This is pretty neat. It's annoying though that this doesn't specify where it searches. When this was discussed earlier on Reddit[0], the person who made it said that it (as of 2 months ago) searches Wikipedia, Wolfram MathWorld, NIST DLMF, MathOverflow and PlanetMath, but that should probably be specified somewhere on the site. And as lots of people have already said, to be really useful this should search arXiv.
But this is nitpicking. This is still the first time I've seen someone make something like this, so, again, really neat!
Good job! I have a similar project - http://uniquation.com - which also is based on a thesis work.
I failed to attract users, lost interest and abandoned it several years ago, nevertheless it has the best (among math search engines) support of equivalence between formulas. For example, the following formulas are equal to uniquation: "x_{n+1} = r x_n (1 - x_n)" and "k y_i (1 - y_i) = y_{i+1}".
I'm going to open source it and maybe it can be merged into your project. What's about your project if it is proprietary or open source?
I have long dreamt of a reasonable open source math search engine. My wife is getting a graduate degree in computational linguistics, and whenever she or one of her friends doesn't know what project to do, I try to convince them that mathematicians would love to have one.
In other words, I'd appreciate it if you (and others) open sourced it.
It's pretty smart: x^{n}+y^{n}=z^{n} - the standard formulation of Fermat's last theorem - returns mentions of it.
Now if I search for b^{n}+c^{n}=d^{n} which I shouldn't except to see anywhere, it returns results for a^{n}+b^{n}=c^{n}. It would have probably been better if both results were merged but still!
You should try my pet project based on my thesis - http://uniquation.com It works as you expect. The sad part I aborted it several years ago, the good part that I'm going to open source it this year.
It would be cool if you could search arXiv.org with this or maybe even subscription math/physics journal sites like JSTOR. Paid search results might be an opportunity for partial funding but I know that's not everyone's cup of tea.
We, at KWARC, developed a search engine capable of searching arXiv, see [1]. It can search latex-style formulae and CMML formulae.
In addition, one can also use "query variables", which means that if I search for, e.g., "?a + ?a" (?a is a query variable), I will get all sums of two equal terms.
?a could be an integral, a fancy function, or just a one-letter variable.
A search for "tensor" gives a java.lang.NullPointerException. I realize this may not be intended use case, but I certainly won't be the only one typing in text.
Wow this is wonderful. I'll definitely use it now and then.
Some ux nitpicks:
* I dislike the onhover fold/expand. That half second waiting for it to expand makes the interface feel slow and frustrating. The best thing imo would be if it was always expanded. The next best thing would be to make it expand instantly.
* If you type quickly in the search box, the display lags behind. A common way of fixing this is to detect fast typing and disable rendering until typing stops or at least becomes sufficiently slow.
This is pretty cool. It would be awesome if at some point we could enter any problem and find generalized approaches towards solutions, but enabling search of this kind must count as a first step in that direction. Bravo.
For example: I'd like to be able to look into understanding 4D objects (hypercubes), but I'd have a hard time finding trivial (recreational) puzzles about them, let alone their solutions and mathematical formulations.
I'm doing lots of math searching lately. I'm not sure there's any formula that would really help. There aren't that many really formulas going into whatever advanced math theory.
Plus setting up a single formula to search for looks like the usual ordeal whereas typing some key words is really easy. It would be nice to some detailed logical and/or sorting, though Google scholar overall does a decent job.
Bug report: sticking Unicode characters in the query URL results in a NullPointerException. (Blame a momentary lapse of thought for my trying to do that in the first place.)
But this is nitpicking. This is still the first time I've seen someone make something like this, so, again, really neat!
[0]https://www.reddit.com/r/math/comments/2po75u/searchonmath_a...
Edit: See rhambasan's comment which has a link to an arXiv math searcher; I haven't tested it to see how it compares.