The "Problem of the Year".

This program solves a math problem that is given to elementary school students in public schools in Ontario. It's also used in the Spirit of Math program. I observe it's also mentioned in http://mathforum.org/yeargames/ so it's not just an Ontario thing

Given four single digits, generate expressions that evaluate to the numbers from 1 to 100, using each of those four digits once only. In some variations, the order of the digits may not change. Each year a different set of digits is used (I suppose most places just use the four digits of the current year.)

It is permitted to concatenate digits, thus if the digits are 1, 2, 3, and 4, expressions such as 12 + 34 as allowed

It is also permitted to rewrite the digit as the number divided by 10, e.g. to rewrite 1 as ".1". They call this "expressing as a decimal" which seems like a strange way to talk about it. It is also permitted to add the overbar for repeating decimals. So far as I know this is only useful for .9, since .999.. is equal to 1,

For Spirit of Math the set of operators allowed is explicitly limited to addition, subtraction, multiplication, division, exponentiation (binary operators) and square root and factorial (monadic). Modular division, logarithm and absolute value operators are not allowed.

To run the program:

ruby poty.rb --digits 1492

(1492 just happens to be the year that was used in my kid's class. I'm aware that the events of the year 1492 caused some significant problems for the people living in the continents now called "America".)

Two implementations here.

The first (tree) generates all possible expression trees using the five binary operators, applying the monadic operators to the digits and to the intermediate values in the tree. That's about 30 million expressions in total, worse case. The second one (graph) grows expression trees from the four given leaf nodes, upward. It's must faster that the tree algorithm. At present, it also generates the exact same tree more than once, that's a little wasteful but not a serious problem.

When using the digits in the current year (2016), The tree searcher finds only 54 solutions. I do not know whether POTY is known to be solvable for all years. I am sure the tree algorithm does not generate all possible trees, but the combinatorics were getting out of hand (which is why I made the graph solver. The graph solver eventually finds 60 solutions, after examining 4 million expresions.