# Most useful heuristic?

As opposed to the most harmful heuristics, what are the most useful heuristics which

• are hand-waving,

• are conducive to proper mathematical education, and

• you have seen taught or taught yourself?

In this context:

• Hand-waving means imprecise, intuitive, ambiguous, with a purpose of impressing or convincing.

• Proper mathematical education means that a person can understand, use, discuss, and derive the learnt mathematical claims after finishing the education process to the levels (a) advertised by goals of the education process and at the same time (b) having, up to some allowed degree of ambiguity, the same, widely accepted meaning in the community. Example: "Real Calculus" could mean "basics of differentiation and integration over the functions $\mathbb{R}\to\mathbb{R}$".

• Seen taught means you closely observed or participated as a learner in the educational process.

• Taught yourself means you were a lecturer or an author of used educational material.

• Woops deleted my comment, I read that you were asking for harmful heuristics. But I guess Euler's approach to the Basel problem was a useful heuristic, though that one might also be considered harmful as the solution to his approach requires the Weierstrass factorization theorem, which is quite deep. Oct 21, 2016 at 7:22
• Interesting combination of up- and downvotes (five of each at the moment) for this opinion based (hence closeable) question several people thought worth answering. Oct 21, 2016 at 13:09
• @EthanBolker "opinion-based (hence closeable)" Almost every soft question is opinion based. Some are "primarily" opinion based ("what's your favourite number?"), others can be reasonably answered with evidence and are relevant, interesting, applicable queries. These judgements are made case-by-case. Oct 21, 2016 at 13:26
• @EthanBolker Besides, if an identical (modulo one adjective) question is good enough for MathOverflow, a site for professionals, why isn't this question good enough for math.se? Oct 21, 2016 at 13:28
• Leon, you're question is clear, I just miread it. Oct 22, 2016 at 7:17

I like Richard Feynman's heuristic of understanding a generality via a simple (but nuanced enough), well-understood example. The start of an inductive proof is sort of an application of this: Convince myself that the statement is true for some simple cases, and see where there might be a general pattern for every such case.

Teaching and using differentials in elementary calculus. They help with linear approximations, the product rule, the chain rule, arclength, Cavalieri's principle, applications of integration. In each case the handwaving can be made rigorous, but the effort at rigor obscures the underlying idea.

Picking up on Tao's comment there, one of the most useful heuristics is thinking of exponentiation as iterating an infinite number of infinitesimal multiplications. This is a useful heuristic not merely in Lie groups but any time one is dealing with the infinitesimal generator of a flow. In fact the flow can be thought of as the shadow of a walk by infinitesimal steps (of course infinitely many of them).

At a more elementary level, thinking of $\frac{dy}{dx}$ as a ratio and ignoring the boos from the audience :-)

I like testing probability deductions with real life experiments. Especially dice problems are really illustrative for people who are just starting out.

A picture is worth a thousand words.

For young students, it is helpful to introduce the concept of multiplication by arranging objects into groups having an equal number of objects in each group, or by arranging the objects into a rectangular array of the desired number of rows and columns.

For algebra students, we can illustrate that $(a+b)^2 = a^2 + 2ab + b^2$, by drawing a square that is $a+b$ on each side, and dividing it with a horizontal and vertical line into four regions: squares that are $a^2$ and $b^2$ on each side and two rectangles that are $a$ by $b$ in size.

For calculus students, when introducing the concepts of derivatives or integrals, it is helpful to illustrate the problem we are trying to solve using a graph, then approximate the solution using finite methods and consider how we might converge to the desired solution using limits.

Thus, we might plot a line tangent to a curve at a specific point and ask "How can we determine the slope of the line?" Then, introduce a finite approximation, such as the secant method, and observe that we obtain a better approximation as the two points are moved closer together.

A similar strategy can be used when introducing definite integrals, by asking how to determine the area under a continuous curve over a closed interval. Introduce the midpoint method as a way to approximate the area and consider how the approximation is improved as we reduce the width of the rectangles.

When introducing Fourier Series, plot examples such as $sin(x)+sin(3x)/3$, then $sin(x)+sin(3x)/3+sin(5x)/5$, etc. to show how the sum more nearly approaches a square wave as the number of terms increases. This also provides an opportunity to discuss topics such as overshoot and ringing, or how a low-pass filter might affect such signals.