Zero is overrated

Define a sequence of functions Pi:R+R, iN

P0(x)=ln(x) Pn+1(x)=Pn(x)dx

I found a beautiful closed formula for Pi, that I haven't seen before.

Integrating by parts, it's easy to calculate first few Pi:

P1(x)=xlnxxP2(x)=x22lnx34x2P3(x)=x66lnx1136x3

which leads to a hypothesis, that

Pn=xnn!lnxKnxn

for certain constants Ki, K1=1.

Again integrating by parts, obtain:

Kn+1=1n+1(Kn+1(n+1)!)

from where

Kn=1n(Kn1+1n!)=1n(1n!+1n1(Kn2+1(n1)!))=1n!(1n+1n1)+1n(n1)Kn2=1n!(1n+1n1+1n2)+1n(n1)(n2)Kn3==1n!(1n+1n1++1np+1)+1n(n1)(np+1)Knp

Substitute p=n1:

Kn=1n!(1+12++1n)

Substitute this in the hypothesis:

Pn=xnn!(lnx(1+12++1n))

This nicely contains fragments of exponent, nth-harmonic number and, after a diagonalisation, the Euler constant:

limn+n!nnPn(n)=γ

Why Pi are interesting at all? Because if one completes them for negative indices as

Pn=(1)n1(n1)!xn

then mth-derivative of Pn is Pnm for all non-negative m:

(nZ)(mN)mPn=Pnm

and similarly

(nZ)(mN)mPndx=Pn+m

where m is repeated integral.

This is in contrast with powers xn, n0, which, under repeated derivation, eventually pathetically collapse to a constant and then to 0, so that negative powers are not reachable from positive and other way around.

It's interesting, which other families (ϕi)iZ are there such that

(mN)mϕn=ϕnm (mN)mϕndx=ϕn+m

and

(nm)ϕnconstϕm

(the latter condition is to avoid degenerate cases)?