Block’n’Role

How low can you go? When it comes to standard bases in mathematics, you can’t go lower than 2. But base 2, or binary, is unsurpassable for simplicity and beauty. With only two digits, 1 and 0, you can capture any integer you like:

• 0, 1, 2, 3, 4, 5... -> 0, 1, 10, 11, 100, 101, 110, 111, 1000, 1001, 1010, 1011, 1100, 1101, 1110, 1111, 10000, 10001, 10010, 10011, 10100, 10101, 10110, 10111, 11000, 11001, 11010, 11011, 11100, 11101, 11110, 11111, 100000, 100001, 100010, 100011, 100100, 100101, 100110, 100111, 101000, 101001, 101010, 101011, 101100, 101101, 101110, 101111, 110000, 110001, 110010, 110011, 110100, 110101, 110110, 110111, 111000, 111001, 111010, 111011, 111100, 111101, 111110, 111111...


Here are a few famous decimal numbers in binary:

• 23 = 10111 in binary
• 666 = 1010011010 in binary
• 1492 = 10111010100 in binary
• 2001 = 11111010001 in binary

As you can see, there’s a problem with binary for human beings. It takes up a lot of space and doesn’t look very distinctive. But that’s easy to solve by converting binary into octal (base 8) or hexadecimal (base 16). One digit in octal is worth three digits in binary and one digit in hexadecimal is worth four digits in binary. So the conversion back and forth is very easy:

• 23 = 10111 → (010,111) → 27 in octal
• 23 = 10111 → (0001,0111) → 17 in hexadecimal
• 666 = 1010011010 → (001,010,011,010) → 1232 in octal
• 666 = 1010011010 → (0010,1001,1010) → 29A in hexademical
• 1492 = 10111010100 → (010,111,010,100) → 2724 in octal
• 1492 = 10111010100 → (0101,1101,0100) → 5D4 in hexademical
• 2001 = 11111010001 → (011,111,010,001) → 3721 in octal
• 2001 = 11111010001 → (0111,1101,0001) → 7D1 in hexademical

But there’s another way to compress a binary number: count the lengths of the runs of 1 and 0. For example, 23 = 10111 and 10111 → one 1, one 0, three 1s → (1,1,3) → 113. That’s not much of a compression, but it usually gets better as the numbers get bigger:

• 2001 = 11111010001 → (5,1,1,3,1) → 51131

From the compressed form you can easily re-create the binary number:

• 51131 → (5,1,1,3,1) → (11111,0,1,000,1) → 11111010001

This block-compression doesn’t work with any other standard base. For example, the compressed form (1,2) in ternary, or base 3, is ambiguous:

• (1,2) → (1,00) → 100 in base 3 = 09 in decimal
• (1,2) → (1,22) → 122 in base 3 = 17 in decimal
• (1,2) → (2,00) → 200 in base 3 = 18 in decimal
• (1,2) → (2,11) → 211 in base 3 = 22 in decimal

The higher the base, the bigger the ambiguity. But ambiguity exists with binary block-compressions too. Look at 51131 ← 11111010001 = 2001 in decimal. Out of context, 51131 is infinitely ambiguous. It could represent a number in any base higher than 5:

• 51131 in base 06 = 006751 in base 10
• 51131 in base 07 = 012419 in base 10
• 51131 in base 08 = 021081 in base 10
• 51131 in base 09 = 033643 in base 10
• 51131 in base 10 = 051131 in base 10
• 51131 in base 11 = 074691 in base 10
• 51131 in base 12 = 105589 in base 10
• 51131 in base 13 = 145211 in base 10
• 51131 in base 14 = 195063 in base 10
• 51131 in base 15 = 256771 in base 10
• 51131 in base 16 = 332081 in base 10
• 51131 in base 17 = 422859 in base 10
• 51131 in base 18 = 531091 in base 10
• 51131 in base 19 = 658883 in base 10
• 51131 in base 20 = 808461 in base 10...

But that ambiguity raises an interesting question. Does the binary block-compression of n ever match the digits of n in another base? Yes, it does:

• 23 = 10111 in base 2 → (1,1,3) and 113 in base 4 = 10111 in base 2 = 23 in base 10

113 in base 4 = 1*4^2 + 1*4 + 3*4^0 = 16+4+3 = 23. You could call this “Block’n’Role”, because the blocks of 1 and 0 allow a binary number to retain its identity but take on a different role, that is, represent a number in a different base. Here’s a list of binary block-numbers that match the digits of n in another base:

• 10111 → (1,1,3) = 113 in base 4 (n=23)
• 11001 → (2,2,1) = 221 in base 3 (n=25)
• 101100 → (1,1,2,2) = 1122 in base 3 (n=44)
• 111001 → (3,2,1) = 321 in base 4 (n=57)
• 1011111 → (1,1,5) = 115 in base 9 (n=95)
• 1100001 → (2,4,1) = 241 in base 6 (n=97)
• 11100001 → (3,4,1) = 341 in base 8 (n=225)
• 100110000 → (1,2,2,4) = 1224 in base 6 (n=304)
• 101110111 → (1,1,3,1,3) = 11313 in base 4 (n=375)
• 111111001 → (6,2,1) = 621 in base 9 (n=505)
• 1110010111 → (3,2,1,1,3) = 32113 in base 4 (n=919)
• 10000011111 → (1,5,5) = 155 in base 30 (n=1055)
• 11111100001 → (6,4,1) = 641 in base 18 (n=2017)
• 1011101110111 → (1,1,3,1,3,1,3) = 1131313 in base 4 (n=6007)
• 11100101110111 → (3,2,1,1,3,1,3) = 3211313 in base 4 (n=14711)
• 10111011101110111 → (1,1,3,1,3,1,3,1,3) = 113131313 in base 4 (n=96119)
• 111001011101110111 → (3,2,1,1,3,1,3,1,3) = 321131313 in base 4 (n=235383)
• 100000111111111000001 → (1,5,9,5,1) = 15951 in base 31 (n=1081281)
• 101110111011101110111 → 11313131313 in b4 = 1537911
• 1110010111011101110111 → 32113131313 in b4 = 3766135
• 1011101110111011101110111 → 1131313131313 in b4 = 24606583
• 11100101110111011101110111 → 3211313131313 in b4 = 60258167
• 10111011101110111011101110111 → 113131313131313 in b4 = 393705335
• 111001011101110111011101110111 → 321131313131313 in b4 = 964130679

The list of block-nums is incomplete, because I’ve skipped some trivial examples where, for all powers 2^p > 2^2, the block-num is “1P” in base b = (2^p – p). For example:

• 2^3 = 08 = 1000 in base 2 → (1,3) and 13 in base 5 = 8, where 5 = 2^3-3 = 8-3
• 2^4 = 16 = 10000 in base 2 → (1,4) and 14 in base 12 = 16, where 12 = 2^4-4 = 16-4
• 2^5 = 32 = 100000 in base 2 → (1,5) and 15 in base 27 = 32, where 27 = 2^5-5 = 32-5
• 2^6 = 64 = 1000000 in base 2 → (1,6) and 16 in base 58 = 64, where 58 = 2^6-6 = 64-6

And note that the block-num matches in base 4 continue for ever, because the pairs 113… and 321… generate their successors using simple formulae in base 4:

• 113... * 100 + 13
• 321... * 100 + 13

For example, 113 and 321 are the first pair of matches:

• 10111 → (1,1,3) = 113 in base 4 (n=23)
• 111001 → (3,2,1) = 321 in base 4 (n=57)

In base 4, 113 * 100 + 13 = 11313 and 321 * 100 + 13 = 32113:

• 101110111 → (1,1,3,1,3) = 11313 in base 4 (n=375)
• 1110010111 → (3,2,1,1,3) = 32113 in base 4 (n=919)

Next, 11313 * 100 + 13 = 1131313 and 32113 * 100 + 13 = 3211313:

• 1011101110111 → (1,1,3,1,3,1,3) = 1131313 in base 4 (n=6007)
• 11100101110111 → (3,2,1,1,3,1,3) = 3211313 in base 4 (n=14711)

And so on.

Block n Rule

One of my favourite integer sequences uses the formula n(i) = n(i-1) + digsum(n(i-1)), where digsum(n) sums the digits of n. In base 10, it goes like this:

1, 2, 4, 8, 16, 23, 28, 38, 49, 62, 70, 77, 91, 101, 103, 107, 115, 122, 127, 137, 148, 161, 169, 185, 199, 218, 229, 242, 250, 257, 271, 281, 292, 305, 313, 320, 325, 335, 346, 359, 376, 392, 406, 416, 427, 440, 448, 464, 478, 497, 517, 530, 538, 554, 568, 587, 607, 620, 628, 644, 658, 677, 697, 719, 736, 752, 766, 785, 805, 818, 835, 851, 865, 884, 904, 917, 934, 950, 964, 983, 1003…

Another interesting sequence uses the formula n(i) = n(i-1) + digprod(n(i-1)), where digprod(n) multiplies the digits of n (excluding 0). In base 10, it goes like this:

1, 2, 4, 8, 16, 22, 26, 38, 62, 74, 102, 104, 108, 116, 122, 126, 138, 162, 174, 202, 206, 218, 234, 258, 338, 410, 414, 430, 442, 474, 586, 826, 922, 958, 1318, 1342, 1366, 1474, 1586, 1826, 1922, 1958, 2318, 2366, 2582, 2742, 2854, 3174, 3258, 3498, 4362, 4506, 4626, 4914, 5058, 5258, 5658, 6858, 8778, 11914, 11950, 11995…

You can apply these formulae in other bases and it’s trivially obvious that the sequences rise most slowly in base 2, because you’re never summing or multiplying anything but the digit 1. However, there is a sequence for which base 2 is by far the best performer. It has the formula n(i) = n(i-1) + blockmult(n(i-1)), where blockmult(n) counts the lengths of distinct blocks of the same digit, including 0, then multiplies those lengths together. For example:

blockmult(3,b=2) = blockmult(11) = 2
blockmult(28,b=2) = blockmult(11100) = 3 * 2 = 6
blockmult(51,b=2) = blockmult(110011) = 2 * 2 * 2 = 8
blockmult(140,b=2) = blockmult(10001100) = 1 * 3 * 2 * 2 = 12
blockmult(202867,b=2) = blockmult(110001100001110011) = 2 * 3 * 2 * 4 * 3 * 2 * 2 = 576

The full sequence begins like this (numbers are represented in base 10, but the formula is being applied to their representations in binary):

1, 2, 3, 5, 6, 8, 11, 13, 15, 19, 23, 26, 28, 34, 37, 39, 45, 47, 51, 59, 65, 70, 76, 84, 86, 88, 94, 98, 104, 110, 116, 122, 126, 132, 140, 152, 164, 168, 171, 173, 175, 179, 187, 193, 203, 211, 219, 227, 245, 249, 259, 271, 287, 302, 308, 316, 332, 340, 342, 344, 350, 354, 360, 366, 372, 378, 382, 388, 404, 412, 436, 444, 460, 484, 500, 510, 518, 530, 538, 546, 555, 561, 579, 595, 603, 611, 635, 651, 657, 663, 669, 675, 681…

In higher bases, it rises much more slowly. This is base 3:

1, 2, 3, 4, 6, 7, 8, 10, 11, 12, 14, 16, 17, 19, 20, 21, 22, 24, 26, 29, 31, 33, 34, 35, 37, 39, 42, 44, 48, 49, 51, 53, 56, 58, 60, 61, 62, 64, 65, 66, 68, 70, 71, 73, 75, 77, 79, 82, 85, 89, 93, 95, 97, 98, 100, 101, 102, 103, 105, 107, 110, 114, 116, 120, 124, 127, 129, 131, 133, 137, 139, 141, 142, 143, 145, 146, 147, 149, 151, 152, 154, 156, 158, 160, 163…

And this is base 10:

1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 90…

Note how, in bases 3 and 10, blockmult(n) often equals 1. In base 3, the sequence contains [141, 142, 143, 145]:

blockmult(141,b=3) = blockmult(12020) = 1 * 1 * 1 * 1 = 1
blockmult(142,b=3) = blockmult(12021) = 1 * 1 * 1 * 1 = 1
blockmult(143,b=3) = blockmult(12022) = 1 * 1 * 1 * 2 = 2

The formula also returns 1 much further along the sequence in base 3. For example, the 573809th number in the sequence, or n(573809), is 5775037 and blockmult(5775037) = blockmult(101212101212021) = 1^15 = 1. But in base 2, blockmult(n) = 1 is very rare. It happens three times at the beginning of the sequence:

1, 2, 3, 5, 6, 8, 11…

After that, I haven’t found any more examples of blockmult(n) = 1, although blockmult(n) = 2 occurs regularly. For example,

blockmult(n(100723)) = blockmult(44739241) = blockmult(10101010101010101010101001) = 2
blockmult(n(100724)) = blockmult(44739243) = blockmult(10101010101010101010101011) = 2
blockmult(n(100725)) = blockmult(44739245) = blockmult(10101010101010101010101101) = 2

Does the sequence in base 2 return another example of blockmult(n) = 1? The odds seem against it. For any given number of digits in base 2, there is only one number for which blockmult(n) = 1. For example: 1, 10, 101, 1010, 10101, 101010, 1010101… As the sequence increases, the percentage of these numbers becomes smaller and smaller. But the sequence is infinite, so who knows what happens in the end? Perhaps blockmult(n) = 1 occurs infinitely often.