Physics for scientists and engineers

What Hacker News says

1. mardiyahApr 2023
   Without knowing what is your current level it is very difficult to suggest any book.

   If you are talking about textbooks, it really depends on what topic you want to learn (Algebra, Calculus, Mechanics, Electromagnetism…).

   If you want a generalist one, many colleges' choice has proven it's:

   Physics for Scientists and Engineers

   (the latest Edition is better of course) by Paul A. Tipler (Author), Gene Mosca

   Conceptual Physics by Hewitt

   http://www.conceptualphysics.com/

   Show morethread
2. foothebarMay 2022
   Physics for Scientists and Engineers Paul A. Tipler; Gene Mosca

   https://www.macmillanlearning.com/college/us/product/Physics...

   thread
3. GrumpyYoungManDec 2020
   If you're after a MOOC-like experience, I'd recommend CalTech's highly-regarded "The Mechanical Universe" series: https://www.youtube.com/playlist?list=PL8_xPU5epJddRABXqJ5h5... It covers basic mechanics, E&M, and a bit of QM.

   As others have pointed out, you won't really get much out of it without having the necessary mathematical background and doing the physics exercises. The Mechanical Universe textbooks are still available on Amazon. If you prefer a more recent textbook, Serway's "Physics for Scientists and Engineers" is well regarded, as is Halliday & Resnick, though the latter is aimed more at career physicists.

   [EDIT] Regarding the Feynman Lectures on Physics mentioned elsewhere, while Feynman indeed intended it for physics undergraduates, if you read his biographies and the views of his contemporaries, it was famous for being utterly impenetrable for actual undergraduates. (One might even say the FLoP was a flop.) It would probably not be a good idea to use it as your text; I've never heard of any undergraduate physics course that uses it.

   Show morethread
4. kevstevMar 2020
   I see a lot of people recommending Halliday and Resnick, but I used Serway- Physics for Scientists and Engineers in college and that textbook was one of the best I felt I ever read. Its been quite some time since I was in college though, maybe its fallen out of favor?
   thread
5. SteuardApr 2017
   Tom Moore's Six Ideas That Shaped Physics is one of my favorites. It upends the traditional order of topics in an intro physics class: it starts with conservation laws rather than F=ma, both because they're more fundamental (you're not going to use F=ma in quantum mechanics, but you'll sure use energy) and because students taking calculus concurrently will have learned about derivatives there by the time you get to acceleration and really need to start understanding them. Lots of other pedagogically novel features, too.

   Another favorite of mine that might be especially interesting to the HN crowd is Chabay and Sherwood's Matter and Interactions text. One key feature is its tight integration of simulations using "VPython", which students start to use within the first few weeks of class. That makes it viable to have students study realistic forces rather than just the simplest cases. (E.g. adding air resistance to projectile motion is just one more line of code, rather than requiring a course in differential equations.) That also makes it more reasonable for their book to give true equations from the start (like the equation for relativistic momentum rather than just p=mv) and then state the more familiar forms as approximations.

   Randall Knight's Physics for Scientists and Engineers is a much more standard text than those two, but it's still worlds better than most traditional texts that I've used: it changes the old standards in smaller ways, but still incorporating lots of research-based improvements. (Knight's little book /Five Easy Lessons: Strategies for Successful Physics Teaching/ was a great introduction to Physics Education Research for me when I first started teaching, too.)

   Eric Mazur, one of the real leaders of the Physics Education Research community, wrote a book not long ago called /Principles and Practice of Physics/. I haven't had a chance to really go through it in detail yet to get a feel for everything that makes it unique (one big structural change from tradition is that he focuses on students getting rock solid on 1D physics before introducing the complications of vectors), but he's such an expert in this stuff that I wouldn't want to leave his book off of the list. (It's organized in a bit of a weird way, mind you.)

   I'm sure there are others that I haven't thought to list. Debora Katz has a new book that I just heard about that emphasizes case studies, for example. But yeah: almost any of these books are fundamentally different than the classics in some way, and worlds better in my opinion.

   Show morethread
6. candiodariFeb 2024
   You forgot the corollary. What transformers fundamentally reason about is N x partition of input x token embedding size (N = number of attention heads). That's the "latent space" between 2 layers of a transformer, that's what attention produces (which is, in almost all transformers the same across all layers except the first and the last). Now if you look at this, you might notice ... that's pretty huge for a latent space. Convolutional AI had latent spaces that gradually decreased to maybe 100 numbers, generally even smaller. The big transformers have a latent space. For GPT-3 it will be 96 x 4096 x 128. That is a hell of a lot of numbers between 2 layers. And it just keeps the entire input (up to that point) in memory (it slowly fills up the "context"). What then reasons about this data is a "layer" of a transformer which is more or less a resnetted deep neural network.

   But convnets were fundamentally limited to "thinking" in, the biggest I've seen were 1000 dimensions. Because we couldn't keep their thinking stable with more dimensions. But ... we do know how to do that now.

   You could look at this to figure out what transformers do if you radically simplify. Nobody can imageine a 100,000 dimensional space. Just doesn't work, does it? But let's say we have a hypothetical transformer with a context size of 2. Let's call token 1 "x" and token 2 "y". You probably see where I'm going with this. This transformer will learn to navigate a plane in a way similar to what it's seen in the training data. "If near 5,5 go north by 32" might be what one neuron in one layer does. This is not different in 100,000 dimensions, except now everybody's lost.

   But ... what happens in a convnet with a latent space of 50,000? 100,000? 1,000,000? What happens, for that matter, in a simple deep neural network (ie. just connected layers + softmax) of that size? This was never really tried for 2 reasons: the hardware couldn't do it at the time, AND the math wouldn't support it (we didn't know how to deal with some of the problems, likely you'd need to "resnet" both convnets and deep neural networks, for example)

   Would the "old architectures" just work with such an incredible massive latent space?

   And there's the other side as well: improve transformers ... what about including MUCH more in the context? A long list of previous conversations, for example. The entire text of learning books things like a multiplication table, a list of ways triangles can be proven to be congruent, the periodic table, physical contexts, the expansion rules for differential calculus, "Physics for scientists and engineers", the whole thing. Yes that will absolutely blow out the latent space, but clearly we've decided that a billion or 2 of extra investments will still allow us to calculate the output.

   Show morethread