NumPy Tutorial (2022): For Physicists, Engineers, and Mathematicians
ฝัง
- เผยแพร่เมื่อ 20 ก.ค. 2024
- Check out my course on UDEMY: learn the skills you need for coding in STEM:
www.udemy.com/course/python-s...
This from-scratch tutorial on NumPy is designed specifically for those in physics, mathematics, and engineering. In the future, I will be making tutorial videos on all the essential python packages, so subscribe for more!
All code can be found here:
github.com/lukepolson/youtube...
0:00 Introduction
3:43 Array Operations
8:28 Indexing and Slicing (1 Dimension)
15:18 Calculus and Statistics
21:28 Examples
47:18 Multi-Dimensional Arrays
52:22 Functions on Multi-Dimensional Arrays
56:26 Linear Algebra: Matrix Operations
58:33 Linear Algebra: Systems of Equations
59:53 Linear Algebra: Eigenvalue Problems
1:02:02 Examples
1:28:42 Basic Datasets - วิทยาศาสตร์และเทคโนโลยี
![[SUB] 베몬이 교관님 멘탈을 탕탕 후루후루~٩(๑❛ᴗ❛๑)۶ FOREVER 웃게 되는 갓기들의 수련회 | 돌들의침묵 | BABYMONSTER(베이비몬스터) 'FOREVER'](http://i.ytimg.com/vi/FISAO7BwIxE/mqdefault.jpg)
This dude, this channel: is a blessing.
At university I've been taught Matlab but I always wanted to switch to python in order to get rid of licensing problems and fully embrace the "open-sourceness" of this programming language. This video is a blessing!
Believe me, you are one of the most useful channels about python. Thank you so much.
You would deserve much more subscriptions.
What about a cython playlist?
I have shared this link with almost 75 students of mine in the MAchine learning with PYthon class. You have made my job so much easier.. thanks a TON
Super useful and engaging! Looking forward to the scipy and matplotlib ones.
Even after 2 years this is the most fun and informative video for numpy that i needed , really loved those exercises.
thank you for such an amazing content
Very good tutorial, thank you! I was kind of figuring out Python by myself, but this helped me put everything in context. And this finally enables me to get rid of the for loops :'D
It's awesome mate! Your videos about these libraries really help me, and of course these are very useful in my major (physics). Thank you 🙏
Your tutorial are very easy to learn yet very concise, informative and in-depth. Please keep making such contents.
Wow, thank you so much! Because of your explanation at 29:44 I just understood those slice-operations, I allways struggled with, fell like scales from my eyes :D
Great video. Having some clear and complete tutorial on these topics is very useful!
Just a small remark: at 31:31 we have not found 'exactly' the locations where dydx = 0. We have just found the minimum interval available where dydx = 0 is bound to be. This is important because imagine our number of points wasn't ideally as big as 10000, in most cases we don't have the luxury of having analytic functions. And in most cases this makes it sensitive to discretization errors.
So in reality one additional step would suffice to get the most precise answer. This already gives us the number of sign changes in the interval. So we could simply use interpolation to find zeros between the x[interval],x[interval+1] to get a precise answer.
Like many others before me, I'm just chiming in in order to thank you for these outstanding tutorials.
It's not hard to find numpy tutorials on the internet, but most of them are either very basic because of the target audience (and so they can't push the student hard enough) or deeper but still mostly amounting to a showcase of methods, functions and syntax. What I like about yours is that they can be challenging at times for a newbie but still very much rooted in the scientific problem-solving mindset and that makes them priceless and fun to watch and try out.
Exactly what you said. Chiming x 2
Nice and meaty tutorial. Have been going over it several times.
Cross-referenced many ideas and principles with information from other sources.
It can be said that If nothing else, the effort invested in studying the material,
has produced good fluency in typing LaTex Script,
and consequently, textbook-like IPython Notebook files.
Thank You!
This was really informative! You are a genius, thank you!
I like the hands-on exercises/examples, which I followed along in my own desktop application (VSCode). And I learned that "trick" of how to get all x-values when dy/dx = 0. Incredible how something that looks hard can turn out to be so easy :D
Using Python a lot in my engineering career and NumPy is essentiel in numerical calculations!
I guess I will continue to look at your videos, starting with SciPy!
This is the best channel I have come across for python, as an engineering student, I am so grateful! Thank you !!
Amigo, eres un crack! me sirvió muchísimo el tutorial. Gran calidad de videos :)
I hope the algorithm blesses you. Best tutorial for NumPy on the internet hands down.
U got one more subscriber. Just found you by chance, and I appreciate.
Great tutoria and very clear explanations!!!
Thanks a lot mate. Please keep posting. I find these videos tremendously helpful for my Computational Nano electronics course.
This video series is going to be great!
For my entire undergrad and masters I avoided coding like the plague, always disliked matlab and used maple a few times for checking christoffel symbols in black hole studies, other than that I am extremely inept, this tutorial saved my life, looking forward to trying all of your other videos, from one PhD student to another, god bless you
Good luck man, did the same thing. Didn't feel comfortable coding until I was out of college knowing that deadlines and everything were finally behind me. Now that I'm out though I realize what an amazing resource it is. Bet you'll have a badass thesis down the line
Good luck with your student loans
@@annakquinn7084 ???????? What kind of comment is that
@@alboz1327 real ones based on reality
@@annakquinn7084 the comment doesn't make sense. Why bringing up student loans, what does this have to do with anything in a python video
For sure, you're the best Python teacher I have met so far on TH-cam.
Thank you so much. God bless you
Completed today. Thanks for this great resource Luke!
Awesome! There are no more word to say, your lecture is awesome!
This has been fantastically helpful, thanks. I've used python a lot for the last 5 years or so for many projects, but hadn't properly acquainted with numpy. I'm currently exploring analysing captured audio signals for equipment testing, so playing with numpy and fft a bit. The array multiplication tip for finding zero crossing points is just brilliant. I've just implemented it as another way to get a frequency measurement. 👍
So underrated, what a great channel ! Thank you from a physics engineering undergraduate.
I've learned a lot of thing with your videos rather than in Computer physics classes, thank u Bro
Man. So good. I really wish you the best, you are such a great teacher as well. Thanks for sharing your knowledge.
Referred to your scipy and numpy tutorials, excellent content. Thank you!!
This is one of the best numpy tutorial for engineers on youtube!
Amazing stuff and one of the best teachings I’ve ever seen ❤️
Brilliant presentation! You pretty much covered everything important for doing computational mathematics
Thanks a lot and for questions we have to think of. At that places you stop and think :"Am I really remember all the elements? And did I know how exactly to use them?" All the tricks are not just for the exact implementation of the rules , it was to teach with the way of thinking!
Thank you!
This stuff is pure gold. The curvefit video was excellent. Then the turkey cooking demo caught my eye. Now, you have me playing with numpy, scipy, matplotlib and sympy. Fantastic tools for engineers. I tried Jupyter but I don't see the benefit. There seems to be a big overhead when you consider that Python can be run from the most basic plain text editor.
such a great teacher you are! I am learning a lot here
I am migrating from Matlab to Python. Your videos are a blessing. Thanks
This is amazing! I’m sharing this with everyone I know :-)
thank you so much for these videos. please keep up the good work
These tutorials are really great ! Thanks man !
very nice and in depth presentation. You deserve appreciation and subscription. Thanks
Thanks a lot for all this video. It helped me a lot in my final year project.
😊
I like watching this man, because he is thorough and is great with examples in his teaching, but his language is unique!!!
He is LOUD. He yells to learn you the material. He also repeats himself, and he is a big fan of the imperative mood in English. He's always commanding the listener to do something or remember something. He's so funny 🤣😅🤣.
It's gonna hit a million in no time!
Bro you are a gift for data science & machine learning.
better than most of the other video tutorials on numpy that I have seen...
one of the best if not the best explanation, within 1h:30 u can say that you manipulate numpy well... continuing to scipy and plt
U r the GOAT. I already liked it and I still don't even see it. Do you think it would be possible a serie in Optimization Methods using numpy numba, etc?
The illustration and speed is just amazing😎
Thank you very much for providing this type of content to us.
Very concise explanations thanks you are a lifesaver
Thanks a ton. Please keep making your videos. Just excellent.
Amazing algorithm for roots!! I have watched 3 times and I cetch the idea!! You are cool!!!
thank you Mr.P, as a PhD geology student entering into python this video is amazing,
Dear Mr. P Solver, I am very thankful for your Python tutorial videos. I have learnt much more things. Thank you again.
do you know what software he is using in the video?
great young teacher here. fine fast ideas for my use of numpy in my elliptic curve work.thanky P. Solver
Man you are amazing!!
Great work
great! I am back to python thanks to you. what about the Maxwell equations?
Perhaps the best explanation!!
I need more, you're wholesome.
These videos are great, man.
ravel(), compared to flatten(), will often be faster since no memory is copied, but you have to be more careful about modifying the array it returns. flatten() always returns a copy. ravel() returns a view of the original array whenever possible.
Amazing tutorial thank you very much !!
zooming in: at 13:51 you did something most coding teachers don't take into account, which is zooming in; you yourself forget to zoom in at the beginning of your excellent tutorials , but they could be much better if you remembered to zoom in at the beginning of them; ha ha. you know, some of your thousands of spectators use glasses to watch your tutorials, 13:51, thnkyou very much for your channel for your attention and thanks in advance for zooming in, ;) Ah! and removing the side bar, hiding it: everything that increases screen real state
Thank you, Luke. Can you upload the meshgrids diagram , thanks!
Thanks for this great tutorial.
omg awesome content I am learning for my passion in physics
This dude is an absolute legend
Impressive, wonderful python skills.
I enjoyed learning!! Please do more such tutorials. If possible, please share the codes too:)
could you explain why this code doesn't work for the last part of the 1st exercise? x[dydx==0]
@@renukavelu1701 Because it's never exactly equal to 0. We have to find the value for which it is closest to 0
Excellent Tutorial .... Thank you ...
28:00 Wow, that's an amazing trick! Last semester I had a computational physics exam. If I know this sooner, probably my grade would be better
😅
First things first, your channel is really great, I'm an engineering student learning python and you are making that process so much easier. That being said, did you know that you don't need any library to answer question 2? Just use a list comprehension!
sum([i for i in range(0, 10_001) if i % 4 != 0 and i % 7 != 0])
Very neat! Though I suspect this may be faster with numpy, as list comprehension (which is essentially just a for loop) can be slow in python. I would try this for 10 million numbers and see which method works faster.
@@MrPSolver In my tests, there is no significant improvement other than 1 ms, 100,000,000 was enough to take the runtime to over 7 s. If someone test this and find a significant improvement, I'll like to see cuz even myself thought that with numpy would be faster, lol.
I never compliment people but...this dude is fantastic.
Can you please explain this part E = np.swapaxes(E, 0, -1)? Why is -1 the element or inner axis?
Amazing!! Some real stuff!!
54:45 for =anyone who didnt understand, learn aprtial derivatives and integration multivariable one is constant while other is changing and vise versa
your channel is over the top and your are better then my python prof, so thanks . can you also uplode a video about pandas ? to much lab results in a csv file . and again many thanks
Thank you!!
Thanks for having the chapters built into the video. I do have a question though. Do you see a benefit to learning math applications of python like this instead of using Matlab? My school gives us an edu license and I don't know where to spend my time learning.
Despite having access to both, I would still choose learning python over Matlab, mostly because there are so many resources available online for python. That being said, from what I understand, the two languages are very similar. If you spend a lot of time learning one, it should transfer over to the other (not the notation, per se, but the general way you approach problems). But for all intents and purposes, python has everything you need for an undergraduate degree. When I was in undergrad, my school also offered a full Matlab license, but I chose to learn python instead. Also when working with others people's code after you've graduated (grad school, industry) you'll notice that A LOT of code will be written in python.
Without any doubt: python. Matlab is proprietary bullshit and there is a reason Mathworks (the maker of matlab) is giving out free educational licenses to universities: just to trap you in their ecosystem. Stay clear from matlab unless you actually have to.
In the eigenfunction problem, what do we do to make sure the bounardy conditions are fulfilled? They seem to be, but we don't do anything for it.
Edit: it "comes from the matrix shape"; the first line of the matrix is just "-2f1 + f2" (with f0 missing). I see it now :)
Can you please make a video where you write code for a nudged elastic band calculation? You use nudged elastic band methods to calculate the structure and energy of transition state(s) for a reaction. The classic example for a nudged elastic band calculation is determining the structure and energy of the transition state for H2 splitting.
Very good tutorial, thank you! I
very helpful video, thanks for this. very nice
Thanks, and please continue (:
brilliant series, many people say that scientists use NumPy, scipy but you showed "HOW TO DO IT"
at 22:50 if endpoint=True in linspace() why should we consider the N+1 value?
love the camera dude
Hello Mr.P solver, if it is not much could you do an vedio to explain how to write the Mathematical Expressions in symbolic form for example you wrote sigma, epsilon and others as such. I know I am asking much but I will Highly appreciate your effort thank you in advance
You are a legend
At 24:31 , is there any advantage of using the * operator rather than a plain logical & operator? Both give the same result, but, coming from a programming background, I feel using a boolean operator make the meaning of the operation much clearer. Other operator, like | , don't match so cleanly to a multiplication.
You are amazing dude
I think in question 4，the last term of y-axis should be E(t/k) rather than E(tk)
x = np.linspace(1,10,100)
f=vectorize(lambda x: 1/x**2 * np.sin(x))
dxdy=gradient(f(x), x)
that is where the vectorize comes in
Hey man ,you got one new subscriber 😊
I am struggling define function to uniform random distribution, to count the the number of point in circle from the rectangle.
Hi Luke! Thanks for the awesome to-the-point video. I've got a question for you wrt the last part of the first question which was to find the roots. Although the method is clever, it seems there can be some concerns regarding the result. Does this method assume that none of the discrete elements of x are the roots? because if this becomes the case, it seems that the point will be neglected (due to the < 0 condition) which does not seem so elegant. If we change the condition to
Moreover, in question 4, I believe there is no match between the what you asked and what you plotted which I find not that nice. Moreover, I think the last formula in the question ( E(t)=Integral.... ) is not a correct formula and is weird. IMO, there was no need to use t' here since E is a function of t. In the solution, all t variables could be replaced by t' and then the t' replaced by kt.
If none of the consecutive values is 0 then none of the x positions is a root. Assuming the evaluated function is continuous, there has to be one in between (by the intermediate value theorem), and any of the two x positions is an approximation within the accuracy given by the original spacing. This is accepted, of course, and without further information you can at best take the mean of the two positions in order to double the "precision".
The same precision argument is true for the case where you (numerically) hit a zero. You are correct that
This is Awesome !!
Science girls should be in looove
you must have had a PhD in either Maths or Physics, am I wrong?
thanks yu sooo much ,you are awesome .