Not signed in (Sign In)

Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

    • CommentAuthorHarry Gindi
    • CommentTimeSep 28th 2010 edited

    Someone posted this on meta.math.SE as a resource. It seems pretty nifty. It lets you search the internet in TeX.


    • CommentAuthorRyan Budney
    • CommentTimeSep 28th 2010 edited
    It chokes on a search for $\mathcal C_2$.
    • CommentAuthorHarry Gindi
    • CommentTimeSep 28th 2010 edited

    It's an alpha, give it a break. =p

    I mean, as a permanent pessimist, I am expecting the website to fail, but wouldn't it be something if it succeeded?

    • CommentAuthorMariano
    • CommentTimeSep 28th 2010

    Sadly, anything which tries to match against actual TeX, is not going to be much of a success, for people write TeX in the wierdest (and illogical!) ways: matching against that is only going to bring pain.


    Now if it rendered it first and then searched for matching stuff in PDF/DVIs, then that would be something. But, as Mariano says, people write anything in their documents that looks approximately right. Here's something from someone who shall remain anonymous:

    That is to say, there is a natural isomorphism (\qipfunc \redfunc \pfunc{H} \qipfunc \kiqfunc \cong \pfunc{H} \qipfunc \kiqfunc).

    Hmm, it seems to have's choking on any input. Or is it just me?

    It would be nice to know how it worked (or, at least, how it tries to work). For example, I'd hope that at a bare minimum it was clever enough to check a \Z in a .tex document against a list of \newcommands in the preamble defining \Z as \mathbb{Z}. If it can handle things like this, I think there's hope that this will be a useful tool, Mariano's (very valid) point notwithstanding.
    • CommentAuthorCam McLeman
    • CommentTimeSep 28th 2010 edited
    I talked to the author of the project via email and though I would pass on some information.

    Most importantly, uniquation doesn't (currently) use google to search for tex files. It (again, currently) searches only a few select sites (wikipedia, MO, etc.) where LaTeX is supported. This at least partly alleviates two of the concerns mentioned: First, there are no user-defined commands to worry about, and second, texxing is probably a little more standardized coming from people who are putting it on public display on one of these sites.

    The author suggests suggests it would be very helpful to have a "priority list" of algebraic commands to implement -- his intent, and current functionality, had been to be able to get strings involving basic arithmetic (+,-,*,/,sqrt,frac, etc.), so I'm forwarding on this request (in addition to suggesting \mathbb{} characters).
    • CommentAuthorbbischof
    • CommentTimeSep 29th 2010

    and \mathfrak would be good!

    Hi, I'm the author of the service.

    I've added initial support of abstract algebra recently. Now uniquation supports 'mathbb', 'mathbf' and 'mathfrak' commands and 'oplus', 'cap', 'cup', 'supset', 'times', 'otimes', 'cong' ans 'simeq' operators. Also it recognizes several groups like GL and SO. Some examples are listed on the main page.

    As Cam McLeman said it would be helpful to have a "priority list" of commands to implement. Besides a command it should include an example of a expression that contains the command and short description. It is caused be the engine, It considers the tex expression not as flow of commands but tries to extract a math structure, for example it understands '\mathbb{R}[x]' as a ring of polynomials. Also it is better to specify in the description what symbols are constant and what are variables. For example '\mathbb{R}' is a constant (well known notation for real numbers) and the systems threats '\mathbb{R}[x]/(x^2+1)' and '\mathbb{Z}[x]/(x^2+1)' as different queries, but 'mathbf' is not used to ref to well-known set, so uniquation threat it as a variable and queries '\mathbf{A}[x]' and '\mathbf{B}[x]' and even '\mathbf{D}[y]' are equal.