Factorial number system

The factorial number system is a mixed radix numeral system: the i-th digit from the right has base i, which means that the digit must be strictly less than i, and that (taking into account the bases of the less significant digits) its value to be multiplied by (i − 1)! (its place value).

From this it follows that the rightmost digit is always 0, the second can be 0 or 1, the third 0, 1 or 2, and so on (sequence A124252 in the OEIS). The factorial number system is sometimes defined with the 0! place omitted because it is always zero (sequence A007623 in the OEIS).

In this article, a factorial number representation will be flagged by a subscript "!", so for instance 3:4:1:0:1:0_! stands for 3₅4₄1₃0₂1₁0₀, whose value is

= 3×5! + 4×4! + 1×3! + 0×2! + 1×1! + 0×0! 

= ((((3×5 + 4)×4 + 1)×3 + 0)×2 + 1)×1 + 0

=  463₁₀.

(The place value is one less than the radix position, which is why these equations begin with 5!.)

General properties of mixed radix number systems also apply to the factorial number system. For instance, one can convert a number into factorial representation producing digits from right to left, by repeatedly dividing the number by the radix (1, 2, 3, ...), taking the remainder as digits, and continuing with the integer quotient, until this quotient becomes 0.

For example, 463₁₀ can be transformed into a factorial representation by these successive divisions:

The process terminates when the quotient reaches zero. Reading the remainders backward gives 3:4:1:0:1:0_!.

In principle, this system may be extended to represent fractional numbers, though rather than the natural extension of place values (−1)!, (−2)!, etc., which are undefined, the symmetric choice of radix values n = 0, 1, 2, 3, 4, etc. after the point may be used instead. Again, the 0 and 1 places may be omitted as these are always zero. The corresponding place values are therefore 1/1, 1/1, 1/2, 1/6, 1/24, ..., 1/n!, etc.

The following sortable table shows the 24 permutations of four elements with different inversion related vectors. The left and right inversion counts ${\displaystyle l}$ and ${\displaystyle r}$ (the latter often called Lehmer code) are particularly eligible to be interpreted as factorial numbers. ${\displaystyle l}$ gives the permutation's position in reverse colexicographic order (the default order of this table), and the latter the position in lexicographic order (both counted from 0).

Sorting by a column that has the omissible 0 on the right makes the factorial numbers in that column correspond to the index numbers in the immovable column on the left. The small columns are reflections of the columns next to them, and can be used to bring those in colexicographic order. The rightmost column shows the digit sums of the factorial numbers (OEIS: A034968 in the tables default order).

The factorial numbers of a given length form a permutohedron when ordered by the bitwise ${\displaystyle \leq }$ relation

These are the right inversion counts (aka Lehmer codes) of the permutations of four elements.

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

${\displaystyle \pi }$ ${\displaystyle v}$ ${\displaystyle l}$ p-b ${\displaystyle r}$ # 0 1234 4321 0000 0000 0000 0000 0000 0000 0 1 2134 4312 1000 0001 0010 0100 1000 0001 1 2 1324 4231 0100 0010 0100 0010 0100 0010 1 3 3124 4213 1100 0011 0110 0110 2000 0002 2 4 2314 4132 2000 0002 0200 0020 1100 0011 2 5 3214 4123 2100 0012 0210 0120 2100 0012 3 6 1243 3421 0010 0100 1000 0001 0010 0100 1 7 2143 3412 1010 0101 1010 0101 1010 0101 2 8 1423 3241 0110 0110 1100 0011 0200 0020 2 9 4123 3214 1110 0111 1110 0111 3000 0003 3 10 2413 3142 2010 0102 1200 0021 1200 0021 3 11 4213 3124 2110 0112 1210 0121 3100 0013 4 12 1342 2431 0200 0020 2000 0002 0110 0110 2 13 3142 2413 1200 0021 2010 0102 2010 0102 3 14 1432 2341 0210 0120 2100 0012 0210 0120 3 15 4132 2314 1210 0121 2110 0112 3010 0103 4 16 3412 2143 2200 0022 2200 0022 2200 0022 4 17 4312 2134 2210 0122 2210 0122 3200 0023 5 18 2341 1432 3000 0003 3000 0003 1110 0111 3 19 3241 1423 3100 0013 3010 0103 2110 0112 4 20 2431 1342 3010 0103 3100 0013 1210 0121 4 21 4231 1324 3110 0113 3110 0113 3110 0113 5 22 3421 1243 3200 0023 3200 0023 2210 0122 5 23 4321 1234 3210 0123 3210 0123 3210 0123 6

For another example, the greatest number that could be represented with six digits would be 543210_! which equals 719 in decimal:

5×5! + 4×4! + 3x3! + 2×2! + 1×1! + 0×0!.

Clearly the next factorial number representation after 5:4:3:2:1:0_! is 1:0:0:0:0:0:0_! which designates 6! = 720₁₀, the place value for the radix-7 digit. So the former number, and its summed out expression above, is equal to:

6! − 1.

The factorial number system provides a unique representation for each natural number, with the given restriction on the "digits" used. No number can be represented in more than one way because the sum of consecutive factorials multiplied by their index is always the next factorial minus one:

${\displaystyle \sum _{i=0}^{n}{i\cdot i!}={(n+1)!}-1.}$

This can be easily proved with mathematical induction, or simply by noticing that ${\displaystyle \forall i,i\cdot i!=(i+1-1)\cdot i!=(i+1)!-i!}$ : subsequent terms cancel each other, leaving the first and last term (see Telescoping series)

However, when using Arabic numerals to write the digits (and not including the subscripts as in the above examples), their simple concatenation becomes ambiguous for numbers having a "digit" greater than 9. The smallest such example is the number 10 × 10! = 36,288,000₁₀, which may be written A0000000000_!=10:0:0:0:0:0:0:0:0:0:0_!, but not 100000000000_! = 1:0:0:0:0:0:0:0:0:0:0:0_! which denotes 11! = 39,916,800₁₀. Thus using letters A–Z to denote digits 10, 11, 12, ..., 35 as in other base-N make the largest representable number 36 × 36! − 1. For arbitrarily greater numbers one has to choose a base for representing individual digits, say decimal, and provide a separating mark between them (for instance by subscripting each digit by its base, also given in decimal, like 2₄0₃1₂0₁, this number also can be written as 2:0:1:0_!). In fact the factorial number system itself is not truly a numeral system in the sense of providing a representation for all natural numbers using only a finite alphabet of symbols, as it requires an additional separation mark.

There is a natural mapping between the integers 0, ..., n! − 1 (or equivalently the numbers with n digits in factorial representation) and permutations of n elements in lexicographical order, when the integers are expressed in factoradic form. This mapping has been termed the Lehmer code (or inversion table). For example, with n = 3, such a mapping is

In each case, calculating the permutation proceeds by using the leftmost factoradic digit (here, 0, 1, or 2) as the first permutation digit, then removing it from the list of choices (0, 1, and 2). Think of this new list of choices as zero indexed, and use each successive factoradic digit to choose from its remaining elements. If the second factoradic digit is "0" then the first element of the list is selected for the second permutation digit and is then removed from the list. Similarly, if the second factoradic digit is "1", the second is selected and then removed. The final factoradic digit is always "0", and since the list now contains only one element, it is selected as the last permutation digit.

The process may become clearer with a longer example. Let's say we want the 2982nd permutation of the numbers 0 through 6. The number 2982 is 4:0:4:1:0:0:0_! in factoradic, and that number picks out digits (4,0,6,2,1,3,5) in turn, via indexing a dwindling ordered set of digits and picking out each digit from the set at each turn:

                            4:0:4:1:0:0:0!  ─►  (4,0,6,2,1,3,5)
factoradic: 4              :   0            :   4          :   1        :   0      :   0    :   0!
            ├─┬─┬─┬─┐          │                ├─┬─┬─┬─┐      ├─┐          │          │        │
sets:      (0,1,2,3,4,5,6) ─► (0,1,2,3,5,6) ─► (1,2,3,5,6) ─► (1,2,3,5) ─► (1,3,5) ─► (3,5) ─► (5)
                    │          │                        │        │          │          │        │
permutation:       (4,         0,                       6,       2,         1,         3,       5)

A natural index for the group direct product of two permutation groups is the concatenation of two factoradic numbers, with two subscript "!"s.

           concatenated
 decimal   factoradics        permutation pair
    010     0:0:0!0:0:0!           ((0,1,2),(0,1,2))
    110     0:0:0!0:1:0!           ((0,1,2),(0,2,1))
               ...
    510     0:0:0!2:1:0!           ((0,1,2),(2,1,0))
    610     0:1:0!0:0:0!           ((0,2,1),(0,1,2))
    710     0:1:0!0:1:0!           ((0,2,1),(0,2,1))
               ...
   2210     1:1:0!2:0:0!           ((1,2,0),(2,0,1))
               ...
   3410     2:1:0!2:0:0!           ((2,1,0),(2,0,1))
   3510     2:1:0!2:1:0!           ((2,1,0),(2,1,0))

Unlike single radix systems whose place values are baseⁿ for both positive and negative integral n, the factorial number base cannot be extended to negative place values as these would be (−1)!, (−2)! and so on, and these values are undefined. (see factorial)

One possible extension is therefore to use 1/0!, 1/1!, 1/2!, 1/3!, ..., 1/n! etc. instead, possibly omitting the 1/0! and 1/1! places which are always zero.

With this method, all rational numbers have a terminating expansion, whose length in 'digits' is less than or equal to the denominator of the rational number represented. This may be proven by considering that there exists a factorial for any integer and therefore the denominator divides into its own factorial even if it does not divide into any smaller factorial.

By necessity, therefore, the factoradic expansion of the reciprocal of a prime has a length of exactly that prime (less one if the 1/1! place is omitted). Other terms are given as the sequence A046021 on the OEIS. It can also be proven that the last 'digit' or term of the representation of a rational with prime denominator is equal to the difference between the numerator and the prime denominator.

There is also a non-terminating equivalent for every rational number akin to the fact that in decimal 0.24999... = 0.25 = 1/4 and 0.999... = 1, etc., which can be created by reducing the final term by 1 and then filling in the remaining infinite number of terms with the highest value possible for the radix of that position.

In the following selection of examples, spaces are used to separate the place values, otherwise represented in decimal. The rational numbers on the left are also in decimal:

• ${\displaystyle 1/2=0.0\ 1_{!}}$
• ${\displaystyle 1/3=0.0\ 0\ 2_{!}}$
• ${\displaystyle 2/3=0.0\ 1\ 1_{!}}$
• ${\displaystyle 1/4=0.0\ 0\ 1\ 2_{!}}$
• ${\displaystyle 3/4=0.0\ 1\ 1\ 2_{!}}$
• ${\displaystyle 1/5=0.0\ 0\ 1\ 0\ 4_{!}}$
• ${\displaystyle 1/6=0.0\ 0\ 1_{!}}$
• ${\displaystyle 5/6=0.0\ 1\ 2_{!}}$
• ${\displaystyle 1/7=0.0\ 0\ 0\ 3\ 2\ 0\ 6_{!}}$
• ${\displaystyle 1/8=0.0\ 0\ 0\ 3_{!}}$
• ${\displaystyle 1/9=0.0\ 0\ 0\ 2\ 3\ 2_{!}}$
• ${\displaystyle 1/10=0.0\ 0\ 0\ 2\ 2_{!}}$
• ${\displaystyle 1/11\ \ =0.0\ 0\ 0\ 2\ 0\ 5\ 3\ 1\ 4\ 0\ A_{!}}$
• ${\displaystyle 2/11\ \ =0.0\ 0\ 1\ 0\ 1\ 4\ 6\ 2\ 8\ 1\ 9_{!}}$
• ${\displaystyle 9/11\ \ =0.0\ 1\ 1\ 3\ 3\ 1\ 0\ 5\ 0\ 8\ 2_{!}}$
• ${\displaystyle 10/11=0.0\ 1\ 2\ 1\ 4\ 0\ 3\ 6\ 4\ 9\ 1_{!}}$
• ${\displaystyle 1/12\ \ =0.0\ 0\ 0\ 2_{!}}$
• ${\displaystyle 5/12\ \ =0.0\ 0\ 2\ 2_{!}}$
• ${\displaystyle 7/12\ \ =0.0\ 1\ 0\ 2_{!}}$
• ${\displaystyle 11/12=0.0\ 1\ 2\ 2_{!}}$
• ${\displaystyle 1/15=0.0\ 0\ 0\ 1\ 3_{!}}$
• ${\displaystyle 1/16=0.0\ 0\ 0\ 1\ 2\ 3_{!}}$
• ${\displaystyle 1/18=0.0\ 0\ 0\ 1\ 1\ 4_{!}}$
• ${\displaystyle 1/20=0.0\ 0\ 0\ 1\ 1_{!}}$
• ${\displaystyle 1/24=0.0\ 0\ 0\ 1_{!}}$
• ${\displaystyle 1/30=0.0\ 0\ 0\ 0\ 4_{!}}$
• ${\displaystyle 1/36=0.0\ 0\ 0\ 0\ 3\ 2_{!}}$
• ${\displaystyle 1/60=0.0\ 0\ 0\ 0\ 2_{!}}$
• ${\displaystyle 1/72=0.0\ 0\ 0\ 0\ 1\ 4_{!}}$
• ${\displaystyle 1/120=0.0\ 0\ 0\ 0\ 1_{!}}$
• ${\displaystyle 1/144=0.0\ 0\ 0\ 0\ 0\ 5_{!}}$
• ${\displaystyle 1/240=0.0\ 0\ 0\ 0\ 0\ 3_{!}}$
• ${\displaystyle 1/360=0.0\ 0\ 0\ 0\ 0\ 2_{!}}$
• ${\displaystyle 1/720=0.0\ 0\ 0\ 0\ 0\ 1_{!}}$

There are also a small number of constants that have patterned representations with this method:

• ${\displaystyle e=1\ 0.0\ 1\ 1\ 1\ 1\ 1\ 1\ 1\ 1\ 1\ 1\ 1\ 1\ 1\ 1\ 1..._{!}}$
• ${\displaystyle e^{-1}=0.0\ 0\ 2\ 0\ 4\ 0\ 6\ 0\ 8\ 0\ A\ 0\ C\ 0\ E..._{!}}$
• ${\displaystyle \sin(1)=0.0\ 1\ 2\ 0\ 0\ 5\ 6\ 0\ 0\ 9\ A\ 0\ 0\ D\ E..._{!}}$
• ${\displaystyle \cos(1)=0.0\ 1\ 0\ 0\ 4\ 5\ 0\ 0\ 8\ 9\ 0\ 0\ C\ D\ 0..._{!}}$
• ${\displaystyle \sinh(1)=1.0\ 0\ 1\ 0\ 1\ 0\ 1\ 0\ 1\ 0\ 1\ 0\ 1\ 0\ 1\ 0..._{!}}$
• ${\displaystyle \cosh(1)=1.0\ 1\ 0\ 1\ 0\ 1\ 0\ 1\ 0\ 1\ 0\ 1\ 0\ 1\ 0\ 1..._{!}}$

• Primorial number system
• Combinatorial number system (also called combinadics)
• Steinhaus–Johnson–Trotter algorithm, an algorithm that generates Gray codes for the factorial number system

• Mantaci, Roberto; Rakotondrajao, Fanja (2001), "A permutation representation that knows what "Eulerian" means" (PDF), Discrete Mathematics and Theoretical Computer Science, 4: 101–108, archived from the original (PDF) on 2011-05-24, retrieved 2005-03-27.
• Arndt, Jörg (2010). Matters Computational: Ideas, Algorithms, Source Code. pp. 232–238.

• A Lehmer code calculator Note that their permutation digits start from 1, so mentally reduces o all permutation digits by one to get results equivalent to those on this page.
• Factorial number system