- Видео 21
- Просмотров 59 150
Talk Julia
США
Добавлен 1 янв 2022
A weekly podcast devoted to all things Julia. Join hosts David Amos and Randy Davila every Friday to learn about Julia programming news and resources, listen to us share our experience learning Julia, and to connect with the Julia community.
✨ HELP SUPPORT THE SHOW ✨
If you get value from listening to our podcast each week and would like to support us, you can help us out for free by simply telling someone you know to check out our show.
But if you'd like to support us financially then... Wow! Thank you so much 🙏🏼 Here are two ways to do that:
Tip us or set up a recurring donation on Ko-Fi 👉 ko-fi.com/talkjuliapodcast
Set up a recurring donation on Liberapay 👉 liberapay.com/TalkJulia/
For other ways to support the show, see www.talkjulia.com/donate
✨ HELP SUPPORT THE SHOW ✨
If you get value from listening to our podcast each week and would like to support us, you can help us out for free by simply telling someone you know to check out our show.
But if you'd like to support us financially then... Wow! Thank you so much 🙏🏼 Here are two ways to do that:
Tip us or set up a recurring donation on Ko-Fi 👉 ko-fi.com/talkjuliapodcast
Set up a recurring donation on Liberapay 👉 liberapay.com/TalkJulia/
For other ways to support the show, see www.talkjulia.com/donate
Behind The Scenes of JuliaCon 2022 With Valentin Churavy
Valentin Churavy joins us to talk about the challenges facing organizers of virtual conferences and what he and the JuliaCon 2022 organizers did to meet those challenges and pull off a successful virtual conference.
We also chat about some of our favorite talks from JuliaCon 2022, and dig into Valentin's background in cognitive science and distributed computing.
Episode links are available in the show notes on our website ➡ www.talkjulia.com/21
EPISODE CHAPTERS
00:00 - Hello
00:18 - Who is Valentin Churavy?
05:44 - Language design and applied psychology
17:19 - Organizing JuliaCon 2022
21:44 - How many people will come to JuliaCon 2023?
25:13 - Random social interactions at virtual conferences
29:4...
We also chat about some of our favorite talks from JuliaCon 2022, and dig into Valentin's background in cognitive science and distributed computing.
Episode links are available in the show notes on our website ➡ www.talkjulia.com/21
EPISODE CHAPTERS
00:00 - Hello
00:18 - Who is Valentin Churavy?
05:44 - Language design and applied psychology
17:19 - Organizing JuliaCon 2022
21:44 - How many people will come to JuliaCon 2023?
25:13 - Random social interactions at virtual conferences
29:4...
Просмотров: 687
Видео
From BASIC to Julia: Viral Shah’s Julia Journey
Просмотров 1,5 тыс.2 года назад
Viral Shah, one of the co-creators of the Julia language, join hosts David and Randy to talk about how he got into programming, what led him and his fellow co-creators to create Julia, and his thoughts on the present state and future of the language. Episode links are available in the show notes on our website ➡ www.talkjulia.com/20 EPISODE CHAPTERS 00:00 - Hello 00:34 - How Viral Became a Comp...
Is Julia Better Than JAX For Machine Learning? | Talk Julia #19
Просмотров 2,9 тыс.2 года назад
David and Randy respond to an article that makes the case for JAX over Julia for machine learning, particularly when applied to solving differential equations. David also shares a series of workshops hosted by the Julia Gender Inclusive community, as well as a new package by Elias Carvalho for creating truth tables from Julia expressions, and Randy explores a RUclips series and set of Pluto not...
Plotting in Julia (Plots.jl, GadFly.jl, and More) | Talk Julia #18
Просмотров 4 тыс.2 года назад
David and Randy explore Julia's plotting ecosystem and discuss plotting in Julia using Plots.jl, GadFly.jl, Makie.jl, VegaLite.jl, and more. Episode links are available in the show notes on our website ➡ www.talkjulia.com/18 EPISODE CHAPTERS 00:00 - Hello 00:38 - Plots.jl 22:24 - Julia Visualization Libraries: Which is Best? 23:07 - GadFly.jl 28:35 - VegaLite.jl 30:00 - Makie.jl 35:14 - Algebra...
Julia for Data Analysis (with Bogumił Kamiński) | Talk Julia #17
Просмотров 2,4 тыс.2 года назад
Bogumił Kamiński discusses his new book Julia For Data Analysis (Manning) and his work on the DataFrames.jl package. Get a permanent 35% discount on Julia For Data Analaysis and every other Manning product 👉 mng.bz/pOzw (use code podtalkjulia22) ✨ How to win a FREE copy of Julia For Data Analysis ✨ Share this page on Twitter with the hashtag #JuliaForDataAnalysis and @ mention us (@talkjuliapo...
Fast Optimization Using JuMP.jl (with Miles Lubin) | Talk Julia #16
Просмотров 1,7 тыс.2 года назад
JuMP.jl is a an optimization library written entirely in the Julia language. And it's FAST! JuMP co-creator Miles Lubin join hosts David Amos and Randy Davila to discuss the inspiration behind JuMP, how JuMP become a NumFOCUS sponsored project, the journey to JuMP's version 1.0 release, and the future of the project. Episode links are available in the show notes on our website ➡ www.talkjulia.c...
The Julia REPL is AMAZING (with Miguel Raz) | Talk Julia #15
Просмотров 1,3 тыс.2 года назад
Once you experience the Julia REPL, it's hard to go back to anything else. Join hosts David Amos and Randy Davila as we explore the Julia REPL with Julia's official REPL stan Miguel Raz. Miguel shares his Julia story, how Julia is being used in undergraduate courses at Mexico's UNAM, reflects on the willingness of the Julia community to collaborate with other language communities, and teaches u...
Making New Math With Julia and Christy.jl | Talk Julia #14
Просмотров 8612 года назад
Talk Julia co-host Randy Davila talks about his automated conjecturing program Christy.jl and how he uses it to generate new research problems in graph theory and mathematics using artificial intelligence. Learn how automated conjecturing works, how Randy got into it, and how Christy.jl can be used to generate problems and potential relations in other domains, too. Episode links are available i...
Serving the Julia Community with Logan Kilpatrick | Talk Julia #13
Просмотров 6872 года назад
Logan Kilpatrick discusses his role at the Julia Project, his journey to becoming a Julia developer, and his thoughts on how we can better serve the Julia community and help boost adoption. 🔥 Get early access to future episodes ➡️ ko-fi.com/talkjuliapodcast/tiers 👀 ABOUT THE SHOW Talk Julia is a weekly podcast devoted to the Julia programming language. Join hosts David Amos and Randy Davila as ...
Easy Input and Beautiful Output with Julia | Talk Julia #12
Просмотров 2,4 тыс.2 года назад
This week on Talk Julia, David and Randy share and explore resources from around the Julia community all about file input and output and beautiful terminal output. You'll learn how to read and write files in Julia, work the the filesystem, and see how packages like FIGlet.jl and Term.jl can take command line applications to the next level. Episode links are available in the show notes on our we...
Setting Up VS Code for Julia | Talk Julia #11
Просмотров 17 тыс.2 года назад
Visual Studio Code (VS Code) is a popular code editor from Microsoft, and the Julia VS Code extension makes working with Julia in VS Code a real treat. Learn how to install Julia, VS Code, the Julia VS Code extension, and how to use the Julia VS Code extension features to create a robust and fluid Julia development environment. Episode links are available in the show notes on our website ➡ www....
Deep Neural Networks in Julia With Flux.jl | Talk Julia #10
Просмотров 4,6 тыс.2 года назад
David and Randy explore deep neural networks in Julia using Flux.jl by recreating Grant Sanderson's model for predicting handwritten digits in the MNIST data set. We also show how to visualize model results and training performance in TensorBoard using the TensorBoardLogging.jl package. Episode links are available in the show notes on our website ➡ www.talkjulia.com/10 EPISODE CHAPTERS 00:00 - ...
Learning Flux.jl from a Tensorflow Background | Talk Julia #9
Просмотров 2,6 тыс.2 года назад
Flux.jl is Julia's elegant machine learning library, but it's API is a little different than Tensorflow or PyTorch. This week on Talk Julia, David and Randy dive into Flux.jl, explore some of the big differences between Flux and Python's machine learning libraries, and offer up some tips and tricks for learning Flux if you're coming to it from another ecosystem. Episode links are available in t...
Getting Started With Reinforcement Learning in Julia | Talk Julia #8
Просмотров 2,1 тыс.2 года назад
David and Randy take a look at reinforcement learning in Julia by diving into the ReinforcementLearning.jl package. We talk a little bit about what reinforcement learning is, as well as our thoughts on ReinforcementLearning.jl's design, which taps into Julia's multiple dispatch system. Episode links are available in the show notes on our website ➡ www.talkjulia.com/8 EPISODE CHAPTERS 00:00 - He...
Are These The Best Julia Packages? | Talk Julia #7
Просмотров 1,7 тыс.2 года назад
This week we asked Julia Twitter what their favorite Julia packages are, and the responses don't disappoint! We picked two packages to investigate this week: Randy dives into DataFrames.jl and David explores the amazing Revise.jl package. See the Twitter thread here 👉 talkjuliapod/status/1490786858858696705 Randy also shares an example of building a single neuron machine learning mo...
Exploring Graphs.jl and JuMP.jl in Julia | Talk Julia #6
Просмотров 1,9 тыс.3 года назад
Exploring Graphs.jl and JuMP.jl in Julia | Talk Julia #6
Functions, Methods, Structs, and Style Guides | Talk Julia #5
Просмотров 4,5 тыс.3 года назад
Functions, Methods, Structs, and Style Guides | Talk Julia #5
The Julia Language is Music to Our Ears | Talk Julia #4
Просмотров 1,1 тыс.3 года назад
The Julia Language is Music to Our Ears | Talk Julia #4
Penguins, Packages, and Polluting Namespaces | Talk Julia #3
Просмотров 1,1 тыс.3 года назад
Penguins, Packages, and Polluting Namespaces | Talk Julia #3
Jumping Into Julia From Another Programming Language | Talk Julia #2
Просмотров 2,5 тыс.3 года назад
Jumping Into Julia From Another Programming Language | Talk Julia #2
Welcome to Talk Julia | Talk Julia #1
Просмотров 1,5 тыс.3 года назад
Welcome to Talk Julia | Talk Julia #1
Great Video!
talking about chaining and not mentioning R (the pipe) shows quite a lack of existing technologies, unfortunately.
29:03 are you sure this is juliamono? I installed it and it does work with == and => and so on, but not yhat or x_i
this Tutorial is amazing ! , it really helped me on one of my assigments.Thanks
0:15 "... we'll get back on a regular cadence..." When you hear this in a podcast you can be sure that you'll never hear from the podcast again... as is the case with Talk Julia.
😕
I like this podcast. Would be great if they pick it up at some point, again.
I love julia and that IS BECAUSE of JuMP. :)
finally i understand difference between using and import!
Alas, as of Oct 2023, the blog has only two posts (both Julia), although nicely one is a geospatial mapping post.
Is there any new video? I'm still waiting
Gracias. Muy bueno el contenido. Me llamó la atención el revuelo que causa el uso de letras griegas. Yo lo veo con simpatía por la estética que aproxima lo escrito a lápiz y papel en matemáticas y la programación. Entiendo que Edelman es matemático, co-creador de Julia. Podría ser su impronta. Preguntémonos por la deuda de la informática a las matemáticas. Es tan poco el pago con usar la letra eta o cualquier otra del alfabeto griego. Además, es una pequeña muestra de cultura general reconocer esos bellos caracteres que ayudan a comprimir conceptos abstractos con sentido.
Should you expose your students to Julia or teach them more python you ask? I think you should ask for (quantitative) feedback after each course. What do your students think? That is what matters most. That being said, MIT students do have courses in Julia. Make sure
Its funny watching these again now that I'm a full time Julia programming. The style of my code examples in Talk Julia videos is definitely Python styled and not true Julia styled 😂. We need to make more videos with our new knowledge.
Thank you! I have been struggling with that, reading the docs and not grasping it and in general practically be really confused. I have been installing packages in base and then stuff went wromg oj different PCs and honestly gave up several times
This is so helpful -- thank you!!
As far as I understand this weekly podcast has stopped its life right?
Not yet! We have been busy working but have recorded a new episode and hopefully will post soon!
@@TalkJuliaexcellent content, waiting for more !
It's a great video and really helpful. Can you tell me how can I evaluate the performance of a neural network after training on a testing dataset using Flux.jl
I need some help. I have installed "JuliaMono" and the examples at ruclips.net/video/KesuPOlBB_o/видео.html (line 3+4) work. Unfortunately not the line 1+2. What exactly do I have to do to make this work?
same issue
JuliaUP is great!
Nice job!
It might be easy to make fun of COBOL but a quick search on the internet shows that still today 95% of ATM transactions involve COBOL. There are millions upon millions of lines of COBOL still running in the world. Probably for a couple reasons. 1. It's hard to replace all that with any new language no matter how easy the language is to read or write. 2. COBOL might be the best language to use for some purposes. Yes, it's verbose. It was intended to be somewhat self documenting and it is. Everyone always makes such a big deal of the white space in Python, but it always just looked normal to me. That's the way we were trained to write it. That's the way everyone I ever worked with wrote it. I can't imagine writing code of any language any other way. It's much more maintainable that way.
I cannot get Julia REPL on typing 'Julia' as in ruclips.net/video/KesuPOlBB_o/видео.html. The message I get is 'zsh: command not found: Julia' The executable is correctly filled inside the Julia: Executable Path of VS Code preferences.
When people talk about 10x engineers. They are talking about people who have found the Julia REPL. It's such a huge force multiplier.
Great video. Is it possible to view plots when running the code in a "New Process", it only appears to produce plots in VS Code up when running the code in the REPL. Maybe a different backend for plots would do it.
Thanks for making this video. Here are my comments regarding the questions you had for the viewers: I use `import` almost all the time, unless I use methods from a package a lot in my current project or package (e.g., functions from LinearAlgebra). Like Dave mentioned in the video, It has to do with the cognitive overhead of remembering which function came from which package. I feel more comfortable having more control of what is in my current global namespace by doing `import`; this is just a personal style preference. For teaching, I suggest saying "we now load package XYZ from the active environment", instead of "we now import/using XYZ". This forces the students to remember that they need to add & instantiate XYZ in their active environment, and make sure the correct environment is activated for the task at hand.
Great episode as always! I tried to find the repo for Christy.jl but couldn't find anything. Is it available somewhere? Thanks again for the great podcast!
In a non-package form you can clone: github.com/RandyRDavila/Christy.jl but I encourage you to checkout: github.com/RandyRDavila/Conjecturing.jl Once there, run the example julia scripts and see some conjectures 🙂
@@TalkJulia thanks a lot! I'll definitely have a look at it
Thank you for this behind the scenes view on Julia and for the introduction to Valentin Churavy. Super interesting. Huge challenges, I have come to understand. I agree with the point you make about the importance of a single digital platform (e.g. Gather only, or Discord only). The JuliaCon Gather was fantastic but, like, no one was there. The map interface was far more friendly than Discord, so I would only encourage it - unless Discord introduces the map (game-like) interface.
Great video, thanks! I would also add one my experience: After opening Julia: Start REPL (for example with ALT+J ALT+L) and trying to run the Debug, a new terminal was opening useing again the base enviroment and not the workspaceFolder enviroment. I fixed this problem by changing the "juliaEnv" attribute in launch.json file as this: "juliaEnv": "${workspaceFolder}"
Great talk guys, preatty step by step across different pkgs. Which is the page for the colors selection?
Thanks, Mauricio! Glad you enjoyed the talk! The colors reference is at juliagraphics.github.io/Colors.jl/stable/namedcolors/.
If you are interested in JuMP used for solving traveling salesman problem, have a look at ruclips.net/video/zkF69rF6EkU/видео.html (latest of series of 3 videos)
Valentin is the best ♥️
TalkJulia Is back
I'm glad TalkJulia is back! Great video as always!
Thanks, Elias! It’s great to be back!
great video. really well organized and easy to follow.
This video was awesome and really helped me understand.
It’s “functional” to skip return from functions as functions need to return values. Nonfunctional style programmers find it odd but then they expect functions to not return values as well. Those are considered side effects that are discouraged in functional style. From that perspective, “return” is superfluous except in conditional construct. Returning values also allow for chaining or piping, which is the reason why Julia packages work very well together.
Maybe I'm wrong, but it seems to me that it could also be useful to game developers. Say a video game that keeps evolving interesting challenges?
Finally a tutorial that actually works. Subscribed.
Dude no kidding.
This was quite enjoyable and informative thanks!
OMG! is the pluto output sane now? does it show below the cells like 100% of the rest of the world’s agreed upon standard for command output? I’m gonna cry!
I’m sorry I really want to like Julia but there’s some eccentric quirkiness in this community which is a turn off for me. This whole “eta” symbol thing is an example of that. Another example is Pluto, where the output shows above the cell and the developer is super resistant to a plethora of users begging him for years to give them an option to change this but he refuses to listen and all of Julia community stands behind him and this shows that there is this weird quirkiness which will eventually not be a good thing and i don’t want to invest any time in an ecosystem that actively ignores its users for the sake of “opinionated ness”
So you don't want to learn a language because of has a weirded symbol in it (?), And an editor, which you are under no obligation to use ( like really, you can use any editor), has some developer with strong opinions? Have you looked into why Julia is different as a language, what it can offer ? Sure, I am on the fence about Julia as well, I have python stockholm syndrome - but one would be really shortsighted to just ignore a language because "it didn't feel good when I looked at it".
@@f3arbhyTL;DR: Julia seems like a really nice language. The community is a little too nice and that usually becomes a breeding ground for sociopaths and capricious techbros; something which eventually runs the whole project into the ground. I think you should re-read my comment carefully. My problem is with the attitudes, not with the language. One, the community is too permissive and "nice" for potential bad actors who act in the name of "innovation and opinionated-ness". Use of these symbols is a gimmick and should not be encouraged. It is also an equity issue. A text based source file is more accessible to a larger audience. I don't want to make my variable names emojis either, for similar reasons. Not being able to use fancy math symbols is NOT what's holding Julia back tbh. The second criticism is also about community and certain developers being disrespectful towards the users. This is the case of Pluto.jl where despite decades of convention of putting output under the input (every single interactive program out there,) the developer (who seems like a very nice and quiet person btw) arbitrarily decided to put the output _above_ the cell. This causes all kinds of issues when it comes to UX. He could've given users the option to configure this, but no, he forced this "opinion" on everyone without so much as a fig leaf of a consultation process. The Julia community, despite this anti-social and disrespectful behaviour by the developer, continues to support this choice meekly and without protest (spinlessly would be too strong a word.) Pluto is not "an editor" btw, and appears to be destined to become the standard notebook platform for Julia. So, it's not really a personal choice for long when Julia community will be spending all its time improving Pluto instead of making Julia related enhancements to Jupyter (which ironically, was there first to support Julia , the "Ju" in the name represents Julia btw) . If I want to work with Julia in data-science, then I have to deal with Pluto and the (frankly, boneheaded and capricious) decisions made by one person without any community oversight. I refuse to be held hostage to individual developer whims. Using a language is not about just the technical merits, it is also about the community. If the community stands by silently when certain developers (despite how "nice" they seem) impose their will -- contravening decades of UX standards and best practices -- in an arbitrary and capricious manner, and refuse to listen to repeated pleas by the users to at least give them a configuration option to choose the output method, then I think there's a deeper issue there, and I won't invest any time in a language where the general prevailing attitudes of people in charge are degenerative and harmful to the long term viability of the language. I've been around the block long enough to recognize such things as I've lived through some of these in other language communities. Just look at how LISP destroyed itself.
@@laalbujhakkar I understand what you said, and it does make some sense. I have never used Pluto - the Jupyter integration was good enough and honestly, the repl of Julia is good enough for most prototyping. As for the UI design choices, I was looking into an issue on the Pluto GitHub issue, and it was honestly a mix bag of emotions - but it is really an ass backward way to not include an option to choose between top and bottom output. I wholeheartedly agree. As for symbols, this is where I kind of disagree. One of the main highlight of Julia is that it is more tailored for the scientific community than anyone else. It replaces the idea using python+ compiled language for performance gain paradigm which is so prevalent in the scientific community. And most often, what we would do is reproducing/ translating academic paper into libraries. To me, and I am sure it is the case of everyone who does this - the existence and support of mathematical symbols with its intended meaning makes this whole experience more enjoyable. Imo it is more clean and elegant. So it is a design choice by the Julia community to target those relevant audience who finds meaning in having those symbols. As someone said in a blog post - Julia is designed and maintained by academic "programmers". The lack of professional programming design choices are clear - but that is something we can improve upon.
It's not really quirky to use η for the learning rate, it's the symbol that has been used for the learning rate in the papers for a long time. Julia supports unicode symbols, there is nothing wrong with using them when it makes sense (in this case, it does). The symbol "η" is a lot shorter than "learning_rate" and more descriptive than the potentially ambiguous "lr". If anyone is confused about what η means in this context, the docs are always easy to get to in Julia. Frankly, I really don't get what you're on about. If any language is needlessly opinionated, it's Python. If you have a problem with Pluto.jl, that's one package and not the whole language. My experience with Pluto.jl has been much more enjoyable than the bloated mess that is Jupyter. Julia is really not "quirky", the community is very pragmatic from what I've seen. The Julia community is less quirky than, say, the Rust community. Julia is principled where it makes sense, and pragmatic where it makes sense. If it was a "big idea" language, it wouldn't be multi-paradigm.
@@laalbujhakkar The whole output above or below the cell issue in Pluto.jl is just a matter of opinion, though I do agree it should be configurable. Pluto.jl didn't have things for a long time I wanted, that I thought would never be added, but eventually were. For example, being able to open a notebook without everything immediately running. So I don't know for certain that the feature you want won't be eventually added as well. Whether the output of a cell is above or below it just isn't that big of a deal to me, I don't think it has anything to do with "decades of UX standards and best practices". It's just your preference vs someone else's. Neither is objectively correct. Also, you don't seem to realize you don't need to use Pluto.jl. Jupyter is well supported as you well know, there is no "standard notebook platform" for Julia. The idea of one monolithic thing being "the standard" is a Pythonic and anti-Unix way of thinking that is best abandoned in the Julia sphere. Julia is more about composible and extensible code. It's just that many prefer the lightweight and unbloated Pluto.jl. It is nowhere close to the only serious option. In fact, Pluto.jl was originally made explicitly for teaching purposes and not as a scientific notebook thing. It just happened to evolve into a scientific notebook tool because people like it.
BASIC is the best first language for beginners.
Of all the languages I’ve programmed in, BASIC, FORTRAN, C, Assembly, ADA, C++, and MATLAB, I’ve never encountered a language that was as easy to get thoughts into code as in Julia. Looking forward to when our industry adopts it more for real-time embedded system simulation and design and execution.
QBasic was my first language too in the mid 1990's. Fortran later then C, perl and now R, python. I still fondly remember those QBasic days as it was quite easy and versatile
What a great interview with Viral!
Good to know that Viral uses Emacs! I know VSCode is very user-friendly, but Emacs is such a versatile tool... it would be great if Julia ecosystem in Emacs got a little bit more of love. (The same goes for vim/neovim, of course)
Thank you so much for mentioning TruthTables.jl in this episode!
Parabéns!!! Show esse pacote criado por vocês.
Invaluable insights, thanks a lot for these podcasts!
Thank you for this interview. It was inspiring to see you all discussing the possible future impact of Julia.