like to encourage and invite your engagement with the institute. So we have a table outside and I like you to sign up for our students, graduate, undergraduate students and post out to sign up for joining the Data Science Students Association, which is just established last week. You are going to be the first members to join them. Affiliated faculty for our UT faculty, HIM faculty for faculty from our partner institutions welcome. And industrial affiliate. If you are in industry, government and non-profit organizations, we would love to engage you as our industry affiliate program. And we also have a data science mainList that I encourage you to subscribe. You will get information, dissemination from all different kinds of activities, seminars, and events.
2023 Darwin Symposium_Welcome & Opening Remarks
From Robert Diiorio February 27, 2023
8 plays
8
0 comments
0
You unliked the media.