Big Compute - How COVID-19 Spreads Indoors

Episode Date: September 29, 2020

Welcome to Series #2 of the Big Compute Podcast!  Ever wondered how COVID-19 can spread through the air indoors?  Is 6 feet apart enough distance?  Do plexiglass barriers protect us?  We talk to undercover superhero, Jiarong Hong of the University of Minnesota, about his discoveries from simulating the movement of aerosol particles in different indoor spaces and how it can affect our everyday lives.  We also dive into his revealing research about what musical instruments may be spreading the virus more than others, including whether or not tuba concerts are worth the risk during this pandemic.

Just play. It's so hot in here. Because you're in the closet. Hello, everyone. I'm Jolie Hales. And I'm Ernest DeLeon. And welcome to the Big Compute Podcast. Here, we celebrate innovation in a world of virtually unlimited compute power. And we do it by shining a spotlight on the undercover superheroes of today. We're talking about scientists and engineers who are embracing the power of high-performance computing to better the lives of all of us.

From the products we use every day to the technology of tomorrow, high-performance computing and simulations are making it all happen, whether people know it or not. So if you've listened to this podcast before, you might not recognize our voices. That's because we shuffled some things around and have reimagined this podcast with a little bit different style. We hope you like it. Special thanks to the hosts who came before us and all of our subscribers. We are definitely standing on the shoulders of giants. So today we're going to talk about something that's affected pretty much all of us. With health officials here in the U.S. moving fast to prepare for what the CDC has called an inevitable spread.

This is an evolving situation. Literally every day we learn a little bit more about it. In the last 24 hours, we had the most cases in a single day. COVID-19 is something that just kind of jumped out of the bushes and quickly impacted lives around the world. There's been a lot of talk about COVID since the outbreak first began, especially since some of the big players in the pharmaceutical industry have gotten closer to a vaccine. But what we don't hear about as often are the many researchers and scientists who are working behind the scenes to actually make these vaccines and therapeutics possible. There are heroes out there working around the clock to get us answers and to get us help, and they're using modern technology to accelerate the process. I mean, imagine what the

outcome of the 1918 pandemic could have been with today's technology. Oh, seriously. I mean, it's crazy to think what can change in 100 years. I mean, heck, what can change in 10 years these days with tech evolving as quickly as it does? It feels like forever ago that COVID first started being talked about in the media. Yeah, it does. In fact, I'm going to roll back the clock a little bit and tell you about my experience. So back in January, air traffic to China was closed down, right? And we were starting to hear about this virus. And I'm clearly not an epidemiologist, but I do remember doing the math in my head and thinking, how are we legitimately going to stop this thing? You know, all it takes is one infected person

who apparently could be asymptomatic and they just need to come over here to the United States and then infections could increase exponentially. But all we could do at that time was order more food storage. I know I ordered more and hoard toilet paper and then hope for the best. But during the second week of February, we actually had a big compute conference in San Francisco that I was supposed to emcee, which is basically the in-person version of this podcast with thought leaders and speakers from companies that are doing amazing things through high performance computing. for research at Microsoft, head of HPC business development and go-to-market for AWS, cloud solutions architect at Google Cloud, Sam Altman, CEO, OpenAI. But I live in Southern California,

so I was going to have to fly to San Francisco. And again, this is the beginning of February before there was really talk of virus being on our soil. And I remember having a teeth cleaning done a few days before. And on my way out the door, I saw this box of disposable masks that was out on the counter that the hygienists use. And they were pink, which made me happy because yay for pink. So I actually embarrassingly ended up asking the hygienists if I could take one

because I knew I was going to be on this flight to San Francisco later. We laughed and we have a good relationship and they joked with me and I told them, I'm like, hey, you never know. These masks could be worth a lot of money soon. And and they gave me a mask and I took it with me. And then a few days later, I was on my way to San Francisco and I was wearing this pink mask I had gotten from the dentist's office. And I was one of only two people on the flight who was wearing a mask. Again, this is beginning of February. And this isn't to say that I'm smart. Really, it's just because I'm overly cautious and I had a baby at home and I just wasn't about to take any chances. There wasn't a lot of research out yet on how it affects children, for instance. But on the flight, I could see people staring at me. And I did kind of feel

like a moron, but that's okay. I embraced the moron feeling. And I'm like, whatever. But to think that even then I had no idea of the magnitude of really what was coming and coming quickly. And that was literally the last time I was on an airplane was for that trip. Just a few days after the conference, everything started to shut down. In California, the notoriously busy highways are nearly empty. The hustle and bustle of New York is at a standstill. Governors in four states have now asked residents not to go out unless absolutely necessary. And Ernest, I know that your family's been affected by this as well.

I mean, how did you first find out about COVID-19 and when? Yeah, so I used to work within the federal government community in a former life, and I still keep my ear to the ground in those circles. Still have a lot of friends that work across federal government and various agencies, and I still keep my ear to the ground in those circles. I still have a lot of friends that work across the federal government and various agencies, and we talk. So I heard the first whispers about it in late December, early January. And the reason I remember that timeline specifically is because there's a massive conference that takes place every year in Vegas at the beginning of the year.

It's CES, the Consumer Electronics Show. Yeah. CES has always delivered on the vision of what tomorrow can be. And it's doing that again this year. And this conference, it's in the United States, in North America, it's the largest tech conference for consumer electronics. Obviously, there's a bigger one in Taipei, right, that happens every year as well. But this one is massive.

And if you've ever been to CES, there's tens of thousands of people at this conference. And the vast majority of them are from China because the manufacturers of so many of these electronics are headquartered in either Hong Kong or China. Interesting. That makes so much sense. And I haven't actually been to the conference, but that's the time of year where you get to see all the cool news stories about robots and 4K, 8K, 100K TVs. It seems like it's going up and up and up. That's right. But I didn't think about the tide of China. That's right. And so when I started hearing about this, I remember the first thing that came into my mind was, oh no, CES already happened. And if this virus was let loose in China, then chances are it has already made it to U.S. soil. And I think at that point, it had already made it. As a matter of fact,

in retrospect, I believe there are several articles that have been written that the virus was here in January. Some people are suspecting even before that, but at least by January. And one of the articles specifically called out CES as one of the vectors by which this happened. So I remember that being the first thing in my mind. CES was probably the original Petri dish, at least in the Western United States. Now, they've subsequently done studies to find out that there were, at that point, two different strains. One of them had made its way through Europe. The other one came directly through Asia. And they're thinking that the strand that hit New York, at least, was from Europe. And they're unsure about the rest of them. So I knew about the virus much earlier than most, right? And that being said,

my wife and I were in California for a short bit. We're typically based out of Texas. We spend most of the year in Texas. However, we were here in California when this quote unquote made the news. We happened to come back to California just before the general lockdown happened and we got trapped here, if you want to call it that. My wife at the time was already five months pregnant with our first child when the big lockdown came. So we had to deal with that. On top of everything else, the pandemic changed for everyone else.

Oh my gosh. As you can imagine, we were stuck inside all the time and I had to leave the house for anything that we needed because I did not want to expose my pregnant wife to the virus. With some luck, neither of us contracted the virus. My sister, however, who is a registered nurse, did catch it while at work. Oh.

Yeah. She fully recovered and is fine now. But I want people to know that my sister and every other first responder and member of the medical community out there are true heroes. Amen. They stepped up when America needed them most, and many gave their lives to give their patients a chance to survive and return home to their families. Oh, it's so true. It's so true. I mean, the images that we've been seeing out of the various hot spots and just the intense stress that our medical workers must be under. Thankfully, it seems to be lightening quite a bit and getting much better. But during that first big wave,

I mean, the real heroes, like you said, were seriously those nurses, those doctors, and even in some cases, volunteers who were out there putting their lives on the line for the sake of other people. So if you're a medical professional and you're out there listening to this, big round of applause for you. Just another night. You can see everybody's cheering for health care providers. Such amazing support from the community. I just, I can't even believe it. So in the media, we've been hearing predictions and health advice like crazy. You know, wash your hands, socially distance, all of that. But we don't hear much about how scientists have discovered which advice to give. And that's where Big Compute comes in. So to kick off our podcast reboot, we've spoken to some of these

incredible scientists, including this guy. Jorong Hong, I'm an associate professor at mechanical engineering at the University of Minnesota. Jorong studies fluid mechanics, which is basically the study of how liquids, gases, or plasmas respond to forces that are exerted upon them. For Jurong, while a majority of his work is on the experimental side, he also does a lot of computational work to understand how fluid moves and how particles are transported in different flow environments. But that's not all he does. I play soccer, I play tennis, play badminton. Whenever I have time, I like to travel. And as a father of two, and I have a lab with 20 people. And right now it's just so hard to find time for myself.

Jurong is a self-proclaimed workaholic with a full daily schedule. I kind of sleep like four or five hours a day. Four or five hours? That's my typical schedule, yeah. And this pandemic has affected his family and work life the same way it's affected many of us. Both my wife and I are actually from China, so we do have a lot of relatives in China. So, this outbreak started in China, and we learned about COVID-19 around January. And we know that Chinese governments have been taking very aggressive actions to stop the virus. And we did not expect the virus will actually migrate so quickly to other parts of the world. But Zhirong watched with the rest of us as the virus did spread across the globe, eventually crossing into the United States.

And by mid-March, his children's school in Minnesota had shut down and his lab sent people away to work remotely. At first, the hope was that this virus was going to die out quickly. But as we know, that didn't happen. Days turned into weeks. And by mid-April, Geron realized that this could be their new way of life for a long time. I started thinking about other creative ways to do some research during this period because our lab had to shut down. And with his lab shut down, Jurong, the workaholic, started to get restless.

Yeah, you know that I'm a workaholic. If I don't work, it becomes very uncomfortable. But there was one way the university would let him back on campus. Research that was specific to COVID-19 was still allowed to continue in the lab, which was perfect because Jerome wanted to help fight the virus and that would get him back into the workplace. So I was thinking about some creative way to do some work related to COVID-19 that I can get back to my lab. That's how it got started. I love that he came up with research idea because he had to get out of the house again. Yeah, right. I think we can all relate. I still only leave my house like once a week

to go running at the beach on Saturdays. I'm legitimately recording this right now from my bedroom closet and it's like a thousand degrees in here. I also only leave my house once a week to get necessary supplies and I still take every precaution to ensure I don't contract the virus as I have an infinite home now. So far I've been lucky, but I make sure that I don't leave anything within my control to chance. Oh, seriously. So while Jerong was trying to come up with COVID related research, there were these rumors that the virus might be spreading by way of contaminated surfaces, like the flu often spreads. But then some early studies started to

show that it might actually be primarily passing from person to person through the air by talking or singing or even breathing. Wasn't there a choir in Washington where just about everyone contracted the virus after one person showed up with cold-like symptoms? Yes, there was. 61 people attended two choir practices in early March, and 53 of them walked away infected. That is 87% of the choir. But in early March, I mean, there really wasn't a lot of evidence out there to say how the virus was spreading, so members of the choir didn't really know what kind of danger they might have been in.

Research on the subject was in its infancy and honestly, more evidence was needed. Enter Jerong and his supercomputer. For airborne transportation, it's a fluid mechanical phenomenon. And it's a particle that moves in the fluid and being dispersed in the different regions. So if the disease is airborne, it becomes very challenging to actually reduce the risks based on the social distancing because the particles can stay and suspend in the air for a long period of time, right? So they can transport a long distance. Even if you are six feet apart, it does not necessarily mean you can get away from those

particles because they can be transported through the ventilation system in the rooms through the airflow and natural airflow in the room. That's interesting that we as a public are always being told to practice social distancing. And we're told that social distancing basically means six feet apart. But he's saying it's not that simple because particles can stay in the air for a long time and can travel even further than six feet. Yep. Kind of interesting that every single sign that we see in public places insisting on staying six feet apart might not actually be universally accurate depending on the physical space that you're in. So we're thinking about can we actually use our measurements and computations to quantify how aerosol particles transport in these different indoor spaces.

Because when people talk about social distancing, talk about different guidelines, they talk about this in a very quantitative way. But there is no really quantitative information to tell you the risk level under different indoor spaces. And if the person stays there for two hours, three hours in a room of different sizes, what is the corresponding risk levels are? Because this quantitative information can give people very precise guidelines. So what they should do under different circumstances. So this is where high performance computing comes in.

Yes. So we are doing so-called computational fluid dynamics simulation. So we're actually using our computers to solve equations that govern the motion of the fluid. And so then we put particles in the fluid and solving the equations that determine the motion of the particle in that fluid. So this allows us to track those particles and track their motions and study how they disperse in that space. So we use the supercomputer provided by the University of Minnesota. It has more than 10,000 nodes.

Each node has 128 cores. Well, of course, we're not using the whole supercomputer. We're just using one single node and with 128 cores to do the computation. To start, Zhuang and his team picked three different indoor environments to computationally simulate how aerosol particles carrying the virus could spread from an asymptomatic individual, depending on the physical indoor environment. This includes where people are located in that space, how long they are there, and where the air vent is in the room. It's interesting that they took the ventilation system into account. I don't think a lot of us consider that when we walk into a room. Right. And as Jerome says,

there's not a lot of quantitative risk assessment out there. A lot of the advice that we hear is just this general six feet social distancing thing, which is better than nothing, but it's a broad stroke for an unlimited number of circumstances. So Durong's team ran simulations specifically for a classroom, an elevator, and a grocery store. So let's talk about the classroom study first. I do have a collaborator I would like to give credits to Professor So Young, who was doing the simulation work with us in this study. Okay, so picture this. A classroom that is 5 meters wide, 10 meters long, and 3 meters high. But let's be honest, if your metric system is stupid like I am, that doesn't make any sense.

We're talking about 16.5 feet wide, 33 feet long, and almost 10 feet high. So a typical grade school classroom. I would guess pretty average. Now imagine a teacher. What's 40 plus 20? We'll call him Mr. Teacher because we're really creative. And he's standing at the front of the room looking down the length of the room where there are rows of desks set up for students. Now Mr. Teacher doesn't know it, but he's an asymptomatic carrier of COVID-19. Maybe he picked it up at a birthday party or he went on a date with someone who had it and was also asymptomatic.

I don't know. But now he's talking and he's breathing at the front of this classroom. And as he's doing that, aerosol particles or tiny droplets of saliva are projecting from his mouth and then they're floating around in this classroom. So, Ernest, out of all of those rows of desks, who do you think is at most risk of contracting the virus? Well, where are the air vents placed? Oh, you're a smart one, Ernest. You're smart. I'm so glad you asked because they are a huge factor in this equation. In fact, Jurong said that the biggest learning they came away with was on just how important ventilation placement is in either mitigating or increasing the risk level of contracting COVID-19 in an indoor space.

People have talked about social distancing. People talk about a mask using. People talk about, you know, disinfection. And the ventilation, I think, is another important factor that was brought up by our study. We were experimenting the ventilation at different locations and different flow rate in different indoor settings. We're looking at how the ventilation can help or change essentially the risk level, spatially and temporally in that space. So first, with this classroom scenario, Zhirong and Su Yang, his colleague, they use high-performance computing to run the simulation of aerosol particles in a classroom with an air vent in the ceiling just behind the head of Mr.

Teacher at the front of the room. You with me so far? Yep. Cool. And we're talking about a return vent, which is the kind that sucks air out of the room, not the kind that blows like hot or cold air into the room. So they ran the simulation and then they ran the same simulation again, but with the air vent on the opposite side of the room in the ceiling behind the desks. And guess what they found? When the air vent was behind the teacher, the aerosol particles were projected enough to reach the front row of desks, but didn't really go beyond that. So it was like a splash zone at SeaWorld. The front rows may get wet or in this case get COVID. Exactly.

But then when the vent was in the back of the room behind the students, the results were quite different. Rather than a SeaWorld splash zone, when the air vent was at the back of the room, the air circulation actually pulled Mr. Teacher's aerosol droplets all the way back across the room, hitting every row along the way, which means potentially every student. This particle has to travel all the way across the entire classroom to reach to that ventilation site. So this helps actually spreading the particles across the entire space. That's probably the only difference between these two scenarios. They've published some of these simulations so people can see them, and we'll post those on bigcompute.org under this episode.

And there's a video here that shows the difference between the two classroom scenarios. So I wanted to show this to you. So in the first classroom, it's pretty interesting. Like you said, Mr. Teacher is standing at the front of the classroom and there's an air return vent right above his left shoulder. And you can see the particles here. They're kind of blue and red and you can see them moving around. And there's a lot of movement around the teacher but as you go further back in the classroom away from him the particles are moving less and less until the very back when they're kind of nearly stationary on the rear wall.

Yeah, it's super interesting because they really do stay close to the teacher and the vent but then there's the other simulation where that vent is moved to the very very back of the room so on the opposite side is the teacher behind the students and and tell us what you see there so like jerong said this one is really interesting because the because the vent is on the other side of the room it's pulling the air across so it's dragging the particles with it so in this one, the same particles that you can see kind of in red and maybe blue are very active throughout the entire room from the front to the back, as opposed to mostly in the front on the previous

one. Right. And I mean, I wouldn't have really considered that. I mean, you always wonder about the airflow, but it's so interesting to actually see it simulated because that's what computational simulation does. It gives us that visual representation of the experiment. And this really does show how critical placement of ventilation systems is. I mean, I didn't really think it would be this critical, but it's a make or break for a lot of these kids who would be in the classroom potentially getting COVID, which is crazy. First of all, I want to remind you, those particles are very, very small. Here you see the red dots. They look very big, but in fact, you know, this is just exaggeration to allow you to see those particles.

In reality, you cannot see any of these particles. They are only like less than five micrometers in size. So those particles are very small, so they are very airborne. So they travel along the trajectory of airflow. So essentially the trajectory of airflow can represent the trajectory of particles, because the particles are very small from the fundamental fluid mechanics point of view. So if someone were to look at this information, it could potentially, I would venture to say, it should affect the way we set up our classrooms as our kids return to school.

That's what I would think. her to say it should affect the way we set up our classrooms as our kids return to school. That's what I would think. Put the person talking like the teacher as close as you can to the air vent and maybe the room will be a safer space. Or maybe even consider where the air vents are when you design a school. I mean, in fact, maybe we should be considering air circulation flow patterns more carefully when we're designing ventilation systems for buildings and for homes. Absolutely. And this goes beyond this pandemic, right? This can go for transmission of particles for the seasonal flu. So yes, we should be putting these things into practice over time. But I know that schools are struggling right now with budgets in general, and a lot of the buildings they're in are just, they were built so long ago, none of this was taken into consideration. Yeah, exactly.

Well, and maybe in the future we will, as technology makes it more possible for us to examine these kinds of scenarios without having to run the experiment in person, which obviously takes time, money, resources, and realistically may not even be possible. Thanks to supercomputing, researchers can simulate real life scenarios in a computer and look at the visual representation of the results, and that can tell us a lot. And since simulations are often quite scalable, you can go as fast as hardware or cloud HPC will let you go. So for that simulation, it took us about three days to run that on a single node with 128 cores. 128 cores. So just to compare for those who don't work with high-performance computing, like me, until a year ago,

if Jerome would have run that simulation on, let's say, an average desktop computer with four cores, it would have taken him... A couple of months, I mean, to run the simulation, I think, on an ordinary PC. And it's because high-performance computing brings a lot more things into play than your typical desktop does, right? So even if your processor, let's say, on your desktop is just as powerful as the 128 cores that are in the supercomputer, the high-speed interconnects for storage that contain the data set that's being crunched in the supercomputer don't exist on a standard desktop PC.

Right. So there's a lot of latency in moving those data sets in and out of RAM to be accessed and crunched by all this. So mathematically, if you look at just core count to core count, it might only look like a couple of months. But, you know, realistically, it's probably much longer than that. So Jerong's team also ran similar simulations on an elevator scenario, which I was super curious about because I would think that elevators are a high risk environment for spreading aerosol germs. You know, you're all standing there, you're in this tiny little space, you're in close proximity to everybody around you. And it's always super awkward because you're aware of everyone else's presence but nobody's got anything to say to each other and frankly speaking i imagine it would be a lot less awkward for the

actual aerosol particles who don't have a problem picking who they socialize with so i would think that this enclosed space would be a high risk environment but when gerong ran the simulations they actually found something different than what I would have guessed. For low-rise buildings, the simulation time within that space is only one minute. Because of the short simulation time, the particle doesn't really spread that much. So in the elevator simulations, ventilation wasn't as critical as the amount of time the individual was inside the space. Right. On average, aerosol particles didn't actually spread much in the typical low-rise building elevator because people only stay in there for like a minute or so. In the majority of the space, within that one minute, the risk level

is very, very low. It's almost less than one particle you could possibly encounter in the majority part of the space during that one minute. So you might encounter one particle, one of those tiny little dots. They did say that there were high risk spots, like hot spots in the elevator, depending on where they put the air vent, but that mostly the high risk area was just mostly centered around the person actually producing the particles and didn't move much beyond that. So as long as you're not in a packed elevator where everyone is up in everyone else's business, you should be okay. For the most part. There were exceptions.

But if you consider elevating for the high-rise buildings like in New York City, the situation could be very, very different because the person could be staying in the elevator for longer than one minute or sometimes multiple minutes, and you could have more staying in the elevator for longer than one minute or sometimes multiple minutes. And you could have more people in the same space. So that potentially increases the risk levels substantially. Have you ever seen the movie Elf with Will Ferrell? Yes, a long time ago.

Well, you're due to watch it again, I think, because it pertains to this particular experiment. There's a scene in Elf that I think of here where Will Ferrell's character, Buddy, goes into the elevator at the Empire State Building, and there's some random person standing in there. And Buddy, who's really never been in an elevator, pushes one of the elevator buttons, and it lights up, and it makes him really excited, and he thinks it's pretty, like a Christmas light. So he pushes every single button on the elevator and then he gets off the elevator and leaves that random person there with all the buttons lit up

because he thinks it looks like a Christmas tree. I think that scene is hilarious. And it just makes me think here that this guy now has to ride on this elevator and suddenly, you know, this is actually a safety hazard, which we never really considered before. So I think the moral of this story is if you need to ride in an elevator in a high-rise building, avoid riding with somebody dressed as a Christmas elf. Or if you happen to see Will Ferrell get in the elevator before you,

wait for the next one. Right. But it is interesting how the longer elevator rides really do increase that risk is what they were finding in simulations. So what about the grocery store? Okay. So for the grocery store study, since all of us still have to go grocery shopping and spend time there, Jurong's team simulated an asymptomatic person basically walking around the store, stopping here and there and being in there for about 30 minutes. And just like the classroom study, they looked at how air vent placement changed that aerosol flow. And in doing this, they discovered that there was one particular person who was at higher risk for becoming infected inside the grocery store as

compared to everyone else. Jerome let me actually guess in our conversation who that might be. Which one you think is the most vulnerable person in the grocery store? I would think, I don't know, the cashier. That's right, the cashier, because this person stands there all the time, right? So he stands in the same place and gets exposed potentially to those aerosols produced by the shoppers and so we look at these two scenarios we found actually when this person he stands closer to the ventilation out he has higher risk of getting infected because all the aerosol particles produced by this asymptomatic individual will eventually be in transport and going into the ventilation out.

So the person, the cashier, that's standing all the time near the ventilation out has the highest risk of getting infected. So again, those viral particles are being sucked up by the air vent and pretty much hitting anything in their path. So if you're a cashier standing near an air vent for a long time and there are shoppers throughout the store emitting these particles, you're not in the greatest position. Which got me thinking. Interesting, because I know that there's plexiglass going up now between the cashier and the customers. And it sounds like that might, I mean, we haven't run any simulations on it, but I imagine that that actually is a good idea, given what you found. Not necessarily. So that's a common misconception because which was

his polite way of saying think about it the the disease is airborne right so airborne means that the particles are floating there they've been transported so in fact the flow can go past the plexiglass you know because this air can disperse in all directions, right? So as long as there's a gap, air can come in there and still disperse the particles. So this plexiglass is usually useful for preventing the direct ejection of larger droplets. For example, this person is sneezing, coughing, and producing a lot of droplets. That can be blocked by the plexiglass. But if this person is just simply breathing and speaking he's producing a very small amount aerosol particles but this

particle very small they are airborne they can be carried by the air and dispersed into all corners in the space so those plexiglass is not so effective in terms of preventing airborne transmission. That's so crazy. It's like a salad sneeze guard, basically. So it's giving us this false sense of security. That's fascinating to me. Right. So we're actually doing some interesting simulation to evaluate this effectiveness of plexiglass. Sometimes I think it does more harm than good because it helps actually creating a circulation

regions behind the glasses that help you accumulate aerosol particles there. Interesting. Really? So actually putting up the plexiglass could be making the situation worse potentially for the cashier because those aerosol particles are getting in there where the cashier is anyway, even though the glass is there. And then it's just kind of sitting there spinning around with them because of this. That's a possibility. Yes, we haven't confirmed that yet, but that's why this airborne transmission is so much different.

But really, how many times have you seen plexiglass installed during these pandemic times or heard about it talked about for like schools, for instance? I know that this morning I was at the dentist, which is the same dentist that gave me that pink mask, and they had plexiglass installed at the front counter. And then they were also talking about how plexiglass is being installed around every individual desk at the schools. So the students, when they go back to school, they have this like protective barrier there. But now science is showing us that it might actually be making the problem worse to have that plexiglass there. They're still running these simulations and we don't have all the results yet, but that's a possibility that they're looking

at. And it's possible also that just wearing a mask is better than going through the trouble and the cost of installing plexiglass. Yeah. And I think part of the story here, right, is that this situation is very much evolving, even where we are today, six months out from this, right? It's still evolving. The science is still evolving. Studies that are providing a lot more data to help guide policy, right, to help reduce the number of infections. And I think a lot of these things that were put in place, like the plexiglass shields and whatnot, were done before we really understood about aerosolization of these particles, right? They were primarily worried, like we said, about large droplets that, you know, someone directly sneezed on someone or someone directly coughed on someone. I don't think they

realized at that point how small the particles could actually be and how they could flow in the air like they do. Exactly. We've seen recommendations change during this pandemic quite a few times because of this, right? Respirators are essential. Oh, wait, maybe it's better not to be on a respirator. This drug is basically a cure. No, it isn't. Don't wear a mask. Everyone wear masks. And while it can feel frustrating at times because it affects how we keep our families safe, maybe this rapid changing of directions is something that we can, dare I say, almost depreciate in a strange way. I mean, obviously there are terrible consequences to bad health information, and we don't want that. But we're really reaching more accurate scientific conclusions, I think, faster than ever before. And since science involves the constant testing of hypotheses, it makes sense that it would be this evolving process.

It's just a very public evolving process right now, which is something that none of us are really used to. And this is why these studies are so critical and why it's so critical that we fund this type of science. A lot of the answers that we're looking for can be found through science, through the scientific method, through testing, through experimentation. But all of that requires people, time and money. And in this scenario, while we can throw all of the compute power at this that we have, we still need the people and the science to drive that. That's a good point. And that's where our supercomputer superhero researchers come into play for sure. They really are helping us learn about this novel, invisible punk virus so that we know more how to handle it. I know that next time I see plexiglass in a grocery store, I'm going to start looking around for the nearest air

vent just out of pure curiosity. But the aerosol study of the various environments isn't the only COVID-related research that Jurong's been involved in. He's been involved in other research that is, shall we say, even more in tune. Oh, man. So dumb, right? Is there such a thing as a mom joke that's like, that's the counter to the dad joke? Because that was it right there. More on that after the break. They call it high performance computing, but is it really living up to its name?

I mean, how much has really changed in the last 15 years? Since the revolutionary jump to cluster computing, there have been new core types, new ways to queue jobs, but no real seismic shift. Until now. Introducing Rescale, the intelligent control plane that allows you to run any app on any infrastructure, totally optimized. Innovators are moving away from the traditional data center only model Rescale. Tomorrow's HPC, today. so we've learned a bit about how aerosol particles flow through a classroom an elevator and a grocery store but there are other ways aerosol particles can be distributed that don't involve using your vocal cords. Ernest, do you play any instruments? I actually play quite a few. Bass guitar is probably the one I play now most, but in the past I played quite a few brass instruments from trumpet all

the way up to my favorite, which is the tuba. You played the tuba and the trumpet? I did for many years. That's awesome. We need to get a picture of that and put it on bigcompute.org so everybody can see Ernest playing the trumpet. I've destroyed all those pictures. Whatever. I'm going to find out your mom's phone number.

We're going to get one. How about you, Jerong? I did try to play a few instruments. So I'm never good at it. I don't know whether you consider harmonica as an all-music instrument. Perhaps it is, right? I play harmonica. I play a little bit of violin at some point.

I played the piccolo a little bit when I was young. That's pretty good. You say you don't play, and then you name off like 10 instruments. But never good at any one of them. Just played it for a year and then dropped it. So Jolie, how about you? I do the singing thing more than anything else, and I tinker around with the piano and the drums.

In high school, I was in a band called Jolie's Our Drummer. So that's my claim to shame. I did study music in undergrad, but I didn't ever play any instruments that involve blowing air into like a mouthpiece or something. So no real aerosol particle instruments. Yeah, we mentioned earlier how choirs have been affected by this pandemic, but this also has not been a good year for orchestras and bands. Yes, which is exactly where the Minnesota Orchestra approached Jurong's team at the University of Minnesota and asked for help. They were extremely concerned about the airborne transmission, especially the musicians that play the wind instruments and you blow a lot of air So Jurong started studying what amount of aerosol particles are being produced by each type of wind or brass instrument and how these particles are being transported through the type of wind or brass instrument and how these particles are being transported

through the air of an orchestra hall. And the hope was that they would be able to provide the Minnesota Orchestra with a quantifiable risk assessment and mitigation strategies so that they can decide how to best move forward during this pandemic. In fact, we just have a paper accepted and CDC got in touch with us because

they wanted to use our information to provide the public guidance about people who play those kinds of music instruments. So are we ready to hear which musical instruments they studied? It was the piccolo, flute, clarinet, bass clarinet, bass clarinet, oboe, trumpet, French horn, bass trombone, and the tuba. They then compared the aerosol particle amounts emitted by each instrument with the aerosol particles that come from like just breathing or speaking normally. And then they used that information to categorize the instrument as either being low risk, intermediate risk, or high risk. So the low risk instruments produced less aerosol particles than normal talking or breathing. The intermediate risk instruments produced about the same amount

of aerosol particles as normal talking or breathing. And the high risk instruments produced more aerosol particles as normal talking or breathing. So we roped Ernest into this conversation, and we took a guess on which instruments would be the highest risk. That's a tough one, because I'm a musician as well, right? And it requires a lot more air to activate, you know, to play a tuba, but at the same time, there's a lot more tubing. And my guess is that a lot of that aerosol would you know stick to the inside of the and then condense pretty rapidly because the metal is cold so my guess is that the shorter instruments are the ones with the least amount of tubing or you know distance between the input and the output

would allow the most amount of aerosol out as opposed to the longer tubes would condense it faster and turn it into essentially water inside the instrument, right? That's my guess. So you're thinking like the shorter instruments like the flute or the trumpet? Yeah, are the worst ones as opposed to the bigger ones. But then again, it's a lot more air for the bigger ones. So you are theoretically putting more aerosol through it in the same amount of time. So it's hard to say. That's true. That's true.

And I think I'm going to echo the same thoughts as you, Ernest. But I'm going to go a little less scientific in my conclusion. When I see somebody play a flute, they look like they're relaxed and having a good old time. Whereas when my brother plays the trombone, when somebody plays the trombone or the trumpet man your face turns so red and it looks like you're just pushing every air particle you have inside your body into this instrument so i'm gonna guess trumpet maybe trombone and tuba you guys intuition is amazing. I think you got exactly right, Ernest and Julie. I think that's exactly what we got. The trumpet?

Trumpet is the highest. The tuba, actually, like Ernest said, you put a lot of gas, but the tubing length is so long, so much longer than the trumpet. So you get a lot of condensation inside the tube. And aerosol particles, a lot of them deposit in the inner surface of the tube. In fact, the tuba is the lowest risk level instrument out of all 10 instruments. The tuba, that means that maybe the annual tuba Christmas concert at Disneyland might still be okay. Have you heard of it, though I've never attended it. Back in the day when I used to have a tuba, I would participate in the annual Tubameister Christmas in San Antonio.

It's been a very long time since I've done that. I'm not even sure if they still have it. If this study tells us that tubas emit less aerosol particles than talking or breathing, which is basically what they're saying because that puts them in the low-risk category, maybe the tuba concert is the one concert that can still proceed safely during this pandemic. So everyone should set up outdoor tuba concerts. But at the same time, if it were a trumpet concert, we have to stay back because apparently the trumpet is the highest risk instrument because it produces, get this, 10 times more aerosol particles than normal speaking or breathing.

And it's not alone. I mean, right up there with it are other high-risk instruments that produce more particles than normal talking or breathing, and that would be the trombone and the oboe. And as for the results of the intermediate risk group, which are instruments that produce the same amount of aerosol as normal speaking or breathing, we have the piccolo, flute, bass clarinet, French horn, and clarinet.

It's actually not that bad. It's not as worse as we thought. And our tuba stands alone in the low-risk category, as the one instrument tested that actually produces less aerosol particles than speaking and breathing. Which is why, according to my own personal opinion opinion which is in no way medical advice every city council should be planning an outdoor tuba concert right now and in that regard san antonio texas my hometown is ahead of the game san antonio but what i was just describing is aerosol generation but there's another part which is aerosol generation, but there's another part, which is aerosol transport. I haven't talked about that. This is actually our phase two study.

It's still ongoing. We're actually using supercomputers to simulate the amount of aerosol produced by those instruments and how they transport in that space. But that's not all he threw at us. Another important factor beyond the ventilation is actually the human thermal plumes. So when you sit in that space, you actually have a higher temperature, body temperature, compared to ambient air. So you're actually generating flow yourself.

You actually have a flow rising from your body. So that flow essentially carries aerosol particles away from your body. What? So you're almost creating like a Krebs cycle or an eddy of air because of the temperature differential. So while I broke that one down in my head, I asked you wrong. Would what you learned from this musical instrument study ever make you give advice to somebody in your family as to like which instrument they should play? Or are the findings to where you still don't think it should change human

behavior necessarily? It shouldn't change. You know, what we're trying to do is just try to keep things the way they are, but to provide some interventions to minimize the risks. You know, we don't want to change people's behavior. So when Jerome was doing this music study, he actually provided the orchestra with some suggestions on where each musician could sit in order to minimize infection risk. I can imagine how that went. Yeah, the musicians were not really big fans of that idea. I mean, which makes sense. Orchestra seating arrangements are determined in such a way that they both project the best sound and they can hear each other well.

Exactly. If you move the clarinet away from the bassoon or oboe, they're probably not going to be able to hear each other as easily, which could affect the way they play and their synchronization. Right. And Jurong's goal isn't to change human behavior. So he and his team have actually been considering other possible unintrusive ways that risk could be minimized. We can put a mask on the trumpet. Those are acoustic transparent masks, but can actually block out a substantial portion of aerosols. The idea of putting a mask on a trumpet, that gave me an interesting visual in my head. But it makes sense. It makes sense if there's a way to do that without muffling the sound in any way, which I'm sure there is.

For example, your microphone has a mask, right? It has a fabric in front of it. So it doesn't affect the sound, but it can help block out some particles. He also talked about possibly adding localized air filters to the orchestra floor that would capture aerosol drops, but be quiet enough to not affect the sound of a symphony, or even playing with temperature control. If we change the room temperature by two or three degrees, we actually can take advantage of natural flow produced by each individual musician that helps us to concentrate aerosol particles moving up instead of spreading horizontally. So that may help us to mitigate the risks as well.

So let's take this back specifically to high-performance computing. I asked Jurong about his expertise using HPC to do simulations and modeling and how that may affect future research. I didn't do much of the supercomputing in the past. I did some supercomputing using DOD computer to simulate, for example, the wind turbines and the super cavitation flows. I didn't use that much. It's not a majority portion of my research, but through this COVID research, I really become more and more appreciating the potential of the supercomputing for really understanding and providing the guidelines for our lives. I think this is going to change significantly our behavior and how we actually operate. Because this provides a lot of scientific-driven guidance rather than very qualitative information on how we should mitigate risks. One of the things that has kind of bubbled to the surface more recently in the last,

let's say, five to 10 years is cloud HPC, right? Where instead of using an on-premise supercomputer that's limited in size and scale, and like you said, in limited amount of time, you can use it. It's also very costly. Now, services like Amazon, AWS, Azure are offering cloud HPC, which have virtually unlimited compute power, right? These are data center scale computers, not the traditional supercomputer. In that sense, right, once you get to the point where you have virtually unlimited compute power, right, where you're not worried about 128 cores

or 10,000 cores. What's possible now that wasn't possible 20 years ago? Right. For a lot of applications, right, it's not just that you need to compute accurately, but you also have to compute fast, right? You want to get the results almost instantly rather than you have to wait for two days or three days or even months, to get the results and the more accurate one model are the the scenario and the more computational power you need and so the the slower it gets so right if you have unlimited computational power i think that's going to change how things work

you know because it's going to get the results instantly and it's going to allow you to directly use that information generated by supercomputing and very high accuracy supercomputing to guide your behaviors, guide your operations. So that's going to change things completely. Jerong went on to say that his lab in particular hasn't seen that kind of compute yet. They haven't been using the cloud for high performance computing, but the idea of virtually unlimited compute was exciting to him. I would imagine that as his studies grow in scale, you know, like well beyond 128 cores on a single node, which I'm sure they will, given the great work that he's doing, Azure, AWS, Google, IBM, Oracle's now in the game. All the cloud HPC providers are going to be even more tempting. Maybe so. Jerome, thank you so much. People like you are absolutely

the undercover superheroes, as we say, because it is such an evolving situation and the public is so anxious to know what COVID is all about and how we can keep our families safe and how we should react to it and respond to it. And if it weren't for people like you and studies like this,

I think we would be so much more in the dark and investing in so much more plexiglass unnecessarily so that it's a real honor to talk to you about this. And we're just grateful that people like you exist. I think it's critical, you know, in any scenario where human beings need to make decisions to have the most accurate and the most current data to make good decisions. And the data that you're putting out there allows us and allows policymakers and public health officials to make better decisions and to keep us safe. So like Jolie said, you're definitely an undercover superhero. And I just want to thank you for doing this critical work.

Thank you very much, too. It's good to know that our work can generate this level of impact. That's what we keep talking about in the scientific community. We want our science to generate the impact. It's my pleasure. To learn more about these studies and to keep up with the latest from Jurong and his team at the University of Minnesota, you can visit bigcompute.org and check out the episode notes where you can find their website, their Twitter information. They have a YouTube channel and we'll provide links to pictures and videos and more great stuff. That's going to do it for this episode of the Big Compute Podcast.

To everyone out there, thanks for joining us and we'll catch you in the next episode. See you next time. you