Lah number

Below is a table of values for the Lah numbers:

The row sums are ${\textstyle 1,1,3,13,73,501,4051,37633,\dots }$ (sequence A000262 in the OEIS).

Let ${\textstyle x^{(n)}}$ represent the rising factorial ${\textstyle x(x+1)(x+2)\cdots (x+n-1)}$ and let ${\textstyle (x)_{n}}$ represent the falling factorial ${\textstyle x(x-1)(x-2)\cdots (x-n+1)}$. The Lah numbers are the coefficients that express each of these families of polynomials in terms of the other. Explicitly,$${\displaystyle x^{(n)}=\sum _{k=0}^{n}L(n,k)(x)_{k}}$$and$${\displaystyle (x)_{n}=\sum _{k=0}^{n}(-1)^{n-k}L(n,k)x^{(k)}.}$$For example,$${\displaystyle x(x+1)(x+2)={\color {red}6}x+{\color {red}6}x(x-1)+{\color {red}1}x(x-1)(x-2)}$$and$${\displaystyle x(x-1)(x-2)={\color {red}6}x-{\color {red}6}x(x+1)+{\color {red}1}x(x+1)(x+2),}$$

where the coefficients 6, 6, and 1 are exactly the Lah numbers ${\displaystyle L(3,1)}$, ${\displaystyle L(3,2)}$, and ${\displaystyle L(3,3)}$.

The Lah numbers satisfy a variety of identities and relations.

In Karamata–Knuth notation for Stirling numbers$${\displaystyle L(n,k)=\sum _{j=k}^{n}\left[{n \atop j}\right]\left\{{j \atop k}\right\}}$$where ${\textstyle \left[{n \atop j}\right]}$ are the unsigned Stirling numbers of the first kind and ${\textstyle \left\{{j \atop k}\right\}}$ are the Stirling numbers of the second kind.

${\displaystyle L(n,k)={n-1 \choose k-1}{\frac {n!}{k!}}={n \choose k}{\frac {(n-1)!}{(k-1)!}}={n \choose k}{n-1 \choose k-1}(n-k)!}$

${\displaystyle L(n,k)={\frac {n!(n-1)!}{k!(k-1)!}}\cdot {\frac {1}{(n-k)!}}=\left({\frac {n!}{k!}}\right)^{2}{\frac {k}{n(n-k)!}}}$

${\displaystyle k(k+1)L(n,k+1)=(n-k)L(n,k)}$, for ${\displaystyle k>0}$.

The Lah numbers satisfy the recurrence relations$${\displaystyle {\begin{aligned}L(n+1,k)&=(n+k)L(n,k)+L(n,k-1)\\&=k(k+1)L(n,k+1)+2kL(n,k)+L(n,k-1)\end{aligned}}}$$where ${\displaystyle L(n,0)=\delta _{n}}$, the Kronecker delta, and ${\displaystyle L(n,k)=0}$ for all ${\displaystyle k>n}$.

${\displaystyle \sum _{n\geq k}L(n,k){\frac {x^{n}}{n!}}={\frac {1}{k!}}\left({\frac {x}{1-x}}\right)^{k}}$

The n-th derivative of the function ${\displaystyle e^{\frac {1}{x}}}$ can be expressed with the Lah numbers, as follows^[7]$${\displaystyle {\frac {{\textrm {d}}^{n}}{{\textrm {d}}x^{n}}}e^{\frac {1}{x}}=(-1)^{n}\sum _{k=1}^{n}{\frac {L(n,k)}{x^{n+k}}}\cdot e^{\frac {1}{x}}.}$$For example,

${\displaystyle {\frac {\textrm {d}}{{\textrm {d}}x}}e^{\frac {1}{x}}=-{\frac {1}{x^{2}}}\cdot e^{\frac {1}{x}}}$

${\displaystyle {\frac {{\textrm {d}}^{2}}{{\textrm {d}}x^{2}}}e^{\frac {1}{x}}={\frac {\textrm {d}}{{\textrm {d}}x}}\left(-{\frac {1}{x^{2}}}e^{\frac {1}{x}}\right)=-{\frac {-2}{x^{3}}}\cdot e^{\frac {1}{x}}-{\frac {1}{x^{2}}}\cdot {\frac {-1}{x^{2}}}\cdot e^{\frac {1}{x}}=\left({\frac {2}{x^{3}}}+{\frac {1}{x^{4}}}\right)\cdot e^{\frac {1}{x}}}$

${\displaystyle {\frac {{\textrm {d}}^{3}}{{\textrm {d}}x^{3}}}e^{\frac {1}{x}}={\frac {\textrm {d}}{{\textrm {d}}x}}\left(\left({\frac {2}{x^{3}}}+{\frac {1}{x^{4}}}\right)\cdot e^{\frac {1}{x}}\right)=\left({\frac {-6}{x^{4}}}+{\frac {-4}{x^{5}}}\right)\cdot e^{\frac {1}{x}}+\left({\frac {2}{x^{3}}}+{\frac {1}{x^{4}}}\right)\cdot {\frac {-1}{x^{2}}}\cdot e^{\frac {1}{x}}=-\left({\frac {6}{x^{4}}}+{\frac {6}{x^{5}}}+{\frac {1}{x^{6}}}\right)\cdot e^{\frac {1}{x}}}$

Generalized Laguerre polynomials ${\displaystyle L_{n}^{(\alpha )}(x)}$ are linked to Lah numbers upon setting ${\displaystyle \alpha =-1}$$${\displaystyle n!L_{n}^{(-1)}(x)=\sum _{k=0}^{n}L(n,k)(-x)^{k}}$$This formula is the default Laguerre polynomial in Umbral calculus convention.^[8]

In recent years, Lah numbers have been used in steganography for hiding data in images. Compared to alternatives such as DCT, DFT and DWT, it has lower complexity of calculation—${\displaystyle O(n\log n)}$—of their integer coefficients.^[9][10] The Lah and Laguerre transforms naturally arise in the perturbative description of the chromatic dispersion.^[11][12] In Lah-Laguerre optics, such an approach tremendously speeds up optimization problems.

• Stirling numbers
• Pascal matrix