[jacobolus]

None of the professional numerical analysts I know work with a handheld calculator.

The point of learning a slide rule is not that it is a particularly important practical tool, but that understanding how it works has independent pedagogical value.

If someone built their own electronic calculator from discrete components, programmed one on an FPGA, or even implemented a bunch of mathematical functions on an existing computer, that would be similarly educational (though teaching different things than the slide rule), but just knowing how to navigate the interface of an electronic calculator doesn’t teach anything.

[mkl]

Professionals don't, but students need to, especially in study groups, tutorials, tests, and exams, when computers are not available. Calculators are also much more approachable than computers for this material for students with no programming knowledge.

The point of using a calculator is to skip over the tedious unimportant details when learning other things e.g. Newton's method or Euler's method. The calculator itself is a tool, not an educational destination.

Learning a slide rule as you say makes sense in a history of maths course, and implementing one's own calculator makes sense in an electronics or computer science course.

[jacobolus]

Computers are plenty available in “study groups and tutorials” (for one thing almost all college students and many high school students now have smartphones, and most college students also have laptops and/or tablets, but if you want a cheap computer just for math class, get a netbook or cheap android tablet and external keyboard or something made from a raspberry pi or ....)

There’s really not much pedagogical value in using a handheld calculator to apply Newton’s method to some root-finding problem or apply Euler’s method (forward differences) to model a differential equation. Both of these are very simple and students can learn enough of some simple programming language to implement them both in a very short amount of time. That time is much better spent than doing 4 or 5 examples of each with a handheld calculator. If a general-purpose programming language seems too much, get them implementing these simple tools in desmos or geogebra.

On a timed in-class exam in an introductory calculus course, there are much better ways of judging someone’s understanding than making them perform a bunch of tedious and error-prone number crunching. (For example you could give the students rulers and printed graphs of a function – without any symbolic expression written down – and ask them to sketch approximately what a solution using Newton’s method would look like).

If you want a nice introductory calculus book organized along more computer-focused and conceptual lines, take a look at http://www.math.smith.edu/~callahan/intromine.html

In a post-introductory-calculus “numerical analysis” course, the exams should consist of writing proofs, not performing algorithms.

The important thing for numerical analysts about different root-finding methods (etc.) is their convergence speed, numerical stability, computational complexity, and so on. In the 1960s and before it might have made sense to get students performing the role of human computer, but nowadays it is anachronistic.

> Learning a slide rule as you say makes sense in a history of maths course

No, learning how to use a slide rule makes sense in an algebra course for ~15-year-old secondary math students who are learning about logarithms, and for 15–17-year-old secondary science students. They’ll end up with a better intuitive understanding of logarithms and significant digits and error bounds after regularly using a slide rule for even a few weeks than any amount of reading about it or doing formal algebraic manipulation.

Electronic calculators give students a very misleading impression that all of the digits printed on its display are meaningful. But in high school chemistry, physics, etc. courses there is pretty much no experiment ever done with better than about 2 digits of precision.