I hate most calculators. I inevitably hit the wrong button, or
realize that I made a mistake somewhere while entering the
equation, and I have to start all over again. Sure, there are
those fancy calculators where you type in the equation and
you can go back and change it if needed - but why
settle for
that when your calculator could do everything itself?

To use my calculator, one would press the "Scan" button, wait
for the
display to read "Ready", drag it across the equation,
and await an answer. After 60 seconds of idle time, it
would just shut itself off again. No messing with small, poorly
described buttons - everything it needs to know is taken from
what you wrote.

The design of the unit would permit the calculator to be very
small and lightweight. It would require only a few buttons
("Scan", scroll left, and scroll right) and it would be entirely
powered by solar cells. As well, the
display would have multiple lines, for cases where it could not
solve an equation, but merely simplify it.

Good handwriting recognition software already exists, as
seen in the NetwonOS, Mac OS X's Ink, the tablet edition of
Windows XP, and several non-Graffiti PDAs. How much
more difficult would it be to add recognition for fractions,
exponents, and radicals?

The unit would be designed to handle advanced algebra - it
would know that sqrt(-1) is sqrt(1)*i, for instance. Variables and
irrational numbers such as Pi and c would be built in, and
functions like tan() would be recognized as well. Whenever
possible, it would solve for variables, and print them out on the
display.

almost what yor looking forhttp://www.cs.swan....RealCalculator2.mov only diff is that you write the equations instead of scanning them [mawgadog, Aug 08 2005]

My palm pilot has a very nice calculator built in. This makes it a calculator with handwriting recognition. Not the same as your idea, but a solution to your problem.

It does, however, know that sqrt(-1) is sqrt(1)*i, has Pi, c, and every useful conversion you can think of, full trig functions, and a whole bunch more.

What phundug said. I'd have to proof its scan to make sure that it made no mistakes. Since I can type (even slowly and carefully) into a calculator faster than I can proof two equations/strings of numbers against each other, I'd be better off with a calculator with buttons.

I'd like [Worldgineer]'s solution, if I knew it didn't require any intermediate steps. That is, write out the formula and touch <solve> to get the answer.

I guess if the calculator displays on screen what it thinks it's recognizing, then you can press CONFIRM and then COMPUTE and accept the answer as valid.