🌟 Artificial Intelligence meets Nanoscience with Andy Sode Anker: Episode 196 of Under the Microscope 🔬

What to Expect:

In this episode, Andy Sode Anker delves into his innovative research on using artificial intelligence in nanomaterials research. Andy shares his journey from studying chemistry in Denmark to working on AI-driven data analysis at the University of Copenhagen. He discusses the potential of AI to revolutionize the field of nanoscience.

About the Guest:

🌟 Key Takeaways from This Episode:

• AI in Nanoscience: Using artificial intelligence for data analysis in nanomaterials research.
• Career Journey: From studying chemistry in Denmark to integrating AI in nanoscience research.
• Favorite Experiment: Compressing high-dimensional data to visualize similarities in chemical structures.

🔬 In This Episode, We Cover:

Andy’s Research :

Andy’s research focuses on using artificial intelligence to analyze data from nanomaterials research. By developing AI tools, he aims to handle the large amounts of data generated in synchrotron experiments and uncover new insights into the properties of nanomaterials.

Andy’s Career Journey :

Andy’s academic journey began with a Bachelor’s in Chemistry in Denmark. He pursued his passion for data analysis, leading him to integrate artificial intelligence into his research on nanomaterials. His diverse experiences have enriched his research perspectives and expertise.

Andy’s Favourite Research Experiment :

Andy’s favorite experiment involves compressing high-dimensional data to visualize similarities in chemical structures. By using AI to reduce the complexity of the data, he can better understand the relationships between different nanomaterials and their properties.

Life as a Scientist: Beyond the Lab

Andy values the collaborative nature of scientific research and enjoys engaging with the global scientific community. He is passionate about teaching and mentoring the next generation of scientists and values the opportunity to work at the intersection of AI and nanoscience.

Andy’s 3 Wishes

1. Increased funding for research: Andy wishes for more financial support to advance innovative research projects.
2. Access to advanced computing resources: He hopes for access to the most powerful computers to further his AI research.
3. Improved public understanding of scientific research: Andy emphasizes the importance of public awareness and support for scientific advancements.

Andy’s Time on @RealSci_Nano:

Andy will be taking over the RealSci_Nano Twitter account to share his research on using AI in nanomaterials research. Followers can expect to learn about the innovative techniques and tools his work focuses on, as well as insights into the future of AI in science.

Transcript

[00:00:00] Hi, so just finished recording a podcast with Andy, who is a PhD student at University of Copenhagen, Department of Chemistry, and his research is really cool. Artificial intelligence meets nanoscience or nanomaterials. And he basically uses artificial intelligence, uh, together, of course, with his team to, uh, to, to do the data analysis and basically trying to make, uh, future materials, functional materials for the future.

And yeah, his journey has been quite fascinating. I think it’s the first time we have, uh, a guest who is working, uh, On these two giant fields like a merge of these two giant fields artificial intelligence and nanoscience So can’t wait for you to check out the episode[00:01:00] 

Hi everyone, i’m prananti your host of under the microscope podcast and today we have with us andy so Anger. I am sure I’m not pronouncing the name properly. Who is a PhD student at the department of chemistry at the university of Copenhagen, not in the group, which got the Nobel prize for chemistry, the click chemistry this year, but, uh, in the same department and in the same university in Denmark.

So hi, Andy, welcome to under the microscope. How are you? Thank you. How good. Thanks. Thanks for inviting me. Look forward for this. Happy to have you. I’m really, really happy to have you. One of our last curators is also from, uh, your department. So, and she gave us a sneak peek into the Nobel prize. It’s a, uh, winning ceremony at the University of Copenhagen and it was quite heartwarming to see that, um, and it just coincidentally happened that, uh, uh, in the [00:02:00] week that she’s, she was taking over the account that the Nobel Prize in Chemistry was, uh, announced.

And then, of course. Uh, we have, uh, part of it, I think one half of it is going to uni Copenhagen, uh, one third, one third. Okay. Yes. One third. Yes. It’s a third shared among three. Yeah. But it was quite special day. So, yeah, I was, I was there as well. We were actually just eating lunch, like normally. And then, um, there was another PhD student, he, uh, he came in running to the lunchroom and he just yelled, have you seen it?

Have you seen it? Like Martin, he won.

Yeah. Then we, so, um, some of us walked to. So his office more than me less office and saw what was going on and he was just busy answering the phone and there’s so many people and I think there was not a lot [00:03:00] done of work that day for the entire department. It’s just one big party. Celebration is part of work.

Okay. Yes. But it was quite fascinating and I talked with one of his PhD students that talked about like, yes. Some of the work that had been done in the group, and it’s fascinating. Yeah, it is. Oh, I’m so happy. Awesome. Let’s talk about your research. Andy, let’s do that. What do you do? What is your research at University of Copenhagen at the chemistry department?

It’s not click chemistry. What is it? No. So I would say it’s, uh, it’s in materials chemistry. So our goal is to synthesize some new materials that has some functional properties. Um, we normally do that in the lab and then we take them to synchrotrons, which is like [00:04:00] these big rings with electrons moving around where you can get a lot of energy from them.

So you can generate x rays and we shoot that into our. Chemistry and follow direction while it’s happening, um, using them. And then when we get the data, like, obviously we get a lot of data when we have such big machines, then I analyze it with artificial intelligence. And for me, that is really the core of my research is like, how can we make new and better tools to analyze this data?

Okay, wow, that’s so it’s basically artificial intelligence and nanoscience and the material science, like all the expertise coming together, um, to, to make materials for the future or functional materials over the future. Whatever the application might be, is that fair to say? Yeah, I think that’s, that’s [00:05:00] very fair to say.

And I think that’s also what makes it interesting that nobody can do this kind of research alone, right? You need someone from all fields to come together and someone that knows about the applications, someone that knows about the synchrotrons, the experiments you do, and some that know about artificial intelligence.

Yeah. Yeah. Also, I’ve always wondered, uh, so it’s good that we met you and we are talking right now that at the synchrotrons at the beam line, beam time that you get like whatever, four or five days or one week, or you’re working 24 seven. So you’re gathering humongous amount of data. Uh, and in order to, uh, do the data analysis, uh, and also storing the data.

It’s you must need like supercomputers and, uh, some really sophisticated, uh, tools or softwares. Um, so it’s really interesting to know that, uh, finally, at least for me, uh, I’m getting to know [00:06:00] this, that there you guys are using machine learning or artificial intelligence to, uh, go through this data that you collect at the beamlines.

Um, So yeah, that’s quite interesting. I always wondered how that works. Um, so that’s, uh, that’s really cool. So, um, Andy, tell me your research is super interesting. Uh, so how did you end up, uh, doing a PhD at University of Copenhagen, um, with artificial intelligence and nanoscience? Like that’s a very like, uh, This is the first time I’m hearing about it.

So tell me, tell me about your journey. Yes. I think like my journey is, is both like super boring and super exciting because I, I started the first year of my bachelor, like doing some lab work in, um, in, uh, in a materials chemistry group. So the group I’m still in, um, where we also, [00:07:00] where we do a lot of nanomaterials.

And, um, in the beginning, I was, um, Like going a lot in the lab myself, um, to, to do the synthesis part of it, but I found out that I’m just too clumsy to go to the lab. I’ve got to be honest. And, um, and I’m also not patient enough. Like you, you need a lot of patients to go to lab. So, um, so I really appreciate those that like have those abilities to go to the lab and, and, um, And this is, um, so actually I kind of switch over to do more of the experiments.

So we went to synchrotrons and I just found out that it’s super fascinating, um, like getting all that data. And I remember my first project, I just. Got like, it was, uh, between 50, 000 and a hundred thousand data sets, like straight away. And that takes a long time to analyze. [00:08:00] Um, so, uh, so, um, I, I kind of began to, to program, uh, more than I already did.

And, um, did some more programming courses. And then I found out that I think like through programming, I became more and more aware of Uh, AI and what it could do, and then I begin to use it and I found out that it’s amazing like how much it could do for my research. Um, and then I started getting collaborators.

I have a really strong collaborator at the Department of Computer Science in Copenhagen and, um, So we have, uh, been working together for the last three, four years. Mm-Hmm. . Um, so fing, I’m from, from my, my background is very much from like nanoscience and materials chemistry. And then I moved more and more into, um, ma Yeah.

Like programming and AI and, [00:09:00] and now that’s, that’s most of what I’m doing. See if I can make some new tools to Mm-Hmm. to analyze our data and, Mm-Hmm. . Mm-Hmm. . Yeah. Wow. That’s, that’s really cool. Yeah, I think that’s how it’s gonna be in the future or even now. That’s, that’s, uh, that’s usually like, that’s typically, uh, the nature of our work that we started.

Probably the bachelor’s or master’s was, I don’t know, in nanoscience or, uh, chemistry or computer science or whatever. And then because, you know, These, uh, giant fields have to merge or have to cross lines at some points. Uh, there are lots of people in the, at these cross section, cross sections, cross sections, uh, having these unique journeys like you’ve had, um.

That sounds really cool. And why leave [00:10:00] Copenhagen if all the amazing talent from around the world is coming to you? I mean, there’s no point in leaving, right? Yeah. But, but there is, there is still like after finishing your PhD, you can always, depending on what your plans are, you can, uh, you can move. Uh, so are you from Copenhagen?

No, I’m, I’m from a small island called Bornholm, so it’s um, it’s a little bit south from Sweden. It’s closer to Sweden than the rest of Denmark, um, so, so people that, they also like pick a little bit on me because I like, I have a Swedish kind of dialect, um, but, um, the eastern part of Denmark. Nice. And you spoke about synchrotrons.

Do you have a favorite one for whatever reasons? You don’t have to disclose your reasons. Of course we have a favorite, but you don’t [00:11:00] have, I’m just curious. I think it would be, uh, ESF in Grenoble, uh, in France, because it’s just so beautiful there. Yeah. Yeah. And it’s, it’s in a valley with the mountains around it and it’s just breathtaking, right?

Yeah, yeah, no, that’s an amazing place Fair enough And they also have the cheese fondue that I lost. Oh my God. Yes. Now it’s the time for cheese fondue. Winter is coming. Yeah. You haven’t been times coming up anytime soon. Actually, not right now. I’ve just been on two and now a little bit more relaxed. Yeah.

Drop me a line next time when you have a beam time at Grenoble and then I can, uh, join you guys as like the, uh, to record videos and interview you all. Um, what is that?

Um, no, I’m kidding. Uh, or maybe half kidding. Uh, but, uh, [00:12:00] no, that sounds, that sounds, uh, really cool. I mean, you’ve had quite a unusual journey from like starting with nanoscience and now doing, you’re still working close to nanoscience or nanomaterials. Uh, but more of like. Not the conventional, uh, nano scientists that we have.

So I’m curious about your different kinds of experiments, because if you say you’re self proclaimed clumsy, impatient person, um, uh, so What, what is that one experiment or research project that comes to your mind when I ask you that the most quirky or fun or the one that you’re most proud of? Uh, do you have one and can you explain it to us in super simple words in the section we call in other words?

Yes, yes, I think, um, so there was this kind of the first [00:13:00] project that I got involved in where we use like where we used AI, but, um, and I think I’m quite process. I’m quite proud of the entire process of like learning this AI and then using it on some real data. Um, in itself, um, but also, so what we did there was that we took like a lot of data that was in 3000 dimensions.

So obviously that’s something that you would normally have a difficult time like plotting and showing, like, see what is the similarity of. Um, so all this data was of chemical structures and then also scattering data. Um, and then we could pass it through some AI that compressed it into two dimensions.

[00:14:00] So from 3000 dimensions to two dimensions. Yes, and I think that was quite amazing because that means now you can plot it you just make two axes and you plot it and then you can actually you can get similarities between structures and how they are different and so on. So one of those things that we could do was take FCC and ACP structures.

Which is um, so you can say an HCP structure is a A B A B layer called A B A B and FCC is A B C A B C. Um, and then normally what you have. So what does FCC stand for, for people who are not aware of these terms? It’s base centered cubic. Base centered cubic, yes, that’s the name of the crystal structure. Uh, and the HCP is hexagonal, uh, closed packed.

Right, hexagonal, closed, packed. So HCP. So FCC and HCP, these are the [00:15:00] two terms. Okay, sorry, continue. Yeah, no, that’s great. And then, like, normally, if you have something in between, so that can be that first you have, like, A, B, A, but then you get a C or something. So you have something that is not either FCC or HCP, then you normally just call it a stacking fault.

And we, we don’t know so much about it. Uh, but, and, and it is quite difficult to understand them because like 3000 dimensions, right? But when you compress it to two dimensions in this space, you can actually see that the machine also learns that this is something in between. FCC and ACP and you can begin to understand or this stacking fault is different from this one up here and you can you can begin to see how Much is it similar to to other chemical structures?

And I think that’s it’s quite beautiful where math is like meeting chemistry and you [00:16:00] can begin to Yeah, this sounds so cool. I do you have like, I don’t know. I mean, of course, the two dimensional picture picture or the plot or that graph, uh, you must have, but is there a way to visualize these like 3000 dimensions, uh, sort of a data?

So basically what I’m looking for is before the analysis, like the raw data and after the analysis, what. Uh, came out of it, uh, after passing through your amazing, uh, machine learn money, machine learning, uh, do you have, is that, is that, uh, available? Like, so you, so you’re kind of thinking if we didn’t have AI, if there would be any other method to do this.

Yeah. Yeah. Yeah. That, that’s also interesting. Yes. Uh, what I was asking though, was. Any way to visualize, like showing basically like without [00:17:00] AI, um, this data would look like a mess. So it would, because you would need to need to visualize so high dimensions. I have no idea how to do it. 3000 dimensions. Yes.

That’s yes. Like normally I would have no idea how to visualize that, but when you pass it through to. I mean, you get it down to two, then it’s just a normal plot as, as we know it. And probably a rookie question, probably a stupid question. Um, 3000 dimensions. So when you say 3000, so I know the X, Y, Z, and then time can be a dimension.

So what are these 3000 dimensions that we are talking about here? Um, So in this case, it’s, it’s a little bit equivalent to every data point you have and the data sets. Um, so if you take just one data sets. [00:18:00] Then, then you can just plot it as it is in a, so you have like an X dimension and a Y, and then you have 3000 data points here.

But then when you have a lot of data sets, like, uh, then let’s say you have thousand data set with these three thousands, then it’s beginning to be difficult to see similarities and differences. Then you, then you take all these three thousands. Then it becomes 3000 data points or 3000 dimensions. You compress it to two by the use of AI.

That means now you can plot that just as you can visualize it just with a point for each structure or data set. Okay. All right. And then when you say dimensions, uh, the 3000 dimensions, we are not talking about. Who is it? I was watching some marvel movie or [00:19:00] something Oh, yeah, okay, so you could get confused by like is it the the dimension?

Yeah It’s not like time and we don’t have a multi word Exactly spider man movie which is like based in europe like venice and wherever and there is the bad guys like i’m from You Not from this dimension. I’m from a different dimension and version or it’s I don’t know Eight six eight. Oh, I don’t remember the exact number.

So that’s not the dimension that we’re talking about. No, no, no no, that’s that’s not how I think about but it’s it’s uh I’ve never thought about that. Maybe I should use it in some talks. Yeah Okay, so these are not the dimensions from avengers these are the dimensions In, not in real, well they are in reality, um, it’s, it’s not the multiverse, [00:20:00] uh, it has nothing to do with astrophysics or anything, and the universe is this tiny universe at the nanoscale that we are looking at, the atoms and the, the, the defects and, you know.

Yeah. Yeah. Um, okay. Okay. That’s Oh, my God. That is so cool. Please, please, please tell our followers more about this with pictures and lip. Please explain in as many words or as many tweets as possible. When you’re tweeting from the real scientist, not on Twitter account. I think this is super fascinating.

So it’s kind of like I’m getting the feeling and the that you love the research aspect of being a scientist. Uh, but what else do you like about being a scientist? Uh, yeah, I think I very much like that you have a lot of discussions and conversations with colleagues and peers and, um, yeah, so you, so in research, there is a lot of time [00:21:00] dedicated to like.

discussing science. And I think that motivates me quite a lot. Um, but at the same time, we also have a lot of time to just reflect on our projects alone. So I think that is a quite strong combination that you can do both. Um, I think there’s new aspects in my research that pops up in both settings, um, but I kind of need them together.

So when I have discussed a lot with my peers, I need to reflect on it afterwards on my own. And it’s, it’s quite normal that I take some of my day where I just go for a walk and, um, reflect on my research and, and, or if I have some specific. Problems that I have to solve. And, um, I don’t think it’s all jobs in the, in the world where you have that, uh, opportunity.

Yeah, [00:22:00] yeah, no, you’re right. You’re absolutely. There is like teamwork or team discussion, stimulating, brainstorming or stimulating conversations, but at the same time you can be independent and you can just, uh, process your thoughts. Process your theory, so to say, uh, by being independent or by being by yourself.

Um, yeah, that’s, you’re right. I don’t think a lot of jobs in the world allow you to have like this balance of the two. Uh, definitely, definitely. Sometimes I feel like I’m, I’m kind of cheating and not doing work because I just go for 20 minutes walk and like reflect on a problem. But. It is much better value of the time than if I didn’t do it.

Yeah, exactly. It’s and it’s necessary. That’s how you work. Your work might be taking a walk and doing the thinking Uh while you’re walking and that’s uh, that’s [00:23:00] okay that That’s part of the job. You have to work for your job. Yeah.

Um, I see. So Andy, uh, other than like, I mean, it sounds to me that you have had amazing research experience so far. I mean, uh, having these amazing talents around you, like surrounded by these experts in their fields. Um, if you, however, if you had three wishes to improve your research experience, what would those be?

What would you ask for? And I’m not promising anything here. Okay. Well, that would have been great. Let’s see No, I think um, I think when you’re in in academia and working with uh, like Computers, then you would always wish for more money, more computer power, um, because it’s, it’s just a reality that some of the big companies like meter and Google and, [00:24:00] uh, they, they just have money to do research that we can’t in academia.

Um, and yeah, and it would not be enough with a few extra million science thing. It’s like really large. Computers, you need to do the stuff that they’re doing. Um, so of course that would have been a dream to do what I do now, but with the access to the largest computers in the world, how much does it cost?

Tell me, tell me a number. Yeah, I don’t know. I think we, it’s, uh, It’s probably, uh, like it’s in the billions, but, uh, how many, I don’t know. Okay. So getting access or getting your hands on these, uh, amazing, uh, super computers or these giant, uh, computers. Um, okay. That’s your first question. I think that would be a dream.

I’m not even [00:25:00] sure what to use them for yet, but it would be. You will find ways. Like, yeah. Why, why stop at 3000 dimensions? Go to 3 million dimensions. Hello.

Yeah. Um, yeah, I think the other thing would be to, um, to, to start my own research group. Um, I, I think I had some amazing opportunities to, uh, choose what I’m working with, uh, in the group that I’m now and got a lot of freedom to move into. Artificial intelligence, um, and that also like, yeah, now I appreciate it so much that I would love to have my own research group And be able to do Do my own research, yeah That makes sense.

Yeah, and Research then you can either get the Hopefully get the funding for these giant, [00:26:00] like amazing computers, or at least work with the companies or groups who have these kind of, uh, tools. Okay, research group is second, third and last. Andy, tell us. Yeah, I think the last one is probably also a difficult one.

So, um, So I told you that I’m impatient, um, and, uh, yeah, that’s not the best thing in academia because everything is so slow. I would, I would love a faster feedback loop. Like when you, when you come up with a new idea, do some, some. Experiments, analyze the data and so on, like, I would like that to happen much faster.

Um, like today you need to come up with an idea. If you go to singletons to do an experiment, you need to write a proposal. It’s might take you like a year before you can do the experiment. If you get the time branded, [00:27:00] then you do the experiment. You have to analyze the data, write the paper. It can take many years for a project.

Um, yeah. And by then someone else has already done what you wanted to do. Yeah, yeah, hopefully not, but it’s, um, yeah, it just takes a long time. It would have been amazing if we could cut down on that, uh, time. Timing. I think it would be even more fun if it’s only so the time between ideation and the results or the data analysis, so to say, to shorten it from years or months to, uh, Maybe a few.

Okay. If you’re wishing for it, then maybe a few days. Yeah, like the best thing would be instantly, right? Yeah, that would, that would be amazing. As fast as possible. Um, yeah, that’s one of the good thing about research [00:28:00] in AI, because you, you’re not limited to, um, to going to a synchrotron or stuff like that.

So you can maybe simulate some data and work on that and thereby, um, the research goes away faster, but then again, like you can do very cool experiments at synchrotrons that, um, That you can’t really simulate. Yeah, exactly. So, um, Yeah. That makes sense. Well, all three of your wishes, Andy, I think, yeah, they seem, they seem, uh, too far, but I hope that they are not too far in the future, especially getting your research group.

And once you have your own research group, then maybe not for every idea, but maybe some of the ideas you can have this accelerated process of, uh, Uh, the fast, uh, Turnaround time, uh, on [00:29:00] like validating the different ideas. Um, I would, I wouldn’t say your wishes are completely unrealistic. I could, I could see that happening in a few years time.

Why not? Uh, very, very realistic. Um, awesome. Andy, this has been wonderful. And, um, what can the followers expect in the week that you are tweeting from the real scientist nano account? Uh, Give us like a sneak peek into what’s coming. Yeah. So of course they can expect a lot about my research and what I’m doing.

Um, but actually that lucky that I traveled to New Orleans to a very big machine learning, uh, AI conference that week. So, um, so obviously I will. Yeah, I will tweet live from New Orleans, uh, and, uh, yeah, show some, some of the cool stuff that is happening there. And then the end of the conference, I’m [00:30:00] attending a workshop that is about, uh, AI for materials discovery.

Oh, wow. Yeah. Now we’ll present a poster there as well. Um, so, uh, yeah, so that would be very interesting. So that would kind of be the end of the week. Uh, uh, what is it called? And I guess celebration of the, yeah, that is so cool. That is awesome. Awesome. I, I. I personally am looking forward to your tweets and your time at the conference and also the week leading to the conference because it all sounds so cool, so fascinating, your research, um, and the ideas that you have.

So this has been amazing, Andy. Thank you very much for speaking with me and really, really excited to have you on Real Scientist Nano. Thank you. Thank you. I’m looking forward for the week.for listening. To know more about us, do visit our website, [00:31:00] their science talk. com. And do consider giving us a review or a rating or follow, depending on wherever you’re consuming this content. Thank you very much.

Podcast title : Artificial Intelligence meets Nanoscience