Brian mcut is a principal R&D engineer at the naval nuclear lab and he focuses on developing and delivering emerging digital technology in support of the US Naval nuclear propulsion program he currently leads the quantum technology integration efforts at NL as well as future technology development in super Computing and AI enabled scientific software today he's here with us to talk about the vum qsl collaboration with NL and the project we are hosting for this summer program that is using quantum AI for climate Earth and environmental simulations over to you Brian all right thanks foron all right so I'm back after uh you heard my talk last week a little bit about what nnl does but the project that uh we contributed to omniums program this year actually doesn't have a lot to do with what we do at nnl at all it's actually a very broad topic that I think would be interesting to a lot of the um a lot of the audience on in the program this summer and that's Quantum and AI for Earth system modeling climate modeling and things related to that and this project is a very uh is going to be more open-ended and more research focused um than some of the other ones and so if this is something that uh you might be interested in maybe going into a research field or getting trying to get a feel for what it's like to actually research something where there's not a clear answer to it this might be something um that you're interested in pursuing further so uh the main background here is that the Earth and it's all of its systems are some of the most challenging systems that we we have to simulate um there's very highly complex physics involved you know ranging from the formation of water droplets and ice crystals and how light interacts with them um all the way up to the full Globe scale of all these interacting systems where you have these nonlinear equations that are coupled with each other they feed back on each other um and they require some of the biggest supercomputers on Earth in order to simulate uh but at the same time thinking about you know what where are we going in the future with our Energy System um in our environment a lot of these challenges require intensive modeling simulation and data analysis of all these different Earth Systems and so from that standpoint it's possible that Quantum and AI are going to be enabling Technologies for this um so when you're doing a very large simulation uh you could either use these things to either get faster results get more accurate predictions or perhaps do new types of calculations that aren't possible right now and so the purpose here is to try and research in the literature and either um you know document what folks are working on right now out there like kind of like what is the frontier of research in this area and also come up with ideas of your own of figuring out how to extend the research that's being done today into the future so the project then um is really to show how AI Quantum and or even Quantum inspired methods can be applied to Scientific Computing in this um Earth and climate atmospheric science context so this is uh could be things like simulation of weather climate other Earth Systems so all the way down from those atomic scales so like atmospheric Trace gases and things like that all the way up to the global scale so figuring out how can we use AI or Quantum Computing to accelerate the types of calculations that are being done when you're simulating uh the weather and the climate um another potential option would be to analyze and perform machine learning on weather climate or other environmental data uh there's large open open source data sets out there on you know things related to um whether it's cloud formations precipitation um severe storms there's all sorts of data sets out there that you can sort of grab and you know play around with and try different machine learning techniques on to see okay can I use this to do uh new types of predictive um predictive analysis that wasn't possible before so that's potentially One Direction um another direction is maybe not related to you know simulating the Earth itself but looking at okay some of these sustainable technologies that are being talked about now um either related to energy materials um batteries photov voltaics the power grid itself agriculture fertilizers are there applications of quantum and Ai and these things as well since they are also very much related to the topic and so the goal then is to try and you know uh look into the research find report on and try and reproduce some of these examples that are out there um to sort of get a feel for how does this stuff really work in practice or even if you're feeling particularly ambitious try and come up with your own ideas to get started started on as potential future research directions but this is very meant to be very open-ended so research the things that interest you the most not trying to be constraining in a particular thing that you want to um you want to focus on this is really meant to be a very broad project that appeals to a wide range of people that are interested in this topic so as you do this project then the questions you want to be able to answer at the end of this um and then in providing your submission would then be why is this problem important so what made you pick it what what was the you know what spoke to you about the particular topic within this area that that you picked and so you'll want to say like you know kind of give an explanation of why um this problem is worth solving in the first place um and then based off the research that you you've done try and say something about how is this problem being addressed currently today are we using giant supercomputers to do it um are we not doing it all because it's too hard um is it something where there's not a lot of research out there at all so nobody's really looked at it um try and say something about you know sort of what is the state-ofthe-art and then to take it a step further try and say something about how would Quantum Computing or AI improve the way that this problem is solved so if a quantum computer could solve a particular um equation faster than is being done right now that's something that we'd want to you know you'd want to understand as this project same thing if you can use an AI method to either do a better type of calculation or discover better types of materials that's another thing that might be worth talking about in your submission um and then finally and this is the um you obviously a very open-ended part is how would you implement an example app using this solution um we only have about a month so it's kind of hard to write a code depending on what sort of compute resources you have access to here but if you know if you're not able to actually write a code um you know say something about how if you're developing an application how would you actually implement this what would you do um in line with the way the problem is currently being solved today to then extend it and apply AI or Quantum to that problem um and if there are let's say you find some research and there is an example code out there or a software tool out there um like vardon said earlier you know using those tools you know site the fact that that tool exists and you used it but try and do some sort of newer analysis beyond what was just done in the research that you found so try and extend that work um and so the purpose of all this then you sort of the learning goals here are really to just gain this experience navigating the research literature I'm just sort of watching the messages go by over the past couple of weeks here a lot of people are interested in doing research um and so if you haven't done that before it's kind of a learning curve to get over like okay how do I how do I find the things I'm looking for in the first place how do I know what's good quality research what's not good quality research being able to you sort of understand like what it's like to navigate the research is a very important skill to have if you plan on going into a research field um and then more specifically you know learning about how scientific simulations and machine learning are being done in practice um there's a lot of um you know sort of a lot of stuff in the popular media today about you know how AI or Quantum could be good for types of different types of simulations but it often gets glossed over of like what it actually means to do a simulation right now and so part of this is going to be um to learn how that's being done in practice and try to get a little bit more um I guess a little bit more grounding in what are we having to work with today and where could we be going in the future um and then sort of the core of this is then discover how Quantum Computing and AI can or maybe can't improve scientific models it could very well be the case that um these Technologies don't provide an advantage to a particular application but as with most things in research knowing the negative is sometimes just as important as knowing the positive because it it can give you information on where you should be focusing your efforts going into the future um and then finally this is also one that um comes up a lot you know in the research field is understanding how to turn ideas and math into code into a practical application a lot of the time we see um you know equations written down or maybe somebody has an idea of how a particular um a particular technology could help with simulations but then actually you know doing the work and turning that into code that's a skill that is very important to have um really not even in research but in any any um technological Endeavor so some examples and this is by no means exhaustive you can certainly come up with your own or um you feel free to ask questions about this um but one could be show how historical climate data um and simulation outputs could create a fast running AI model of the earth's climate that runs on U very limited Computing resources so like a laptop or a PC uh this is a very hot topic these days and a few of the um Hardware vendors are even starting to put some of these out right now uh basically like they call it an emulator or a surrogate model of a very large coupled system of equations um propose or develop a Quantum algorithm for simulating batteries or photovoltaic materials um this is one of the many cited applications of quantum Computing in the near term so you know maybe consider putting into practice how would you do this for um a new type of material with an energy application um modeling the interaction of light with water droplets and clouds using Quantum or AI um one of the you know the common things that comes up a lot in sort of the modeling and simulation Community is the uncertainty That clouds introduce um into um sort of these predictive models of of the the atmosphere and so if if there's a better way of understanding where those uncertainties are and perhaps addressing them with these technologies that would be something that an interesting project could be um on the side of quantum machine learning discover if quantum machine learning algorithm performs or outperforms classical machine learning and say say identifying severe storms so like in this picture here of a radar image of a tornado right you can use machine learning to identify these on radar to provide early warning against those so does a Quantum machine learning algorithm provide any benefit to something like that um and then another one maybe perhaps more practical is compare contrast and quantify the energy consumption of different types of quantum Hardware so do superc conducting cubits draw more electricity than say neutral atom cubits in a a Quantum hardware setup that that might be an interesting thing to know about as more and more data centers look to integrate Quantum Computing in the future um some other just general tips you know looking at this um you know be bold here right this is a very open-ended project this is really your opportunity to you know pick something that interests you and try and go you know take it as far as you can um you know and be curious you know try and you know try and uncover as much as you can about your topic and really try and understand what um like what makes this a hard problem and you know maybe how the technology can um can address that similarly be creative right so try and try and push the boundary of where the research is right so if you see if you find stuff in your research that you you're sort of confident where the Frontiers are try and take it a step further and um you do as much as you can to extend that out into the future um and ultimately you know as Von talked about with the code of conduct and all that you know be yourself don't just take the project description and drop it in the chat GPT and try and get an answer out of that don't that that you don't learn anything doing that um but that's not to say don't use like generative AI tools to help with understanding right you a very helpful thing to do in research um especially with complicated um topics or terminology that you might not be familiar with is put it into something like chat GPT and have it re-explain it in simpler language that's certainly a absolutely valid way of using that tool um and similarly with you know follow the rest of the code of cont very important to you know sort of make sure you're representing yourself in your work uh the right way but ultimately have fun this is something that's meant to you know sort of enrich your learning get a feel for what it's like to do research out there um so really you know try and have fun with this project because you if you knew what you were doing it wouldn't be called research so um these are a very limited set of resources but I could certainly help people find more if um people have questions about a specific topic um but with that uh thank you um hope to hear what people find out