Embedded - 169: Sit on Top of a Volcano

Episode Date: September 21, 2016

John Leeman (@geo_leeman) spoke with us about geophysics and associated technology. John is one of the hosts of the Don't Panic GeoCast (@dontpanicgeo, iTunes). Some episodes you may like: What if you calibrated your candles differently? Out of the Country (Brad Jolive on moon rocks) "Rock Drills and Beer" Undersampled Radio John is teaching a course at Penn State called Techniques of Geoscientific Experimentation. The information and textbook is online! It uses the SparkFun Inventor's Kit. John has a website with a blog. He has some Cheerson CX-10 tiny drone posts (my favorite, also Alvaro's repo and my posts). John also has a consulting company: Leeman GeoPhysical. Python! Lots of Python was discussed.  Jupyter notebooks (here is a good tutorial) Example of reproducing a figure from a paper John's friction model (repo and talk he gave about it at SciPy2016) Neat SciPy talk about open textbooks SciPy is a Python conference in Austin, TX in July Finally, in lieu of rock puns, here is a neat animation showing many different waves from earthquakes. Contest! Contest ends October 1st and now there are more books! In addition to the ones Bob Apthorpe is sponsoring, John's consulting company will sponsor: Earthquake Storms: An Unauthorized Biography of the San Andreas Fault by John Dvorak and The Soul of A New Machine by Tracy Kidder. 

Welcome to Embedded. I'm Alicia White, alongside Christopher White. This week, we'll be talking to John Lehman, scientist by day, engineer by night, coder by birth, mad scientist at large, also podcaster. We do still have that contest going on until October 1st. You're supposed to be sending me a number. What's even better is even though Chris's number has been guessed, he's getting a new number this week, and John's getting a number this week because he's sponsoring a contest. We'll talk more about that later.

But in the meantime, send me all your numbers. And remember, Graham's number is taken. So maybe in the normal number range. Also, negative Avogadro's number is taken. So maybe in the more normal range. Seriously. Hi, John. Good afternoon. Hi.

So can you tell us about yourself? Sure. So I got a bachelor's in meteorology and in geophysics from the University of Oklahoma back in 2012. So I've always been kind of an earth and geoscience person. And then after I did that, I decided that wasn't enough fun. So I started my PhD over at Penn State, which is where I am now. And I work in the Rock and sediment mechanics laboratory. So I do a lot of kind of hardware, science, software, all mixed together trying to get some answers

to some interesting questions. And what kind of questions? I mean, it's the earth. It doesn't change. What? On a long enough time span, everything changes. That's fair. I have a lot of rock questions, but let's do lightning round first. span, everything changes. That's fair. I have a lot of rock questions, but let's do lightning round first. Oh, sounds great. Okay, I know you listen to the show, so you're pretty familiar with we try to ask short questions, you try to provide short answers, and we try not to ask you for more detail.

Okay. So the first one. Science, technology, engineering, or math? I'm going to say all of them to accomplish anything. You need to have some of all. So yes to all that question. Beach or mountains? Mountains, definitely.

Favorite fictional robot? Oh, this is a hard one. I think I'm going to have to go with Data. I don't think that one's been taken. That's a good one from Star Trek. Yeah, it's a really great example of how interaction could eventually work with our machines. Do you really think we're going to get to the point where they have real natural language processing in such a lifelike Android without them having emotional centers? He a chip yeah he was on the fritz though i don't know it's it's a pretty lofty goal but it would be really great compared to the way that we

interact now which is pretty unnatural io europa or ganymede oh uh europa all right this question has a parenthetical so i guess i'll ask it both ways favorite type of rock okay favorite type of rock i'm going to say pseudotaculite which is something you're only going to get from somebody that works on fault probably it's this it can happen in any type of material, but when you get fault movement, there's this large amount of heating, or sometimes it happens in big landslides, or when you get meteor impacts. And it's just a melt layer, so you have this really glassy appearance. And it's really cool, because the fault surface had to get, you know,

well above 2300 Fahrenheit to make this stuff. That's distressing. The parenthetical is favorite type of rock band. Oh, geez. Well, we'll go, since this is an electronic show, we'll go with ACDC. How about that? Beaglebone Black or Raspberry Pi? Oh, I'm going to actually say Beaglebone Black.

Any particular reason? I've used both, and while the Pi does have some nice integrations, I just like the fact that the BeagleBone is much more open with the design. The Raspberry Pi is closed source. If I wanted to insult a chip by saying it's slow, what geological timescale should I use? So, hmm, that's interesting interesting so there's a geological timeline but that's different i'm going to say maybe tectonic because if you're looking at

something that happens at a tectonic rate you know you said the the earth doesn't change much the earth's plates are moving but only at you know four to six inches a year. So that's pretty slow. I'd say it's tectonic. It's a tectonic pick. Yes. If I want to insult a chip by saying it's old, what sedimentary layer should I use? Let's see how this whole thing is going to go. Yeah.

Okay. Sedimentary. Okay. How about the belt super group? So the belt super group is this really old fine-grained sediment that you find in like Montana and Idaho, Washington, Wyoming. And it's a 19-mile thick packet of sediment that was deposited between 1,450 and 850 million years ago. So that's pretty old. Oh, sorry. I got out of order. What programming language do you think people should learn in their first CS course? Python.

Should that first CS course be in college? I'm sorry? Absolutely not. I think that we should get people introduced to programming much earlier than we do right now. How early? I would say as early as in elementary school, at least some of the programming concepts. I didn't really get into programming until I was an undergraduate. And that was kind of a shame because that's where a lot of

folks get into it. And you've already missed so many opportunities to learn and employ it. Oh, favorite type of wave? Favorite type of wave? that's got to be the rayleigh wave so the rayleigh wave is this surface wave that it the ground motion is retrograde elliptical and it's it's really bananas because people don't think about the ground making these backward circles but when you do a plot of what a piece of dirt would have done from a seismometer data set, you actually see these really gorgeous backward circles being traced out by the ground. That sounds like snail teeth. Christopher, you wanted one more?

Okay, what science fiction technology concept do you think will become real in our lifetime? I think I'm going to connect this back to the discussion about data actually I'm going to say that interaction with the computers may be a little bit more like the ship's computer on Star Trek something a little bit less like Siri you know we've had these discussions about if Siri ran the Enterprise things would have gone much differently in many episodes I'm sorry did you mean photon torpedoes?

oh wait that's you mean proton torpedoes? Oh, wait, that's so worse. Photon torpedoes. Yeah, canceling the self-destruct might not work so well when it says, I'm sorry, I didn't quite get that. I'm sorry, you're out of network range. I don't know what you mean. Right. I'm having some trouble understanding right now. Yes. Okay, so now let's go on to the less rock-tastic part of the show and focus in

on what you were saying about software and usability and all of that and Python. What do you use Python for? Shouldn't you be out with a rock pick? Well, so what I do is mostly geophysics, which is basically I do geology, but I spend a lot more time behind the computer than a regular geologist would. And so I use Python to do a lot of my data analysis because as an experimental geophysicist, I end up collecting huge data sets and the tools to evaluate them generally don't exist until we make them. So I've used Python as my general purpose hammer for data munging and figuring out how to do the best analysis of my data that I can. Okay, we did establish that the Earth doesn't move at fast timescales. So what does an experimental geophysicist do? Set out sensors and wait for tiny shifts?

No, so I actually recreate earthquakes in the laboratory for what I'm doing for my PhD. So we use this giant hydraulic press that can put about a mega newton of force on each axis. We affectionately call it the biax because it has two axes. And we make little fault zones that are 10 centimeters by 10 centimeters, a lot smaller than real faults, and generate earthquakes on them that if you calculate their magnitude, there's something like magnitude minus two and a half. So they're really small. I was going to say, I'm pretty sure I saw this in a movie and it all went horribly wrong, but that was a bigger machine. Yes. So this is a pretty small machine but we can recreate

all kinds of behaviors and then of course extrapolating it to nature is a little bit of a hairy subject depending on who you talk to oh yes because i mean what is your substrate made out of wow that's the word again what is your land mass made out of? Right. Well, so I simulate the very shallow portions of fault zones. So I just use steel blocks, and then I crush some kind of powder in between them that simulates fault gouge, so the crushed up stuff that's inside a fault. In my case, I just use a material called Minucil that the engineers actually love. It's crushed up glass to a really fine grain size. But I've ground up natural rocks and put in it. I've put clays in it. People in our lab have been known to use all

kinds of things. I used baking flour at one point to simulate rock properties, actually. Have you ever used water to get little tsunamis in there? No, we haven't. That's something that would be kind of interesting, actually, is put the whole thing in a bag and fill it up with water. And do you, does it create mountains? Does it create little gaps? What do you see? Ah, no. So I don't simulate the entire plate collision like that necessarily, but I'm just simulating a small part of the fault. So what you see in the lab is actually not that exciting. You just see this sandwich that is steel blocks and some kind of powder, and you see it move very slowly. And

occasionally when you get these earthquakes, you hear a little pop. It sounds kind of like snapping a pencil. And what we really look for is in the data. So we're recording sheer stress, normal stress, displacements, all of this on the machine at a kilohertz or so, so not incredibly fast. And then we can analyze what's going on during these tiny events to try to figure out how the friction actually works. You know, we were told that friction has this, there's static friction, there's dynamic friction, or maybe there's rolling friction. But friction is actually really complicated. It's velocity dependent. It's chemistry dependent.

It's humidity dependent. There's all kinds of weird things that happen. And we try to study those and figure out what is relevant for fault zones. And so do you instrument the steel blocks or do you have little tiny sensors wandering around in the soil? I think they'd be crushed. Yeah, so we do instrument the blocks. We would love to be able to get some kind of instrument in the layer, but we haven't figured out a way,

like Chris said, that they can survive. So we have instruments on the hydraulic rams that we're pushing the blocks with, and then we mount instruments like DCDTs, which are direct current displacement transducers, and capacitive distance sensors and little strain gauges. We do lots of strain gauge work onto the blocks. This is a physical model, and so it's not, like you said, it's steel blocks. It's not the same as tectonic plates or bedrock using an analog for the material in between. Why is this superior to just doing a computer model?

Well, because we don't really understand the physics of how failure works well enough to write a computer model that is going to tell us anything useful. We're still discovering new things about friction that we didn't know and didn't have in our models before. The best model of friction that we have right now is pretty much an empirical model. It's called the rate and state friction framework. And that's actually, I wrote a Python implementation of it that you can go grab on GitHub and play with it. But you'll see that there are these constants that we've creatively named things like A and B that we used to put in there. That's kind of the state of, we have these hand-waving arguments

about what's causing these effects, but as far as an actual physics-based understanding, we're still lacking quite a bit. That's kind of amazing because in popular science, you know, you read about the fringes of physics and you, oh, you know, quantum gravity is the thing and brain theory and, you know, big cosmological stuff. but when you get into extremes of basic physics there's still big unsolved questions and that that's that's really cool to me and i think people don't pay any attention to that it's like well we don't really quite understand you know friction when we right take it to this level of pressure or what have you

well and it there's implications for a lot of things other than just earthquakes. I've talked to somebody that was interested in using my model and they were actually concerned about robots that handle things, very fragile things, like silicon wafers, for example. Because the longer the robot arm sits stationary, there's actually frictional healing that goes on in the joints. And then when it begins to move again,

it can move in this stick slip or herky-jerky fashion that can cause damage to what it's carrying. So people that do robotics also care quite a bit about these weird frictional effects. Well, and there are ways we can use friction to do what we want to slow down a robotic arm. And then there are ways where it is catastrophic, just as you're saying, and it would be better to know how to predict these things. I'm just surprised that we don't already know this. It is pretty surprising. And there's been a joke that's gone around for a long time, that the friction of everything is 0.6, if you were to ask a rock mechanicist, or well, if you're asking a geologist now, the friction of everything is 0.6 if you were to ask a rock mechanicist. Well, if you ask a geologist now, the friction of everything is 0.6.

If you were to ask one of us that works in friction, we would have said it's 0.6 maybe 15 or 20 years ago. But now we know that's not the case at all. There are things that are super weak, and there are things that are also surprisingly strong that we didn't even consider when we were doing this basic fault modeling in the 70s and 80s. And so your research has potential beyond geophysics. Right, absolutely. So like I said, what we do with friction, it applies to pretty much all

engineered systems, I would say, anything that's mechanical anyway. And it's a big problem for people that are even doing machining when they get tool chatter that has to do with stick slip events occurring between the tool and the metal and it's because well maybe your machine or your tool or your fixturing isn't rigid enough and the system is too de-stiffened so it has this unstable behavior that's exactly what happens in fault zones if you de-stiffen a fault zone enough, it can have an earthquake. And if it's stiffer, it just slides stably. Now I feel like I should ask questions about Oklahoma. And so, do your models actually work for earthquakes? I mean, now that you have these models, you can do some prediction of what happens along this fault line or along this sort of tension system.

Does it work? Are you predicting things? Also, what are the lot of numbers? Yeah. So, we're getting a lot better at understanding earthquakes. I would say predicting is probably too strong of a word now or ever, just because of the vast number of unknowns. You're dealing with something that is at minimum five kilometers below you, probably much deeper than that. And any information that we get from a real fault zone is attenuated through the earth.

The earth is this great low-pass filter that just foils a lot of our seismological attempts to learn things about fault zones, because it filters out all the high-frequency information that contains a lot of local information on the fault that we'd want to know. So our models are getting quite a bit better, and we can understand the rupture dynamics of earthquakes quite a bit better. But predicting, you know, on a geologic timescale, we can do pretty well. We can say that sometime in the next 250 to 500 years, it's pretty likely that an earthquake will occur here. But better than that's quite a ways off. Well, and I think we're using

predicting in the sense of, do you see things in earthquakes that look like your model? Oh, yes. Yes, absolutely. So, we're getting to reproduce some of the more complicated things about earthquakes that we haven't understood. The fact that, you know, it's this finite rupture that starts and it propagates and it can slow down and speed up and it can change onto a different fault segment. And we're able to resolve what the rupture is doing in time much better in the field and our models are starting to be able to reflect that. And what good will that do us? Hopefully, if we can understand how the fault zone works, we can understand the seismic hazard better. You know, the United States Geological Survey makes these seismic hazard maps every year,

and they're pretty good estimates of what the seismic hazard for a given area is. We think based on how we know fault zones work and what we know about how that particular fault works. But we're learning that faults work in a lot of mysterious ways. There are some faults that they can have a large earthquake, and then the next time they can have what's called a slow earthquake. And nobody on the surface ever knows, but a lot of that pent-up strain energy still gets released. And we didn't even know these slow earthquakes existed until about 2001 when our instruments got good enough to detect them.

And now we're totally puzzled as to why some faults have those, some don't, and some have both kinds of earthquakes and why they're going to have a certain kind. Is a slow earthquake something we don't feel because it happens so slowly? Yeah, yeah. So, as you can tell us geologists, we're very creative with our naming, geologists and geophysicists. So, the slow earthquake, it could be the equivalent of a magnitude seven worth of energy, but it's actually released over days to weeks to, in some cases, many, many months.

They were first found in the early 2000s in the Pacific Northwest. That entire section of the subduction zone there actually is locked, and then every 14 months, over about two weeks or so, it slides back westward about a centimeter. So that turns out to be a magnitude 7 moment release over two weeks. So nobody knows it's there. And it was found by looking at GPS data. And once we got really high-quality differential GPS,

we saw these funny east-west motions, and people didn't know what to make of them for quite some time. And now we've observed them pretty much everywhere there's subduction going on. That's probably why I feel lightheaded sometimes. Very slow. And that's the reason. And people have tried to connect them to all kinds of things. You know, 14 months there, people said maybe it's related to this thing called the Chandler wobble, which is an orbital

thing. But now we know that they can happen at any time scale. In New Zealand, they happen and they can occur for a year and a half before they stop. I remember in the 80s that the big thing in California was, oh, the big one's coming. We're predicting within the next 20 years. And there were television movies where everything's destroyed every couple of months. And that was prior to kind of understanding

the slow earthquake phenomenon. Does it make larger earthquakes less likely if this pressure is being relieved in a slow way? It can, and we have seen that. But we've also seen slow earthquakes can transfer stress to a different part of the fault since it's this continuum. And we now know that before the 2011 Tohoku earthquake in Japan, there was a slow slip event that had been going on for about three months prior to it that very likely played a role in

loading the fault partially so they can they can help you or they can hurt you okay i'm totally fascinated by this because there's i mean like could you've used that to predict it but But I wanted to ask about the Python. Right, right, back to Python. Right, right. You use it to make pretty big models with lots of data. What's lots of data? What does that look like? Well, so if we're looking at an experiment, we might have 8 to 10 channels of data with several million rows.

Generally, I would say not more than 10 million rows. And we might have 200 experiments like that. Okay. So we're looking at pretty large quantities. We're also starting to do some acoustic, like trying to measure time of flight of sound waves through our fault zone. And those data sets can get into the mini gigs very quickly. Oh, yeah, definitely.

Depends on how many sensors you have, but definitely. We're looking at multiple sensors sampled at 50 megahertz or so. Wow, that's a fast sample. Yes. What happens at that frequency? I mean you because you won't be able to model that back to real earth things right in this case we're trying to figure out what the how the acoustic velocity the speed of sound and the rock changes during the load up and then during the earthquake and how it heals after the earthquake. And our experimental system is so small, we have to use pretty high frequencies to be

able to probe it effectively because we're looking at fault zones that in our model, our physical analog model are, you know, millimeters thick. Oh, okay. So yes, that makes sense. You're looking at small things and so you have to do higher frequency stuff. Right. So that when you look at big things and so you have to do higher frequency stuff. Right. So that when you look at big things, you can expect the lower frequencies. All right, I'll buy that. Although Chris pointed out, actually Chris sort of laughed at me because I did not ask a Python question.

I'll ask a Python question. Okay. So, do you find yourself hampered by speed of execution? Or is this something where it's just like, I'll run this overnight and I don't worry about it? Yeah, it's more that, actually. The speed of execution, of course, is not as fast as if I did it in C or something like that. But it can actually be pretty good,

especially if you're using something like Cython, where you can go and compile your code. And there are all kinds of just-in-time compilers that are coming out now. But I have to sleep at some point. So my philosophy has been, I'll start this and it can run overnight. Okay.

Now I was just thinking these kind of data sets that, you know, if you wanted them to run fast, put them on a GPU, which is another big challenge in programming. But if you cared about getting it done a thousand times faster, that would be the thing you need to do. But if you cared about getting it done a thousand times faster, that would be the thing you'd need to do. But if you don't care, then why bother?

Right, we haven't explored doing things like parallelizing or anything like that, though at some point it'd probably be, well, we're going to have to eventually. So there's NumPy, which I have occasionally called a numpy when I'm mad at it. And there's SciPy, and you just mentioned scython. Right. What are the differences between these,

and which one should people start with? Well, so these are all packages that you can kind of tack on to Python to give it different capabilities. For example, numpy gives you a lot of array math capabilities, and very fast. Numpy is C, C++, Fortran-based, so you don't lose a lot of array math capabilities, and very fast. NumPy is C, C++, Fortran-based, so you don't lose a lot of speed if you're doing your calculations with that.

Another one, like Matplotlib. That is a fantastic plotting tool, and one that I actually make all my publication figures using Matplotlib. Not GNU plot? No, Matplotlib. Not gnuplot? No, matplotlib is a lot better. I just have fond and terrible memories of gnuplot.

You know, you can't add drop shadows like Excel, but that's by design. You can really tackle all these packages and there's a package for everything. There's a seismology package that lets you suck in data from seismometers and do processing on it. I've been working on the friction toolkit package.

There's, like I said, just a package for about anything you can imagine. And if there's not, there are probably a lot of people that are interested in that. So you can start it and probably get some help. The community around Python is fantastic. That's why I encourage people to learn it initially, because it's so versatile. It's fast enough. There's lots of libraries. And there's so many people using it that are excited to help you get in it or help you

solve problems. You went to the NumPy, no, you went to the SciPy conference recently, is that right? Yeah, it's Scientific Pythons in Austin, Texas every July because that was a good idea. Hey, it's cheap. It's true. Now, there's a lot of Python going on in Austin, which is why it's there. InThot, the company that sponsors the conference, is located in Austin, but it is always very hot. But what is the conference like? The conference, yes, the conference itself is amazing. It's grown a ton. I think this year, there were somewhere between 600 and 700 people there. And these are all people that either use

Python to do science daily to solve problems, or they're people that are helping make the tools that the scientists are going to use. So there are core developers from Matplotlib, core developers from NumPy, some of the Python core team are all there. And it's my favorite conference of the year to go to because you're surrounded by so many people who are so very passionate about making good tools,

about making tools for reproducible science, about helping people write good code. And so there's a couple days of tutorials at the beginning, which are a lot of fun to go to. And they have all levels of tutorial. There are some talks in the center. And then there are two days of sprints at the end, where people get together and try to do all kinds of feature implementations, bug fixes, try to get releases out, and that kind of thing. And you learn so much by doing that and by helping out some of these open source projects. So there are developers and there are scientists, and you're more of a scientist. You do a lot of coding, but some of the people you're talking about, they are professional software engineers.

This is what they do for a living. How it's used is awesome and fun and applications are wonderful, but at the end what they do is make code, not make models. Right. And this conference brings both of those groups together, or is it mostly developers or mostly scientists? I would say it's a pretty even split. The conference brings both groups together, and it's scientists that are a little bit more on the developer side that tend to attend. It's not strictly that. And a lot of the developers come from scientific backgrounds. I would say many of them have been in academia at one point or another, or still are involved with academia somehow,

but have transitioned to their primary job being making the tools. So going to software development as their primary role. It's a really interesting mix of people. And something I think is becoming a lot more common as scientists realize that we have to write good code to do our jobs. And you came pretty late to programming. And I would say programming is a tool, but software engineering is a bigger tool, a bigger thing. Right.

A discipline. It's a discipline. It's the idea that code should be reusable and that it should be safe and that you should be thinking about how you design it and not just typing in, dear Stack Overflow, how do I do this over and over again? Many scientists use programs, but not everybody learns computer science. Right. And that needs to change, I believe. There is so much code that has been written. Well, there's so much code that's been written over and over again by different scientists that are trying to actually tested in a routine way, like you would normally test software from a software engineering perspective.

And there's a saying that untested code is broken code. And that's a little bit scary because you might run your model and look at the output and say, yeah, that looks about right and go on. And that's just not good enough in where we are now. So there's this whole effort with the reproducible science to get more generalized, more software engineering-like tools out there that scientists can use, but make sure that the tools that we're using are indeed tested and are giving us the right answers. There is still a surprising amount of code out there there or even operational software in geoscience that runs on some weird combination of these really long AUX scripts

that are shell scripts and Fortran 77 on an ancient machine that is being hobbled along. I remember being appalled in 1995 when we took a scientific computing course in college and it was in Fortran. I was like, we're still using Fortran? 1995, and it's now 2016 and we're still using Fortran.

Well, there's some things Fortran is really good at, and there's so much existing code. That's the thing. It's like Python. There's a bunch of packages, and people just kept using them for decades. Ock, on the other hand, Ock in scientific computing must go. I mean, AUK is worse than Perl for legibility.

And that's saying something. I mean, that's saying something really important there. Yeah, and anybody that says, done, I would say, research for somebody as part of a graduate degree or maybe even undergraduate, like a senior thesis, you know that feeling of the supervisor hands you a flash drive and it's all the code that they wrote during their PhD.

And you are going to learn how to run that and do analysis on some different data or modify it to do something. And it might be C that was before the standard, or it might be some really old Fortran that is not going to compile anymore, or it might be a 2000 line aux script. It's a little terrifying.

Well, and part of your thesis is to get the data and to write your thesis about geophysical things. And instead, you're doing all this software engineering that doesn't make any sense, and you might as well just rewrite it because reading it is impossible. Right. What comments?

That's the common theme in academic software. There are no comments. How are you making this better? Well, hopefully, by using things like Python with all these packages that are tested and are maintained by people that care about it and combining them with have you had much experience with the jupiter notebooks i've had a little bit of experience with the jupiter notebooks and i think your experience was a lot better although what you sort of sent me made made my, shed a light on mine that made, let's go with no. Okay. So Jupyter notebooks are, well, I think one of the best tools for doing science right now,

because they run in the web browser. They can run different kernels. You can run Python, you can run R, you can run Julia, but let's stick with Python. And they have cells. And in these cells, you can have different things. So you can have some code that you're going to run to do something to your data or read or write files. You can have cells that are just documentation. So like Markdown, where you can actually write these nice, you can put your equations in that you're solving in text

so you can get this nice rendering of them. You can put media in. You can put audio clips. You can put videos. You can link to YouTube videos that will play in there. So all of this is in one place. It's kind of like if you think about reading through a report on something except you can click and run the

code as you go and you can modify it. You can say what happens if I tweak this parameter in this person's code a little bit? And you just run it and the entire notebook can re-render and update. All the plots update. The text can update. It a really really great tool to do science and do analysis or even if you're doing software engineering to develop things so when you're kind of sketching out how you might want an algorithm to work it's generally where i start well then you were talking about the uh scientists and reproducibility.

If you give somebody a Jupyter notebook, they have everything they need to reproduce what you did. They have the data. I mean, you put the paper in, and now they have not only that, they can go look for the data, and then usually if you then look for the data, you still can't find the code.

It's somewhere else. But in the notebook, it's all in one place. And that was kind of cool. Exactly. And if somebody hands you some data and some code and all that, generally what accompanies it is an email or maybe a text file that says, okay, well, you run this one code first, and then you go in and you change this variable to be something else. Then you run it again. And then you run this other script that I've got. Whereas in the notebook, it's very linear. You can follow it very easily. And combining all of the media makes it, I would say, even a great way to write an entire paper. I was about to say, this sounds like something that should replace standard papers.

Okay, here's a reproducible object, or at least for modeling kinds of papers. Oh, absolutely. I mean, can you imagine writing a technical book in it if you were writing something on how to program or how to use some kind of package or a task? Just write it entirely in a notebook. Then people can run it. You could even put it on a server,

and people could run it on their browser without having anything installed on their machine. Which there are also offline versions, right? Correct. Yes. So you can run it locally, and you don't have to have a connection to use that. And that's how I, in fact, do most of my work

and then push the notebooks out to GitHub, and they go out with the paper. I was remembering about my notebook experience, and I was pretty sure it was local, and it was through PyCharm and all that, but that's not important. So one way you are helping people get more into this is through talking about it. And one way you talk about it is not only on this show, but you have your own podcast, right? We do. It's the Don't Panic Geocast. Why shouldn't I panic?

Well, it's written in big, friendly letters. That's the whole idea, just like Hitchhiker's Guide. And the idea is geology isn't an exact science. So that's kind of the tagline for our show is don't panic. It's not an exact science. And you have a co-host? I do. I co-host with Shannon Doolin, who's at the University of Oklahoma. And every week we talk about some kind of topic that is sometimes geology, sometimes tech, sometimes where those two meet. And we also do some interviews. We don't have interviews every week, of course. Sometimes it's just us. But we have a lot of fun doing it. We're't have interviews every week, of course. Sometimes it's just us.

But we have a lot of fun doing it. We're up to 80-something episodes now and still going strong. And who's your audience? This one surprised us. When we started the show, we thought this was going to be a show about geoscience and about geotech for people that work in the field. We expected that we would get a lot of professional geologists, geology professors, that kind of thing listening. And we do have some of that. But when we get feedback, which we've been getting quite a bit of recently, it almost always comes

from folks that say, I found your show because I was interested in geology because of this one thing that maybe sometime in the past got them interested. And they searched for it on iTunes and found us. and they said, wow, this is all really interesting. I didn't know all these things about the earth, which it's kind of important because that's where we all end up living. And they've got hooked on the show. So we've have a lot of requests for, you know, can you talk about this one thing I don't understand or things from folks that are not specialists in the field. And that's been a lot of fun. I learned so much doing the show, and it's something I look forward to every week, really. Have you done a show about paleontology yet?

Because if so, please mark that down as requested. We will add that as requested. That is something where we're going to have to get a guest, though, since I especially focus mostly on geophysics, and Shannon is a paleomagnetist okay you're gonna have to explain that so in paleomagnetics what shannon does is looks at magnetic fields that are stored in the rock from when it was formed so say when the minerals got below their curie temperature and the magnetic orientation was locked in and so she puts them

in this giant magnetometer and slowly demagnetizes them by heating them to different temperatures. And she can actually tell at what time those minerals locked in their magnetic signature based on the orientation of this remnant magnetic field vector inside. So she can date the rocks that way and talk about where they moved, where they might have been, that kind of thing. I had a dumb question and I thought better of it. So you're up. What are some episodes people should get started with? Oh, let's see.

One of my favorites is actually an early one. So maybe the production quality wasn't the best, but it was called, what if you calibrated your candles differently? And it's all about timekeeping, which seems like maybe a dull topic at first, but keeping an accurate time is really hard. And it's something that's very important for everything we do in science.

So we kind of dug in and were really amazed at how complicated that science is. So there's that one. We've had an interview with Brad Jolliffe talking about working with moon rocks and lunar meteorites. Oh, I liked that. That was a lot of fun. And we also teamed up with another show called Undersampled Radio. And so we had four of us on at once. That was a little hectic to try to manage, actually. But it was a lot of fun that one's called rock drills and beer i feel like that's what we should title this one uh and you mentioned that you are talk about geotech is that the right word yeah so the the technology that we use to do geoscience but now you're teaching a class about

that yes and this is a class that we were developing from scratch and teaching for the first time this semester called techniques of experimental geoscience what are you going to teach in this class well we're trying to cram a lot in, so we'll see how this goes. But when you do things like make laboratory models of some physical system, there are all kinds of skills required. You have to be able to design something mechanical and draw it and communicate it to the machine shop and figure out what you're going to build it out of,

how you're going to build it. There's some mechanical engineering that we need to go over there. You have to do some sensing on it, so you have to know what kind of sensors you're going to need, a little bit about the signal conditioning that you might need, and some of the data acquisition side. So the goal is to give people a broad overview of all of these topics, so they get some basic electronics, some basic mechanics, so that they know what's available and what tools they can use when they're in the lab, though they may not be a specialist in them. It gives them enough to go and talk to specialists, talk to people who do that every day, and have an intelligent conversation to build up their experimental apparatus.

Are you having them build these tools through the class? Is it going to be one of these classes that at the end you have to turn in your neat new scientific equipment? Sort of. So there is a semester-long project that goes with the class. Unfortunately, in science, I've always joked that anytime you write the word geophysical on something, the price goes up by a factor of five at minimum. And in geology, we do everything order of magnitude. So go ahead and call it a factor of 10. And that is a limiting factor in everybody getting to build scientific apparatus for the class. So we're limiting them to somewhere around $100, $50 to $100 budget to build some kind of prototype. Or it could even be a fun project that is not necessarily directly related to their research. But they have to build something where they have to do a little bit of mechanical design, a little bit of electrical design, a little bit of coding. And we're using, for example,

the SparkFun Inventors kits to get them to have some experience with the electronic side and get them started with Arduino throughout the class. And many of them are using that in their final projects. So a lot of these people are probably predominantly scientists and researchers. Just listening to you describe, you know, the contents of the class made me think, are there things that people who are citizen scientists can build and experiments that can be run? Because geology seems like one of those, and I know you're a geophysicist, so I'm not trying to, I don't know if there's a turf war going on between. No, no, it's not too bad. But it seems like one of those things where, ah, you got it in your backyard and you can learn things if you have a little bit of

knowledge. And now people can fabricate just about anything in their house with sufficient, you know, hobbyist equipment. Are there things that, projects and things that citizen scientists can build? Oh, absolutely. I think so. One good one to start off with, so maybe this wouldn't be necessarily geology, geophysics, but earth science would be doing things, of course, like weather monitoring and that kind of thing. But getting more into the geology side, you can put out geophones, and you can measure local ground movement relatively cheaply now. Now, you're not going to get a full seismometer where you can see earthquakes from around the world

because those can see very tiny ground movements, you know, nanometers over hundreds to thousands of seconds. And your little geophone for $60 from SparkFun isn't going to do that. But you can see all kinds of stuff, like a quarry blast locally if there's somebody quarrying near you. You can see that. You can see all kinds of little, like a quarry blast locally, if there's somebody quarrying near you. You can see that.

You can see all kinds of little local events and things. And you can see human patterns, which we often have thought are noise, but turns out we can use them as a seismic source to interrogate what's going on under your feet. Like the diurnal patterns when people are sleeping and so they're not around and when they're walking around and so it's noisier? Or what are their patterns? Yeah yeah so the diurnal ones on campus we've got a seismometer and i did an analysis a little while back we had some construction going on you could tell exactly what time of day the construction workers started when they stopped you could see weekends very clearly you could see football games very clearly because penn State football is a big deal. And you see all kinds of little things that you wouldn't have thought. There's a seismometer near Disneyland you can see interesting patterns on. So there's lots of human activity superimposed on this that normally

we filter out because it's high frequency. But it's fun to look at it. At ShotSpotter when we did gunshot location systems it was always amusing to look at the patterns. I mean, sure, there's fireworks on Fourth of July and you have to get rid of those, but there's a whole bunch of sudden noises that sound like gunshots. Oh, I see, there's a jackhammer. That's not that interesting. Yes, yes. So there's a lot of filtering of things that you have to do to get to the interesting signal. But as we're learning, you know, one person's noise is another person's signal.

It all depends on what you're trying to do. So you limit them to $50 to $100, and they've got the inventor kit. Right. And I'm trying to think, what could i build given that and a little bit more and there's so many sensors out there's there's all these accelerometers that could give you information and there's um the chemistry mem sensors the spectrometers even a cheap camera could be used to monitor something are they overwhelmed overwhelmed with the idea of, oh my god, they're all of these sensors? What will I actually measure? Or do they come in with the idea of,

I really want to know more about how often rocks fall into the ocean or some thing? It's been a mix, really. So we've had some folks that have come in and said you know there's this neat sensor and i want to try to figure out how we use it we have some folks that come in and they have an exact problem that they're trying to solve and in that case sometimes it's too expensive and i say well why don't you build a simplified version of what you're going to do to solve the problem because it is after all just semester project, and we're already coming up on 25% of the way through the semester. And they just had to turn in their project proposals. So, for example, we've had some people talking about building things to move some sensors.

One person from our laboratory moved some sensors on our big BIAX apparatus. Right now, there's a clunky problem of you have to shut things down, run all the way to the back of the machine, move this one sensor a very small amount, and run all the way back around and start things up again. So he wants to solve that by building this bolted-on linear actuator system that's intelligent and prevents you from having to do that. So some of them are more apparatus things like that. Some of them are scientific questions. Some of them are fun projects like automatic plant waterers or automatic beverage dispensers. We really have had the whole gamut from the group of students this year. That sounds like a fun class. I'm really looking forward to seeing what they build and bring in.

Is it in the geoscience curriculum? Yeah, so it's a geoscience class. It's a 597, which is sort of an open course number for graduate classes that you can shove in pretty much any class that you want into that course number. And you're teaching this as a PhD, almost a PhD candidate. Correct. Yes. So my advisor and I are co-teaching this course. And like I said, we're building all the material from scratch. So it's been quite a challenge, but it's all online. It's all free. The lectures are recorded as well. So you can go look at them. We can put a link in the show notes. And it's all hosted, actually. It was just a bunch of restructured text files in a GitHub repository. And then I use a tool, Read the Docs, to auto-generate the website. And there's a continuous integration engine that goes in there and checks to make sure our links aren't broken. And this is an idea that I took directly from some folks at UBC for making these online open textbooks that everybody can contribute to and volumes of class material that are good because they're maintained by a group of folks. That's going to change how education works. Yes.

It's a much different paradigm than we're used to of you buy the textbook that's been in publication for seven editions and is slowly getting corrected or problem numbers changed to this very broad, open set of content. Well, and you can update it

with new pieces of information far more easily than making a new edition. Yes, and there's nothing more satisfying than having a student send a pull request to clarify something that they didn't understand. But textbooks take a lot of effort. It's a lot of time. It's a time suck. Right. And you're giving it away for free right how are you gonna make money

off this unfortunately no so yeah no this is just something that we wanted to do as an open resource because it's a problem that everybody that does experimental work be it geoscience or not has to deal with and by not doing all of the work by just one or two people, by having other people that help contribute, eventually it can grow to be a really nice resource for folks. This is why I love the internet. I mean, it's just, Wikipedia is awesome and people figure out how to teach things. I think the reason we learn so much more now is because we figured out how to teach calculus in high school. And I mean, before it was something you learned

after a lifetime of study. And now we're going to have ways of, you know, if I want to know more about how to do this, there are going to be classes and there are going to be classes that have been improved upon because they were open source and they could be improved upon. It seems like this direction is amazing, more amazing than even what you said about data earlier.

The information is becoming free and better. I don't know if it's better. It will be after the third person teaches with it right there are definitely a lot of problems getting something like this up and going but I think I've learned so much from the community of people that have put things

out online and have put educational material out there for free it would seem a shame to work on this class and then have it live in a notebook on my shelf and not be available to other people and not be able to be improved on. I mean, we've had folks that are not even at Penn State, that aren't even in the country, that are following along with the video lectures and slides that we're posting

and sending us improvements and filing issues and helping us make it a better class. And there's something rewarding with that. Knowing that what you did isn't being forced upon the small group that had to take this class. Yes. But is being actively looked for by people you've never met and may never meet. Yeah, and you know it's being used, right? In a textbook, a university buys 50 copies and, you know, okay, maybe 50 students are reading it,

but this way you get feedback and engagement that says, oh, okay, people are actually using this and it's valuable. And this is one of those things that when I was at SciPy this year, there was a talk by Lindsay Heagy at UBC about this exact concept of they've been moving some of their intro courses to this model of having it all online, all as a Git repository where anybody can look at it or edit it. And I said, wow, that is a fantastic way to look at how we should deal with educational material. So we immediately transitioned this class to that model. Do you think you have the time to do this because you are a graduate student? Do you think as a professor, you would still be able to do this?

That's a really good question. And I don't know, of course, not being a professor, but it does seem like I have more time than many of the folks that are full-on professors that have to deal with, we're writing these grants, we're doing these quarterly reports, we're teaching 700-person gen ed classes. So I would say it would be much harder to develop this if I weren't a graduate student right now, yeah. And how long are you going to be a graduate student for? Probably not too much longer. I will definitely be fully completed by July of 17, but I'm hoping to be able to finish up at the end of this semester and then defend early next. And then after that, professional podcasting? Yes, professional podcasting all the way. Makes as much money as giving away your textbooks for free right no so after that you know i've thought about the kind of the regular route

of going trying to get a tenure track job somewhere after maybe a postdoc or two depending on what the job market looks like at that time but i'm enjoying doing these tool development activities and developing material to go with them and helping scientists be able to do what they do better, that I'm actually looking at some jobs that are more in that vein. So more doing the software engineering side. So being a scientist that mostly is in software. Ah, yes. getting your PhD to become a software engineer. I have heard this story. Although, you'll be doing some pretty specialized stuff. Yeah, and the process of going through this PhD

since I started in 2012 has been absolutely amazing. I've worked with a really great group of people on a pretty interesting set of problems. And it's given me a chance. I think the wonderful thing about graduate school, for me anyway, was it gave me a chance to develop all kinds of different toolkits and be interested in something and go chase that for a while and see if it goes anywhere.

And the freedom to do that has helped me to develop a lot of neat skills that I hope to be able to help other people use. I had a friend who had a PhD say that before he got a PhD, the area of his topic, which had to do with navigation and control systems, it was something he was interested in. I mean, as far as homework went, it was more interesting than other things. But after he got his PhD, it was something he wanted to do for fun. It moved over from chore to passion.

Right. And you've got to be passionate about whatever you're doing because you spend a lot of time doing it and a lot of very frustrating time sometimes. So you have to really enjoy it and be genuinely curious. You also have a consulting company. Is that right? I do.

Correct. Again, creatively named Lehman Geophysical. As I said, we're very good at naming things. So I do a lot of instrumentation for people. It's been the main thing. Do you have time to do much consulting? Not a lot. It gets squeezed in here and there.

Of course, much less so now that I'm trying to get this whole class up and running. But I am doing some development of some instruments and some development of some control systems for people's laboratory apparatus now. So that is something fun to do in the evenings that you're not still working on your PhD. Wow. Yes. I bet it is a busy time for you right now. And so you're both building widgets and advising other people on how to build them, or mostly doing the construction yourself? Mostly doing the construction is what I've found. The advising role has always been more in helping other people model systems in more of a computer modeling way.

When somebody comes to me with an instrumentation problem, it might be, you know, I need this thing to measure parameter X. And by the way, it has to work in 250 meters of water or it has to work for two years and go underneath some horrible system. Or it's going to sit on top of a volcano and be exposed to all these horrible gases. There's this great picture that someone showed me before asking if they could alpha test one of my instruments of a giant volcanic bomb. So just this large car-sized boulder that had landed on and crushed their instrumentation shack previously when the volcano erupted.

Or I've been fortunate enough to get to send instruments to Antarctica, which was great. And unfortunately, I didn't get to go with them, but it was a really harsh environment to design for. Yeah, a little bit. And did they survive? They did. They ran for quite a while up on a glacier, collected very tiny tilts of the glacier. So, you know, a few microradian tilts that turned out to be tidal, mostly. And they ran for quite a while and luckily were successful.

But it's a scary thing. And in geophysics, a lot of times you put these instruments out, you've got this grant to go do this. And you kind of have that moment where you seal the lid or you seal the the container that everything's in to weatherproof it and you hope that it records because you're probably not going to see it again for at least a year yeah this is one of those situations where sending somebody out to check on it because it isn't reporting is difficult very and you know there's not wi-fi everywhere we can't have an internet of things it's funny how much how much of science is very uh sustained effort and difficult effort to produce things

to measure tiny tiny minute values of angles or temperatures or displacements over long periods it's just these look we have to look at this one value it's very small and we have to watch it for or temperatures or displacements over long periods. Look, we have to look at this one value. It's very small, and we have to watch it for a really, really long time. And we'll build up a data set of a lot of that before we can say something. It's all built up of these little tiny things. And everybody thinks a lot of times that science is this grand thing where, oh, I have this massive theory, and I go out. No, it's a lot of really, really hard work.

And building up data sets out of tiny what seem not very interesting pieces of information to get to an interesting conclusion yes it is and as you said all the things that we try to measure are small and in seismology it's a really fun problem because you're looking at something you say okay if I want to fully capture this say an earthquake locally and at a very far distance you're talking about a system that has nine orders of magnitude and amplitude and nine orders of magnitude and frequency content so you know good luck designing the right set of sensors to do that it's you have to have these very targeted goals and measure all these really small things. One of my favorites is I built a demonstration that we use during outreach. So when we have students come through the labs, high school students or tour groups,

and it's a solid cube of granite that's two inches on a side. And I put some strain gauges on it and made a meter that is roughly two foot by two foot. And it's got this pointer, so it looks like a giant analog meter. And when somebody comes up and picks up that cube of granite, you can squeeze it and see how much you're deforming it on the meter. Because rocks are elastic. They're just like springs. That's where all the energy for earthquakes gets stored.

But if you squeeze that chunk of granite as hard as you possibly can, you're making it shorter by maybe a few microns. So a twentieth the width of a hair or so and that's full scale on our meter okay crushing rocks while squeezing rocks i feel like that sherman guy the commercial don't squeeze the granite that doesn't sound right yeah it's it's always fun to show i have more rock questions

and more rock puns um because you know i looked those up but i actually want because we don't have too much more time i wanted to ask you about your work with drones right so i just recently started writing a monthly column for servo magazine on multi-rotors so all kinds of quads tri-rotors whatever and that's been an interesting adventure and it's apparently because i i can't say no to things that's how i end up with all these different i'm glad different things yeah that's so so we just finished building a quad from complete scratch. So it's a wood frame with some aluminum bars and bought all the parts

put it all together. That was a several part series and now investigating fun things like adding meteorological sensors onto it so you can get shallow atmospheric soundings like how does the temperature and dew point change with height above where you are when the thunderstorm's going past or something like that. Or adding FPV cameras to it.

So a lot of fun stuff planned for that article. But you also took apart the little cheap Cheerson CX-10, which I also took apart, but you did different things with. Yeah. which I also took apart, but you did different things with. Yeah, so I actually looked at your teardown of it and looked at the controller and said, oh, that's interesting. I'm going to pull those joysticks, which are just potentiometers, off. And I'm going to instead pipe in my own analog

or digital-to-analog converter where I can control this from the computer was just a simple serial stream and do things like control it with a flight yoke. Like you would use in a flight simulator or a joystick where you have the throttle and a yaw and all of that on the one stick. So that's something that I'm still getting the loops tuned a little bit. So it's not quite as volatile when you try to fly it,

but that's definitely going to be one of the articles. Ah, and I would direct you to look at some of alvaro's articles which i will put in the show notes because he took a different path and used um he didn't use the joystick he controls the cheersons from his computer with a radio. And so you get rid of the joystick entirely and control it there. It was pretty neat. It seems like there could be a lot of fun ways to try to fly this. I keep trying to come up with something

with different kinds of gloves or camera tracking of your hands or something like that. It'd be very amusing to be a conductor with little drones above you. Yes. Cam you. Yes. Cameras. Excellent.

Okay, so the Servo magazine about the drone, can we get to that? Or do we actually have to buy the magazine? I think you have to actually subscribe to the magazine. They have pieces of articles that go up on their website. And then after some amount of time, some of it does go up for free. But I don't know exactly what that timescale is. Okay, here's a question. Do you anticipate or currently use drones in any of your geophysics?

I definitely anticipate it. It is only just now being kind of toyed with by some folks. One of the most interesting applications I've seen is putting a camera on the front and doing the structure from motion inside of places like volcanic craters, where it's hazardous to go and they want a really nice 3D model of what it's actually like down in there. And I could see maybe even deploying some instruments in scenarios where maybe you want to set instruments out on a piece of ice that's going to calve off. And that would be pretty hazardous to do by yourself.

And hanging from a helicopter to do it is also pretty hazardous. So sending a cheap drone out there to do it would be a much better way. Actually, when you put it that way, there are lots of ways that if you're a drone i mean it's it's basically a delivery mechanism and you land it and it's done and that was what you wanted now you have to maybe fetch it or maybe it just transmits the data exactly yeah treat it like a space probe yeah and a lot of these situations where you're trying to send something very far away to a to a hazardous place maybe you don't ever expect to see the sensor back but if you get the data that could be good enough i have seen some video people have taken with um quad cop you know expensive

quad copters in an active active volcano that was spewing magma or lava that was pretty spectacular but they weren't doing science. They were just, how can we get really terrifying images by destroying drones? It's amazing. They didn't lose more than they did. We've had some volcanologists come here and give talks and they end up with these stories like, well,

we had, you know, here's some, some video or we had this cool data, but then it all got melted by lava. Then it all got melted by lava. Then it all got melted. Yay. Whoops.

All right. Contest, you were going to give away some books. Tell us about these books. All right. So I've got two books that I thought would be fun to give away, one on the geoscience side, one on the more technical side, but neither of them hardcore technical manuals because every now and then we all enjoy reading something a little bit different. The first book is called Earthquake Storms by John Dvorak, which is a history of the San Andreas Fault. And it is one of the best kind of pop-sci books that I've read on this topic.

It's got a lot of really interesting history, everything from trying to figure out how fast seismic waves travel by looking at wiggles in a bucket of mercury on the ground. It tells you how far back that goes to modern science. So that's a really fun one. And the second one is the classic Soul of a New Machine by Tracy Kidder. I just read that. What did you think of it? I liked it. It was disturbing to me how little has changed in my industry. It reminded me a lot of some experiences I had at a silicon

startup. So it was kind of a good groan through and wow this is this is very familiar and also a nice snapshot of of uh you know what what companies actually look like when they're trying to do things they probably shouldn't be because they should be impossible yeah it's just it's a really great story about the development of the data general eclipse and all of the pain that went into it. Cool. And I know Hecate just put Soul of a New Machine on their must-read list. So, lots of people are reading it, and I suspect it is good. It is on my to-read list. Excellent. Cool. So we're going to give them away with the whole number thing. And I'm going to ask John for a number after the show ends.

And Chris is going to give me a number after the show ends. And he's not going to give it away by saying it's perfect or anything. And on October 1st, we will give out books. It will be very exciting. I love books. So, you you know giving them out is exciting and you all listeners should send me numbers and i don't know maybe john will choose negative grams number um so feel free to be odd if you'd like to I guess three is also odd. Feel free to be odd. Yes. Isn't that nice?

You get it? You get it? You get it? Odd? Odd. Numbers? Okay. Yeah.

Yeah. Excellent. And I guess that is our show. Christopher, do you have any questions? I just, having done a physics curriculum at one point, I was curious what- You got a master's degree in physics. Be proud of it. It's not very good. I was curious what a geophysics curriculum looked like.

What kinds of classes? I mean, how many classes are standard physics classes? And where do you branch off and stop taking those? So we have a lot of the standard physics, of course. We go through E&M. Some people take quantum. Some people don't. We do generally kind of up to that point.

And then we go into more field-specific, so seismic data acquisition or potential field methods like doing gravimetry and doing magnetics and some other E&M methods. So we branch off and do some of those. But I would say it's it's physics and math squished together with a little bit of geology okay so and probably it sounds like um signal processing is kind of important for later stage stuff because you're processing a lot of data but you're also doing a lot of sampling stuff and um you know, looking at... You have to look at pieces of signals for kinds of features, right?

Yeah. So there's a lot of processing, pattern recognition, automating as much of that as possible, but unfortunately it's not as much as we would often like. It sounds like a background that could be useful elsewhere, too. I think it's very transferable, the skills that geophysicists have into a lot of other industries, and a lot of other industries use geophysics to look at their problems. Yeah. Alrighty. All right, John, do you have any last thoughts, because I have found myself mixed up in a lot of different things from different times, different projects at all kinds of organizations. And it's something that someone told me at one of them, and I wish I could exactly remember who to credit this to. But they said, you can do everything, just not all at once.

All right. All right. That is very reasonable right and it's something to remember when you get very overwhelmed with lots of projects cool our guest has been john r lehman geoscientist at penn state podcaster at the don't panic geocast and consultant through lehman geophysical thank you so much for being with us. Thank you for having me. Thank you also to Christopher for producing and co-hosting and getting a

master's degree in physics. That was important. And of course, thank you for listening. Don't forget to send us your number or your lot of numbers, either one. No, no,

no, just, just one number, just one entry, and you will win the book. Your number is closest to unless you get it exactly. And then you'll definitely win it, as long as you're first. Does that all make sense?

Send it by October 1st, speaking of firsts. Hit the contact link on Embedded FM if you'd like to say hello, or send us your number. If you'd like to sign up for the newsletter or subscribe to our YouTube channel, also hit the contact link on Embedded FM. Subscribe link. Contact us to email us. Oh, we have more than one link? Yeah, subscribe link is to subscribe and the contact is to contact. Go to Embedded FM and look around and maybe check out the blog and all of that.

And then you can hit whatever links you want. It's okay with me. And now a final thought for you from Catherine Hepburn, all people. The thing about life is that you must survive. Life is going to be difficult and dreadful things will happen. What you do is move along, get on with it and be tough. Not happen. What you do is move along, get on with it, and be

tough. Not in the sense of being mean to others, but being tough with yourself and making a deadly effort not to be defeated. Embedded FM is an independently produced radio show that focuses on the many aspects of engineering. It is a production of Logical Elegance, an embedded software consulting company in California. If there are advertisements in the show, we did not put them there and do not receive any revenue from them. At this time, our sole sponsor remains Logical Elegance.