A mainstay of local elementary schools are word puzzles that challenge and develop the kids' verbal skills. A favorite is the 'how many other words can you make from the letters of this big word' variety.
So, for example, the word "bread" will yield: red, are, ear, dare, dear, read, bar, etc. There are more (at least 17 common words, actually, but we'll get to that). We have fairly verbal kids, who now and then come up with more words than the "official" target. No wonder ...
How to Solve It
How would you get a list of all of the right answers? First, you have to figure out how to ennumerate the complete list of letter subsets found in the "parent"word, and then you have to figure out what words (if any) can be formed by rearranging the letters in any one subset. To generate an initial list of all possible letter subsets from the "parent" word, we can employ a clever hack involving binary numbers. (And as the MIT t-shirt tells us, "There are 10 kinds of people in this world: those who can handle binary arithmetic and those who can't.")
Any word you are going to find involves using some subset of the letters in the "parent" word. Said another way, in any given subset, a letter from the parent word is used (1) or it isn't (0). That means that any given subset of letters can be represented as a string of 1s and 0s according to whether the letter in the parent word was used or not. Consider the word "east":
Map the letters you use to a "1" and any that you don't use to a "0", and the letter combo you use suddenly has a unique numeric "signature".
For a word N characters long, there must be 2N - 1 possible combinations of letters (you subtract one since one of the combos is all 0s ... no letters used ... so it can't ever help).
For a four letter word like "east", that's 24 - 1 = 16 - 1 = 15 combos, and the binary form of each number from 1 to 15 can be used as a map for which letters are included in the subset.
Many "parent" words will have repeating letters, so some of these binary letter subsets will boil down to the same set of letters (just originating from different places in the "parent" word), so we steal a trick from the Jumble Solver, and sort the letters in each subset, and only keep the list of unique combos.
Finally, we recognize each subset may represent more than one word, depending on how the letters are arranged (for example, the subset of "e", "a", and "t" can generate words "eat", "ate", and "tea"). But this is just the Jumble problem we've solved elsewhere, so we use that code to get the set of matches from the dictionary.
The dictionary I use contains over 100,000 words, but also includes a flag for the 5,000 most common English words, and we restrict our search to this subset, which is still pretty expansive for the K-12 crowd. (Although, maybe not ... in our "bread" example, this restriction misses "bead" and (gasp) "bard" ... so maybe it is a little too restrictive!)
One challenge to come home from the 2nd Grade was to find the 39 words hidden in "ornaments". Using this script, we found 105! Time to start talking about extra credit ...