Re: [Maxima-discuss] run of the mill... was Re: Cheating Around Maxima Latex

[Stavros M. <mac...@al...>]

Yes, it's often the really simple, basic things that trip users up -- like
input format -- rather than the hard cases that typically motivate us
insiders.

Another example: math jockeys get excited about having a *solve* function
that uses the Lambert W-function when necessary and provides all solutions
to *2^x=1* (*solve* finds only *x=0*, though *%solve *returns all of them),
but for most users, it would be far more useful to provide numerical
solutions to numerical problems, and to find the real solutions of
*(2/3)^x=5* and *4^x+2^x=20*.

                -s

On Mon, Oct 26, 2020 at 11:35 PM Richard Fateman <fa...@cs...>
wrote:

> Just a brief observation about what "run of the mill" might be.
> Some years ago I had a project, Tilu, that hooked up a web page to an
> integral table
> lookup program (a program much less comprehensive than Albert Rich's Rubi).
> We kept track of what people tried to integrate using Tilu.  The vast
> majority
> ("run of the mill" perhaps) were variations of sin(x).  Like
> sinx,   sin x ,   sin ax, sin[x].
> So we changed our parser so that any of those forms were acceptable.
>
> Consequently, a really weak system, but one bolstered to take erroneous
> input, could answer a vast majority of the cases.  Like a CAS that can
> do high-school algebra will do the vast majority of problems, because
> they are from high-school students. Except the students will not
> be able to type syntactically correct input, so there's THAT problem.
> So we looked at stylus-based input; an interesting problem with
> a number of partial solutions out there.
> I recall that there is one system that promises to do your
> homework problem -- all you have to do is send it a photograph
> of the problem...
>    So I generally think of success at "run of the mill" problem solving
>   as a surprisingly low bar, algorithmically speaking.  This is not
> terrible, just so different from ambitions of CASs.
> RJF
>
>
> On 10/26/2020 4:59 PM, Robert Dodier wrote:
> > On Mon, Oct 26, 2020 at 3:57 PM Raymond Toy <toy...@gm...>
> wrote:
> >
> >> I agree. I want Maxima to produce the correct answers.  Then it's up to
> me to have the TeX form be in the format I want. For example, it's
> impossible for Maxima to know if I want \frac or \displayfrac.  Or what the
> order of the variables should be when I format it.
> >>
> >> I pretty much expect to have to do some editing of the TeX form to
> produce what I want.  And what I want may not be what you want.
> > I dunno, this feels too pessimistic to me, the situation is really
> > more rosy than that. Maxima can already do a decent job with the vast
> > majority of run of the mill expressions, and something comprehensible
> > otherwise, and if Maxima's tex output is really lacking in some way,
> > by all means, let's fix it.
> >
> > Maxima's tex output is the basis for the stuff that's displayed (via
> > MathJax I guess) by Jupyter, and it's more than workable. If I were
> > going to publish a paper, sure, I might tweak the output, but for
> > everyday stuff, Maxima's tex is quite sufficient.
> >
> > best,
> >
> > Robert Dodier
>
>
>
> _______________________________________________
> Maxima-discuss mailing list
> Max...@li...
> https://lists.sourceforge.net/lists/listinfo/maxima-discuss
>