Show HN: I wished for a site with a growing list of math problems, I built it (teachyourselfmath.app)
Good math problems are hidden inside textbooks and online documents. To keep up with all the sources in the world is hard. For someone who just wants to continuously solve problems, finding and going through all the sources feels like a hassle. I wished for a website that could just dump all the math problems available in the world out there. And if I could filter the problems by topics, that would be beautiful.
teachyourselfmath is a side project that was born out of this need. At its core, it is a math PDF extraction engine. The engine has some machine learning going on behind the scenes to extract math problems in LaTeX from any image or document.
A little bit about me: I am Vivek, a software engineer based out of India with a diverse set of interests including math. This project is close to my heart for many different reasons and nothing would make me happier than finding people on the internet who would find this website to be useful.
I’d love to hear your feedback on this. Thanks!
94 comments
[ 4.1 ms ] story [ 170 ms ] threadSome links to math textbooks to help figure out answers would be great.
One of the ideas that came up in feedback is that I could have an AI driven experience to walk a user through the question's topic and give enough pointers to eventually arrive at a solution. I am still thinking about the best way to approach this problem in general.
Thanks for the feedback!
Though, I also went a bit into the other areas to see if I still recalled other areas of math that I haven't studied in years.
When asked to solve and explain something it produced a right answer to a problem with a solid explanation I could easily follow. (Or ask more in depth questions about it).
When asked to push me to the right solution by asking me things / walking me through the steps, it got "confused" and more often than not produced a wrong answer.
Not in the spirit of spoiling the fun or bring any unnecessary tension, I can't help asking: considering that these exercises are being scraped from .pdf sources, would you consider having the source for any given exercise? Of course, it brings problems of exposing possibly copyrighted material. I'm just wondering what your stance is on that.
Glad to hear this!
> Of course, it brings problems of exposing possibly copyrighted material. I'm just wondering what your stance is on that. Given that problems are anyway discussed on the internet in various forms, I wouldn’t see this as a copyright concern. And I intend to keep this list free. Aggregation shouldn’t be a problem, hopefully :)
I think there would be value in knowing a given exercise comes from chapter X of book Y; it might even help track knowledge dependencies, so to say (to solve exercise Z, student probably needs exposure to all chapters between X-3 and X). And it could also be possible to build thorough different levels of exercises in different areas: a set of basic computation exercises (invert this matrix; solve this simple integral) - and these exercises, as you probably know, can be generated easily; some intermediate exercises of theorem applications; and then groups of higher difficulty material (which are probably somewhat more creative exercises).
Anyway, this is really an inspiring project and I hope it brings you lots of joy! and, in case it helps anyone, I've found a good trove of mathematics resources over at https://math.libretexts.org/Bookshelves.
> and these exercises, as you probably know, can be generated easily; Yes, this is on my todo list. To do problem generation instead of problem sourcing.
thank you for all the feedback!
I'm not going to do your work for you and I'm not going to pat you on the back for scraping good content in a sloppy way. So here's a really good example of the problem with what you're doing.
https://teachyourselfmath.app/problem?id=18
I really hope to see this improve because the idea is good. The execution is not
thank you for your feedback!
In either case, you're welcome. I don't mean to be rude :) As I said I think it's a good idea. I just know that poor execution on this will limit the potential a lot.
Basically you mirrored Math's metodology on chained functions :D
i am also intrigued by the idea of enabling search here as pointed out by someone else here [0]
[0]: https://news.ycombinator.com/item?id=39114852
thank you for your feedback!
[0]: https://github.com/viveknathani/teachyourselfmath/blob/maste...
thank you for your feedback!
[0]: https://math.stackexchange.com/
I would like more to see curated hand picked math problems instead of mass imported problems from various sources. Along with solutions and detailed explanations and links to the theory needed to understand and solve the problem.
> Suppose that one digit, indicated with a question mark, in each of the following ISBN-10 codes has been smudged and cannot be read. What should this missing digit be?
Handpicking best problem and writing detailed explanations could be quite hard for one person, but if a lot of people find someones explanation useful and insightful, and vote for it, then one person could handpick from roster of problems with good explanations.
That would also help platform by engaging visitors to not only learn math, but to teach the parts they know well to others. (I also found out that explaining math to others really helps you to understand it better, so win|win)
[0]: https://news.ycombinator.com/item?id=39115376
I wish you had a feature that allowed for bookmarking problems & comments like HN.
https://cemc.uwaterloo.ca/contests/past_contests.html
If you're interested in open sourcing it so collaborators can help contribute to some of the future features you've discussed in the thread here, let us know! I'm sure there are plenty of people who'd like to pitch in.
Technically it is an interesting problem.
But ... just stating the obvious ... you need to be careful with copyright. Make sure to clearly indicate the source of each problem.
Two suggestions that I don't think have been mentioned:
1. It might be nice to at least acknowledge the source of the problems PDF-wise, no doubt somebody put effort into developing then and the supplementary information (assuming the PDFs are open source) would be useful for learning
2. Supporting information for learning the context subject material around the problem would be good: this relates to other comments about curation. This doesn't feel like "teach yourself math" so much as "test your math", which is still really helpful, but if you don't know the material there's nowhere to turn too
Great foundation for an amazing platform, nice work!
Is there a copyright violation? Totally not a copyright lawyer (or any kind of lawyer) but I thought basic facts, mathematical relationships, etc. were not eligible for copyright. At least for something like:
Find the derivative of X^2 + 2x + 3
I think the problem and the solution aren't subject to copyright.
I guess a complex word problem might be though, since the prose part would normally be subject to copyright anyway? Not sure being part of a math problem would change that.
I dunno. I could be totally wrong about this. Any copyright experts around to chime in?
Google's Generative AI feature says the following, but not sure how much trust to put in that.
---
According to Bytescare, mathematical formulas, problems, proofs, and theorems are not copyrightable. This is because math is considered a "truth" and is not necessarily invented but discovered.
However, the content of a math book, including diagrams, word problems, and illustrations, can be copyrighted. According to the Copyright Advisory Network, word problems may fall under copyright, but problems containing only numbers and symbols probably do not.
----
thank you for your feedback!
difficulty of problems will be out soon. though, not in a way where users will be able to assign the difficulty on their own.
thank you for the feedback!
Is there a way to share a single math problem as a link in social platforms?
I'd rather see these problems than all that the algo throws at me currently.