What fun, useful, mathematical fact do you know that is not normally taught at school?

5 Answers
Apr 25, 2017

How to evaluate "towers of exponents", such as #2^(2^(2^2))#, and how to work out the last digit of #2^n,# #ninNN#.

Explanation:

In order to evaluate these "towers", we start at the top and work our way down.

So:

#2^(2^(2^2))=2^(2^4)=2^16=65,536#

On a similar, but slightly unrelated note, I also know how to work out the last digits of #2# raised to any natural exponent. The last digit of #2# raised to something always cycles between four values: #2,4,8,6#.

#2^1=2,# #2^2=4,# #2^3=8,# #2^4=16#

#2^5=32,# #2^6=64,# #2^7=128,# #2^8=256#

So if you want to find the last digit of #2^n#, find which place it is in the cycle, and you'll know its last digit.

Apr 25, 2017

If #n > 0# and #a# is an approximation to #sqrt(n)#, then:

#sqrt(n) = a+b/(2a+b/(2a+b/(2a+b/(2a+b/(2a+...)))))#

where #b = n-a^2#

Explanation:

Suppose we want to find the square root of some number #n > 0#.

Further we would like the result to be some kind of continued fraction that repeats at each step.

Try:

#sqrt(n) = a+b/(2a+b/(2a+b/(2a+b/(2a+b/(2a+...)))))#

#color(white)(sqrt(n)) = a+b/(a+a+b/(2a+b/(2a+b/(2a+b/(2a+...)))))#

#color(white)(sqrt(n)) = a+b/(a+sqrt(n))#

Subtract #a# from both ends to get:

#sqrt(n)-a=b/(a+sqrt(n))#

Multiply both sides by #sqrt(n)+a# to get:

#b = (sqrt(n)-a)(sqrt(n)+a) = n-a^2#

So if #a^2# is a little less than #n#, then #b# will be small and the continued fraction will converge quicker.

For example, if we have #n=28# and choose #a=5#, then we get:

#b = n-a^2 = 28-5^2 = 28-25=3#

So:

#sqrt(28) = 5+3/(10+3/(10+3/(10+3/(10+3/(10+...)))))#

which gives us approximations:

#sqrt(28) ~~ 5+3/10 = 5.3#

#sqrt(28) ~~ 5+3/(10+3/10) = 545/103 ~~ 5.29126#

#sqrt(28) ~~ 5+3/(10+3/(10+3/10)) = 5609/1060 ~~ 5.2915094#

A calculator tells me #sqrt(28) ~~ 5.291502622#

So this is not converging particularly quickly.

Alternatively, we might put #n=28# and #a=127/24# to find:

#b = n-a^2 = 28-127^2/24^2 = 28-16129/576 = (16128-16129)/576 = -1/576#

So:

#sqrt(28) = 127/24-(1/576)/(127/12-(1/576)/(127/12-(1/576)/(127/12-...)))#

giving us approximations:

#sqrt(28) ~~ 127/24 = 5.291bar(6)#

#sqrt(28) ~~ 127/24-(1/576)/(127/12) = 32257/6096 ~~ 5.29150262467#

That's converging a lot faster.

Apr 29, 2017

You can find approximations to square roots using a recursively defined sequence.

Explanation:

#color(white)()#
The method

Given a positive integer #n# which is not a perfect square:

  • Let #p = floor(sqrt(n))# be the largest positive integer whose square does not exceed #n#.

  • Let #q = n-p^2#

  • Define a sequence of integers by:

    #{(a_1 = 1), (a_2 = 2p), (a_(i+2) = 2pa_(i+1)+qa_i" for " i >= 1) :}#

Then the ratio between successive terms of the sequence will tend towards #p+sqrt(n)#

#color(white)()#
Example

Let #n=7#.

Then #p = floor(sqrt(7)) = 2#, since #2^2=4 < 7# but #3^2 = 9 > 7#.

Then #q=n-p^2 = 7-2^2 = 3#

So our sequence starts:

#1, 4, 19, 88, 409, 1900, 8827, 41008,...#

In theory the ratio between consecutive terms should tend towards #2+sqrt(7)#

Let's see:

#4/1 = 4#

#19/4 = 4.75#

#88/19 ~~ 4.63#

#409/88 ~~ 4.6477#

#1900/409 ~~ 4.6455#

#8827/1900 ~~ 4.645789#

#41008/8827 ~~ 4.645746#

Note that #2+sqrt(7) ~~ 4.645751311#

#color(white)()#
How it works

Suppose we have a sequence defined by given values of #a_1, a_2# and a rule:

#a_(n+2) = 2p a_(n+1) + q a_n#

for some constants #p# and #q#.

Consider the equation:

#x^2-2px-q = 0#

The roots of this equation are:

#x_1 = p+sqrt(p^2+q)#

#x_2 = p-sqrt(p^2+q)#

Then any sequence with general term #Ax_1^n+Bx_2^n# will satisfy the recurrence rule we specified.

Next solve:

#{ (Ax_1+Bx_2 = a_1), (Ax_1^2+Bx_2^2 = a_2) :}#

for #A# and #B#.

We find:

#a_1x_2-a_2 = Ax_1(x_2-x_1)#

#a_1x_1-a_2 = Bx_2(x_1-x_2)#

and hence:

#A=(a_1x_2-a_2)/(x_1(x_2-x_1))#

#B=(a_1x_1-a_2)/(x_2(x_1-x_2))#

So with these values of #x_1, x_2, A, B# we have:

#a_n = Ax_1^n+Bx_2^n#

If #q < 3p^2# then #abs(x_2) < 1# and the ratio between successive terms will tend towards #x_1 = p+sqrt(p^2+q)#

Apr 29, 2017

Modular division

Explanation:

Modular division is just the same as division except the answer is the remainder instead of the actual value. Rather than the #-:# symbol, you use the #%# symbol.

For example, usually, if you were to solve #16-:5# you would get #3# remainder #1# or #3.2#. However, using modular division, #16%5=1#.

Apr 29, 2017

Evaluating squares with summations

Explanation:

Normally, you should know squares such as #5^2=25#. However, when numbers get bigger such as #25^2#, it gets harder to know off the top of your head.

I realized that after a while, squares are just sums of odd numbers.

What I mean is this:

#sum_(n=0)^k 2n+1# where #k# is the base value minus #1#

So #5^2# could be written as:

#sum_(n=0)^4 2n+1#

That will give you:

#1+3+5+7+9#

This, in fact, is #25#.

Since the numbers are always incrementing by #2#, I could then add the first and last number and then multiply by #k/2#.

So for #25^2#

#sum_(n=0)^24 2n+1=1+3+...+49#

So I can just do #(49+1)(25/2)# and get #25^2# which is #625#.

It's not really practical but it's interesting to know.

#color(white)()#
Bonus

Knowing that:

#n^2 = overbrace(1+3+5+...+(2n-1))^"n terms" = ((1+(2n-1))/2)^2#

allows us to solve some problems about differences of squares.

For example, what are all the solutions in positive integers #m, n# of #m^2-n^2 = 40# ?

This reduces to finding what sums of consecutive odd integers add up to #40#...

#40 = overbrace(19+21)^"average 20"#

#color(white)(40) = (1+3+...+21)-(1+3+...+17)#

#color(white)(40) = ((1+21)/2)^2+((1+17)/2)^2#

#color(white)(40) = 11^2-9^2#

#40 = overbrace(7+9+11+13)^"average 10"#

#color(white)(40) = (1+3+...+13)-(1+3+5)#

#color(white)(40) = ((1+13)/2)^2-((1+5)/2)^2#

#color(white)(40) = 7^2-3^2#