1 Introduction

In the following, we let \({{\mathcal {D}}}\) be the space of infinitely differentiable functions with compact support, let \({{\mathcal {D}}}[a,b]\) be the space of infinitely differentiable functions with support contained in the interval [ab], and let \({{\mathcal {D}}'}\) be the space of distributions defined on \({{\mathcal {D}}}\).

we define the locally summable function \(x_+^\lambda ,\) for \(\lambda >-1\), by

$$\begin{aligned} x_+ ^\lambda =\left\{ \begin{array}{cc} x^\lambda , &{}x>0,\\ 0, &{} x<0. \end{array} \right. \end{aligned}$$

The distribution \(x_+^\lambda \) is then defined inductively for \(\lambda <-1\) and \(\lambda \ne -2,-3, \ldots \) by \((x_+^\lambda )' =\lambda x_+^{\lambda -1}.\) It follows that if s is a positive integer and \(-s-1<\lambda < -s\), then

$$\begin{aligned} \langle x_+ ^\lambda ,\varphi (x) \rangle = \int _0^\infty x^\lambda \left[ \varphi (x)- \sum _{k=0} ^{s-1} {\varphi ^{(k)} (0)\over k!} x^k\right] dx \end{aligned}$$

for arbitrary \(\varphi \) in \({{\mathcal {D}}}\). In particular, if \(\varphi \) has its support contained in the interval \([-1,1]\), then

$$\begin{aligned} \langle x_+ ^\lambda ,\varphi (x) \rangle = \int _0^1 x^\lambda \left[ \varphi (x)- \sum _{k=0} ^{s-1} {\varphi ^{(k)} (0)\over k!} x^k\right] dx +\sum _{k=0} ^{s-1} {\varphi ^{(k)} (0)\over k! (\lambda +k+1)} . \end{aligned}$$

The distribution \(x_+^{-s}\) is defined by

$$\begin{aligned} x_+ ^{-s}= {(-1)^{s-1}( \ln x_+ )^{(s)} \over (s-1)!} \end{aligned}$$

for \(s= 1,2, \ldots \) and not as in Gelfand and Shilov [5].

It is easily shown that if \(\varphi \) is an arbitrary function in \({{\mathcal {D}}}[-1,1],\) then

$$\begin{aligned} \langle x_+^{-s} , \varphi (x) \rangle= & {} \int _0^{1} x^{-s} \left[ \varphi (x) - {\sum } _{k=0}^{s-1} \frac{\varphi ^{(k)}(0)}{k!} x^{k} \right] \,dx \nonumber \\&\,\,\, -{\sum }_{k=0}^{s-2} \frac{\varphi ^{(k)}(0)}{(s-k-1) k!} -{\phi (s-1)\over (s-1)!}\varphi ^{(s-1)}(0), \end{aligned}$$
(1)

for \(s=1,2, \ldots ,\) where

$$\begin{aligned} \phi (s)=\left\{ \begin{array}{ll} \sum _{k=1} ^s k^{-1} , &{}s\ge 1,\\ 0, &{} s=0. \end{array} \right. \end{aligned}$$

The composition of a distribution and an infinitely differentiable function is extended to distributions by continuity provided that the derivative of the infinitely differentiable function is different from zero, see [1]. Fisher defined the composition of a distribution F and a summable function f which has a single simple root in the open interval (ab) and it was recently generalized in [6] by allowing f to be a distribution.

Now let \(\rho (x)\) be a function in \({{\mathcal {D}}}\) having the following properties:

  1. (i)

    \(\rho (x) =0\) for \(|x| \ge 1\),      (ii)   \(\rho (x) \ge 0\),

  2. (iii)

    \(\rho (x) =\rho (-x)\),      (iv)   \({\displaystyle {\int _{-1}^1 \rho (x)\,dx =1}}\).

Putting \(\delta _n(x)=n\rho (nx)\) for \(n=1,2,\ldots \) , it follows that \(\{\delta _n(x)\}\) is a regular sequence of infinitely differentiable functions converging to the Dirac delta function \(\delta (x)\).

Further, if F is a distribution in \({{\mathcal {D}}}'\) and \(F_n(x)=\langle F(x-t),\delta _n(x)\rangle ,\) then \(\{F_n(x)\}\) is a regular sequence of infinitely differentiable functions converging to F(x).

The following definition for the neutrix composition of distributions was given in [3] and originally called the composition of distributions.

Definition 1.1

Let F be a distribution in \({{\mathcal {D}}}'\) and let f be a locally summable function. We say that the neutrix composition F(f(x)) exists and is equal to h on the open interval (ab) if

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }\int _{-\infty }^{\infty } F_{n}(f(x))\varphi (x) dx = \langle h(x) , \varphi (x) \rangle \end{aligned}$$

for all \(\varphi \) in \({{\mathcal {D}}}[a,b],\) where \(F_n(x) =F(x)*\delta _n(x)\) for \(n=1,2, \ldots \) and N is the neutrix, see [2], having domain \(N'\) the positive integers and range \(N''\) the real numbers, with negligible functions which are finite linear sums of the functions

$$\begin{aligned} n^{\lambda } \ln ^{r-1} n, \, \ln ^{r} n: \lambda > 0, \, r=1,2,\ldots \end{aligned}$$

and all functions which converge to zero in the usual sense as n tends to infinity.

In particular, we say that the composition F(f(x)) exists and is equal to h on the open interval (ab) if

$$\begin{aligned} \mathop {\mathrm{lim}}\limits _{n\rightarrow \infty }\int _{-\infty }^{\infty } F_{n}(f(x))\varphi (x) dx = \langle h(x) , \varphi (x) \rangle \end{aligned}$$

for all \(\varphi \) in \({{\mathcal {D}}}[a,b].\)

Note that if a function f(n) tends to \(\alpha \) in the usual sense as n tends to infinity, it converges to \(\alpha \) in the neutrix sense. The reader may find the general definition of the neutrix limit with some examples in [2].

2 Results

Nicholas and Fisher defined the composition \((x_+^r)^{-s}\) as the neutrix limit of regular sequence \([(x_+^r)^{-s}]_n\) for \(r,s =1,2,\ldots ,\) see [7]. Further Öz c̣ ağ et. al. consider the case \(r=0\), in other words the s-th power of the Heaviside function H(x) defined by \([H(x)]^{-s} =H(x),\) see [9]. Recently, some compositions such as \((|x|^{r-1/2})^{-4s}\) and \((|x|^\mu )^{-s}\) were defined in [4, 8] respectively.

We first of all need the following Lemma which can be easily proved by induction.

Lemma 2.1

$$\begin{aligned} \int _{-1} ^1 v^i \rho ^{(r)} (v) \,dv = \left\{ \begin{array} {cc} 0,&{} 0 \le i<r, \\ (-1)^r r!,&{} i=r \end{array} \right. \end{aligned}$$
(2)

and because \(v^r\rho ^{(r)}\) is an even function, we have

$$\begin{aligned} \int _{-1}^0 v^r\rho ^{(r)}(v)\,dv =\int _0^1 v^r\rho ^{(r)}(v)\,dv = \frac{1}{2}(-1)^rr! \end{aligned}$$
(3)

and

$$\begin{aligned} \int _{-1}^0 v^r\ln |v|\rho ^{(r)}(v)\,dv= & {} \int _0^1 v^r\ln |v|\rho ^{(r)}(v)\,dv \nonumber \\= & {} \frac{1}{2}(-1)^rr!\phi (r)+(-1)^rr!c(\rho ) \end{aligned}$$
(4)

for \(r=0,1,2,\ldots \) where \(c(\rho )=\int _0^1\ln t\rho (t)\,dt.\)

We now prove the following theorem.

Theorem 2.2

The distribution \((x_+^\mu )^{-m}\) exists and

$$\begin{aligned} (x_+^\mu )^{-m}=x_+^{-s}-(-1)^s\frac{(-1)^mm!\left[ 2c(\rho )+\phi (m-1)\right] +s\phi (s-1)}{s!}\delta ^{(s-1)}(x) \end{aligned}$$
(5)

for \(\mu >0, \quad m=1,2,\ldots \) and \(\mu m=s (s\in {{\mathbb {Z}}}^+).\)

In particular, we have

$$\begin{aligned} \left( x_+^\frac{1}{r}\right) ^{-r} = x_+^{-1}-(-1)^rr!\left[ 2c(\rho )+\phi (r-1)\right] \delta (x) \end{aligned}$$
(6)

for \(r=1,2,\ldots .\)

Proof

We first put

$$\begin{aligned} (-1)^{m-1}(m-1)![(x_+^\mu )^{-m}]_n = \left\{ \begin{array}{cc} \int _{-1/n}^{1/n} \ln |x^\mu - t| \delta _n^{(m)}(t) dt , &{} if \quad x\ge 0 \\ \int _{-1/n}^{1/n} \ln |t| \delta _n^{(m)}(t) dt , &{} if \quad x<0. \end{array} \right. \end{aligned}$$
(7)

Then

$$\begin{aligned}&(-1)^{m-1}(m-1)! \int _{-1}^1 x^k [(x_+^\mu )^{-m}]_n\,dx = \int _{-1}^{1} x^k \int _{-1/n}^{1/n} \ln |x_+^\mu - t| \delta _n^{(m)}(t) dt dx \nonumber \\&\quad = \int _{-1/n}^{1/n} \delta _n ^{(m)} (t) \int _0^{n^{-1/\mu }} x^k \ln |x^\mu -t| \,dx \,dt \nonumber \\&\qquad +\int _{-1/n}^{1/n} \delta _n ^{(m)}(t) \int _{n^{-1/\mu }}^1 x^k \ln |x^\mu -t|\,dx \,dt +\int _{-1}^0 x^k \int _{-1/n}^{1/n} \ln |t|\delta _n ^{(m)}(t) \,dx\,dt \nonumber \\&\quad = \frac{n^{m-(k+1)/\mu }}{\mu } \int _{-1}^1 \rho ^{(m)}(v) \int _0 ^1 y^{-1+(k+1)/\mu }\ln |y-v| \,dy\,dv \nonumber \\&\qquad +\,\frac{n^{m-(k+1)/\mu }}{\mu } \int _{-1}^1 \rho ^{(m)}(v)\int _ 1^n y^{-1+(k+1)/\mu }\ln |y-v| \,dy\,dv \nonumber \\&\qquad +\,\frac{n^{m-(k+1)/\mu }}{\mu }\ln n\int _{-1}^1 \rho ^{(m)}(v)\,dv\int _0^n y^{-1+(k+1)/\mu }\,dy \nonumber \\&\qquad +\,\frac{(-1)^{k+1}n^m}{k+1}\int _{-1}^1\ln |v/n|\rho ^{(m)}(v)\,dv \nonumber \\&\quad = I_1 + I_2+I_3+I_4, \end{aligned}$$
(8)

on using the substitutions \(y = nx^\mu \) and \(v = nt.\)

It follows immediately that

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }I_3 = 0\qquad and \qquad \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }I_4 = 0 \end{aligned}$$
(9)

for \(k=0,1,\ldots \) and

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }I_1 = 0 \end{aligned}$$
(10)

for all \(k=0,1,2,\ldots ,s-2.\)

Further

$$\begin{aligned} \int _1^n y^{-1+(k+1)/\mu }\ln |y-v| \,dy= & {} \int _1^n y^{-1+(k+1)/\mu }\ln y\,dy \nonumber \\&+\int _1^n y^{-1+(k+1)/\mu }\ln |1-v/y| \,dy \nonumber \\= & {} I_2' +I_2'', \end{aligned}$$
(11)

where

$$\begin{aligned} I_2' =\frac{\mu n^{(k+1)/\mu } \ln n}{(k+1)}+\frac{\mu ^2[1-n^{(k+1)/\mu } ]}{(k+1)^2} \end{aligned}$$
(12)

and

$$\begin{aligned} I_2''= & {} -\sum _{i=1}^\infty \frac{v^i}{i}\int _1^n y^{-1-i+(k+1)\mu } \,dy \nonumber \\= & {} -\sum _{i=1}^\infty \frac{v^i\mu [n^{-i+(k+1)/\mu }-1]}{i(k+1-\mu i)} \end{aligned}$$
(13)

\(k=0,1,2,\ldots ,s-2.\)

It follows from Lemma 2.1 and Eqs. (11), (12), and (13) that

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }I_2 =\frac{(-1)^m (m-1)!}{s-k-1} \end{aligned}$$
(14)

for \(k=0,1,\ldots ,s-2,\) and it then follows from Eqs. (8), (9), (10), and (14) that

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }\int _{-1}^1 x^k \left[ (x_+^\mu )^{-m} \right] _n \,dx = -\frac{1}{s-k-1} \end{aligned}$$
(15)

\(k=0,1,2,\ldots ,s-2.\)

When \(k=s-1,\) then we have

$$\begin{aligned} I_1= & {} \frac{1}{\mu }\int _{-1}^1\rho ^{(m)}(v) \int _0^1 y^{m-1}\ln |y-v| \,dy\,dv \nonumber \\= & {} \frac{1}{\mu } \int _0^1\rho ^{(m)}(v) \left[ \int _0^v + \int _v^1 y^{m-1} \ln |y-v|\,dy\right] \,dv \nonumber \\&\,\,\, + \frac{1}{\mu } \int _{-1}^0 \rho ^{(m)}(v) \left[ \int _0^{-v} + \int _{-v}^1 y^{m-1} \ln |y-v| \,dy\right] \,dv \nonumber \\= & {} J_1+J_2+J_3+J_4. \end{aligned}$$
(16)

Now making the substitution \( y=uv\)

$$\begin{aligned} J_1 = \frac{1}{\mu } \int _0^1 v^m \rho ^{(m)}(v)\int _0^1 u^{m-1}\left[ \ln v+\ln (1-u)\right] \,du\,dv, \end{aligned}$$
(17)

and using Lemma 2.1, we have

$$\begin{aligned} \int _0^1 v^m \rho ^{(m)}(v)\ln v\int _0^1 u^{m-1}\,du\,dv =(-1)^m(m-1)! \left[ c(\rho ) +\frac{1}{2}\phi (m) \right] \end{aligned}$$
(18)

and

$$\begin{aligned}&\int _0^1 v^m \rho ^{(m)}(v)\int _0^1 u^{m-1}\ln (1-u)\,du\,dv \nonumber \\&\quad =\frac{1}{2}(-1)^m(m-1)!\int _0^1\ln (1-u)d(u^m-1)\nonumber \\&\quad = \frac{1}{2}(-1)^m(m-1)!\int _0^1\frac{u^m-1}{1-u}\,du \nonumber \\&\quad = \frac{1}{2}(-1)^{m-1}(m-1)!\phi (m), \end{aligned}$$
(19)

and it follows from Eqs. (17), (18), and (19) that

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }J_1 = \frac{(-1)^m(m-1)! c(\rho )}{\mu }. \end{aligned}$$
(20)

Similarly, using the substitution \(y=uv\) again we have

$$\begin{aligned} J_3 = - \frac{1}{\mu } \int _{-1}^0 v^m \rho ^{(m)}(v)\int _{-1}^0 u^{m-1}\left[ \ln |v|+\ln (1-u)\right] \,du\,dv. \end{aligned}$$
(21)

Thus

$$\begin{aligned} \int _{-1}^0 v^m \rho ^{(m)}(v)\ln |v|\int _{-1}^0 u^{m-1}\,du\,dv= & {} \frac{(-1)^{m-1}}{m}\int _{-1}^0 v^m \rho ^{(m)}(v)\ln |v|\,dv \nonumber \\= & {} -(m-1)!\left[ c(\rho )+\frac{1}{2}\phi (m) \right] , \end{aligned}$$
(22)

and

$$\begin{aligned}&\int _{-1}^0 v^m \rho ^{(m)}(v)\int _{-1}^0 u^{m-1}\ln (1-u)\,du\,dv \nonumber \\&\quad =\frac{1}{2}(-1)^m(m-1)!\int _{-1}^0\ln (1-u)d(u^m-1) \nonumber \\&\quad =\frac{1}{2}[(-1)^m-1](m-1)!\ln 2-\frac{1}{2}(-1)^m(m-1)!\int _{-1}^0\frac{u^m-1}{u-1}\,du \nonumber \\&\quad = \frac{1}{2}[(-1)^m-1](m-1)!\ln 2+\frac{1}{2}(-1)^m(m-1)!\sum _{i=1}^m \frac{(-1)^i}{i}. \end{aligned}$$
(23)

It follows from Eqs. (21), (22), and (23) that

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }J_3= & {} \frac{[1-(-1)^m] (m-1)!}{2\mu }\ln 2+\frac{(m-1)!}{2\mu }\phi (m)+\nonumber \\&\,\,\,-\frac{(m-1)!}{2\mu }\sum _{i=1}^m \frac{(-1)^i}{i}+\frac{(m-1)!}{\mu }c(\rho ). \end{aligned}$$
(24)

Further

$$\begin{aligned} J_2= & {} \frac{1}{\mu }\int _0^1 \rho ^{(m)}(v)\int _v^1 y^{m-1} \left[ \ln y+\ln (1-v/y)\right] \,dy\,dv \nonumber \\= & {} \frac{1}{\mu }\int _0^1 \rho ^{(m)}(v)\int _v^1 y^{m-1}\ln y\,dy\,dv-\frac{1}{\mu }\sum _{i=1}^\infty \frac{1}{i}\int _0^1 v^i\rho ^{(m)}(v)\int _v^1 y^{m-i-1}\,dy\,dv \nonumber \\= & {} \frac{(-1)^m(m-1)!}{2\mu m}-\frac{1}{\mu m^2}\int _0^1 \rho ^{(m)}(v)\,dv + \nonumber \\&\,\,\,-\frac{1}{\mu }\sum _{i=1, \,i\ne m}^\infty \frac{1}{i(m-i)}\int _0^1(v^i-v^m)\rho ^{(m)}(v)\,dv \nonumber \\= & {} \frac{(-1)^m(m-1)!}{2\mu m}+\frac{\rho ^{(m-1)}(0)}{\mu m^2}+\frac{(-1)^m(m-1)!}{2\mu } \left[ 2\phi (m-1)- \phi (m)\right] \nonumber \\&\,\,\,-\frac{1}{\mu }\sum _{i=1, \,i\ne m}^\infty \frac{1}{i(m-i)}\int _0^1v^i\rho ^{(m)}(v)\,dv \nonumber \\= & {} \frac{\rho ^{(m-1)}(0)}{\mu m^2}+\frac{(-1)^m(m-1)!}{2\mu }\phi (m-1)+ \nonumber \\&\,\,\, -\frac{1}{\mu }\sum _{i=1, \,i\ne m}^\infty \frac{1}{i(m-i)}\int _0^1v^i\rho ^{(m)}(v)\,dv \end{aligned}$$
(25)

because

$$\begin{aligned} \sum _{i=1, \,i\ne m}^\infty \frac{1}{i(m-i)} =\frac{2\phi (m-1)-\phi (m)}{m}= \frac{\phi (m-1)}{m}-\frac{1}{m^2}. \end{aligned}$$

Finally, we have

$$\begin{aligned} J_4= & {} \frac{1}{\mu }\int _{-1}^0 \rho ^{(m)}(v)\int _{-v}^1 y^{m-1} \left[ \ln y+\ln (1-v/y)\right] \,dy\,dv \nonumber \\= & {} \frac{1}{\mu }\int _{-1}^0 \rho ^{(m)}(v)\int _{-v}^1 y^{m-1}\ln y\,dy\,dv-\frac{1}{\mu }\sum _{i=1}^\infty \frac{1}{i}\int _{-1}^0 v^i\rho ^{(m)}(v)\int _{-v}^1 y^{m-i-1}\,dy\,dv \nonumber \\= & {} \frac{1}{\mu }\int _{-1}^0\left[ \frac{(-v)^m-1}{m^2}-\frac{(-v)^m\ln |v|}{m} \right] \rho ^{m)}(v)\,dv+ \frac{1}{\mu m}\int _{-1}^0 v^m\ln |v|\rho ^{(m)}(v)\,dv \nonumber \\&\,\,\,-\frac{1}{\mu }\sum _{i=1, \,i\ne m}^\infty \frac{1}{i(m-i)}\int _{-1}^0 \left[ v^i-(-1)^{m-i}v^m) \right] \rho ^{(m)}(v)\,dv \nonumber \\= & {} \frac{(m-1)!}{2\mu m}-\frac{\rho ^{(m-1)}(0)}{\mu m^2}-\frac{[1-(-1)^m]m!}{2\mu } \left[ \phi (m)+2c(\rho )\right] + \nonumber \\&\,\,\,-\frac{1}{\mu }\sum _{i=1, \,i\ne m}^\infty \frac{1}{i(m-i)}\int _{-1}^0 v^i \rho ^{(m)}(v)\,dv-\frac{(-1)^m(m-1)!}{2\mu m} \nonumber \\&\,\,\,+\frac{(m-1)!}{2\mu }\sum _{i=1}^{m-1} \frac{(-1)^{m+i}}{i} -\frac{[1-(-1)^m] (m-1)!\ln 2}{2\mu } \end{aligned}$$
(26)

because

$$\begin{aligned} \sum _{i=1, \,i\ne m}^\infty \frac{1}{i(m-i)} =-\frac{(-1)^m}{m^2}+\frac{1}{m}\sum _{i=1}^{m-1} \frac{(-1)^{m+i}}{i}- \frac{[1-(-1)^m]}{m}\ln 2 . \end{aligned}$$

Next, we similarly have

$$\begin{aligned} I_2= & {} \frac{1}{\mu }\int _{-1}^1 \rho ^{(m)}(v)\int _1^n y^{m-1} \left[ \ln y+\ln (1-v/y)\right] \,dy\,dv \\= & {} \frac{1}{\mu }\int _{-1}^1 \rho ^{(m)}(v)\int _1^n y^{m-1} \ln (1-v/y) \,dy\,dv \\= & {} -\frac{1}{\mu }\sum _{i=1}^\infty \frac{1}{i} \int _{-1}^1 v^i\rho ^{(m)}(v) \int _1^n y^{m-i-1}\,dy\,dv \\= & {} -\frac{(-1)^m(m-1)!\ln n}{\mu }-\frac{1}{\mu }\sum _{i=m+1}^\infty \frac{1}{i(i-m)} \int _{-1}^1(n^{m-i}-1)v^i\rho ^{(m)}(v)\,dv, \end{aligned}$$

and so

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }I_2 = -\frac{1}{\mu }\sum _{i=m+1}^\infty \frac{1}{i(i-m)} \int _{-1}^1 v^i\rho ^{(m)}(v)\,dv. \end{aligned}$$
(27)

It now follows from Eqs. (8), (9), (16), (20), and (24)–(27) that

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }\int _{-1}^1 x^{s-1}\left[ (x_+^\mu )^{-m} \right] _n \,dx= & {} \frac{(-1)^m(m-1)!}{\mu } \left[ 2c(\rho )+\phi (m-1) \right] \nonumber \\= & {} \frac{(-1)^mm!}{s} \left[ 2c(\rho )+\phi (m-1) \right] . \end{aligned}$$
(28)

We now consider the case \(k=s.\) If \(x<0\) and \(\psi \) is an arbitrary continuous function, then

$$\begin{aligned}&(-1)^{m-1}(m-1)!\int _{-1}^0 x^s \left[ (x_+^\mu )^{-m} \right] _n \psi (x)\,dx \\&\quad =\int _{-1}^0 x^s \psi (x)\int _{-1/n}^{1/n} \ln |t| \delta _n^{(m)}(t)\,dt\,dx \\&\quad =n^m \int _{-1}^0 x^s\psi (x)\,dx \int _{-1}^1 \ln |v/n|\rho ^{(m)}(v)\,dv, \end{aligned}$$

where \(v=nt.\) Thus

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }\int _{-1}^0 x^s \left[ (x_+^\mu )^{-m}\right] _n \psi (x)\,dx =0. \end{aligned}$$
(29)

Next with \(k=s\) in \(I_1, \) we see that

$$\begin{aligned} \int _0^{n^{-1/\mu }} x^s \left[ (x_+^\mu )^{-m} \right] _n \,dx=\frac{n^{-1/\mu }}{\mu } \int _{-1}^1 \rho ^{(m)}(v)\int _0^1 y^{m-1+1/\mu } \ln |(y-v)/n| \,dy\,dv \end{aligned}$$

and if \(\psi \) is any continuous function, then

$$\begin{aligned} \mathop {\mathrm{lim}}\limits _{n\rightarrow \infty }\int _0^{n^{-1/\mu }} x^s \left[ (x_+^\mu )^{-m} \right] _n \psi (x)\,dx =0. \end{aligned}$$
(30)

If \(x^\mu \ge \frac{1}{n}\), then we have

$$\begin{aligned} (-1)^{m-1}(m-1)!\left[ (x_+^\mu )^{-m} \right] _n= & {} \int _{-1/n}^{1/n} \ln |x^\mu -t| \delta _n^{(m)}(t)\,dt \\= & {} n^m \int _{-1}^1 \ln |x^\mu -v/n| \rho ^{(m)}(v)\,dv \\= & {} n^m \int _{-1}^1\left[ \ln |x^\mu -\sum _{i=1}^\infty \frac{v^i}{in^ix^{\mu i}} \right] \rho ^{(m)}(v)\,dv \\= & {} -\sum _{i=m}^\infty \int _{-1}^1\frac{v^i}{in^{i-m}x^{\mu i}}\rho ^{(m)}(v)\,dv, \end{aligned}$$

and it follows that

$$\begin{aligned} \left| (m-1)!\left[ (x_+^\mu )^{-m} \right] _n \right|\le & {} \sum _{i=m}^\infty \int _{-1}^1\frac{|v|^i}{in^{i-m}x^{\mu i}}|\rho ^{(m)}(v)|\,dv \\\le & {} \sum _{i=m}^\infty \frac{\kappa _m}{in^{i-m}x^{\mu i}}, \end{aligned}$$

where \(\kappa _m =\int _{-1}^1 |\rho ^{(m)}(v)|\,dv \) for \(m=1,2,\ldots .\)

If now \(n^{-1/\mu }<\eta <1 \)

$$\begin{aligned}&(m-1)! \int _{n^{-1/\mu }}^\eta \left| x^s \left[ (x_+^\mu )^{-m} \right] _n \right| \,dx \le \kappa _m\sum _{i=m}^\infty \frac{n^{m-i}}{i} \int _{n^{-1/\mu }}^\eta x^{s-\mu i} \,dx \\&\quad =\kappa _m \sum _{i=m}^\infty \frac{n^{-1/\mu }}{\mu i} \int _1^{n\eta ^\mu } y^{m-i+1/\mu -1}\,dy \\&\quad =\left\{ \begin{array}{ll} \kappa _m \sum _{i=m}^\infty \frac{n^{-1/\mu }}{\mu i(m-i+1/\mu )} \left[ (n\eta ^\mu )^{m-i+1/\mu }-1\right] , &{} \mu \ne 1 \\ \kappa _m \sum _{i=m,i\ne m+1}^\infty \frac{n^{-1}}{i(m-i+1)} \left[ (n\eta )^{m-i+1}-1\right] +\frac{\kappa _m n^{-1}\ln (n\eta )}{m+1} , &{} \mu =1.\end{array} \right. \end{aligned}$$

It follows that

$$\begin{aligned} \mathop {\mathrm{lim}}\limits _{n\rightarrow \infty }\left| \left[ (x_+^\mu )^{-m} \right] _n \right| = O(\eta ) \end{aligned}$$

for \(m=1,2,\ldots \) and if \(\psi \) is any continuous function then

$$\begin{aligned} \mathop {\mathrm{lim}}\limits _{n\rightarrow \infty }\left| \int _{n^{-1/\mu }}^\eta x^s \left[ (x_+^\mu )^{-m} \right] _n \psi (x)\,dx \right| = O(\eta ) \end{aligned}$$
(31)

for \(m=1,2,\ldots .\) Now let \(\varphi (x)\in {{\mathcal {D}}}[-1,1].\) By Taylor’s theorem, we have

$$\begin{aligned} \varphi (x) = \sum _{k=0}^{s-1}\frac{\varphi ^{(k)}(0)}{k!} x^k +\frac{x^s}{s!} \varphi ^{(s)}(\xi x) \qquad (0<\xi <1). \end{aligned}$$

Then

$$\begin{aligned} \langle \left[ (x_+^\mu )^{-m} \right] _n,\varphi (x) \rangle= & {} \sum _{k=0}^{s-1}\frac{\varphi ^{(k)}(0)}{k!}\int _{-1}^1 x^k \left[ (x_+^\mu )^{-m} \right] _n \,dx \\&+\frac{1}{s!}\int _{-1}^0 x^s\left[ (x_+^\mu )^{-m} \right] _n \varphi ^{(s)}(\xi x)\,dx \\&+\frac{1}{s!} \int _0^{n^{-1/\mu }} x^s \left[ (x_+^\mu )^{-m} \right] _n\varphi ^{(s)}(\xi x)\,dx \\&+\frac{1}{s!}\int _{n^{-1/\mu }}^\eta x^s \left[ (x_+^\mu )^{-m} \right] _n\varphi ^{(s)}(\xi x)\,dx \\&+\frac{1}{s!} \int _\eta ^1 x^s \left[ (x_+^\mu )^{-m} \right] _n\varphi ^{(s)}(\xi x)\,dx. \end{aligned}$$

Using Eqs. (15) and (28)–(31) and noting that the sequence \(\{ \left[ (x_+^\mu )^{-m} \right] _n \}\) converges uniformly to \(x^{-s}\) on the interval \([\eta ,1]\), it follows that

$$\begin{aligned} \mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }\langle \left[ (x_+^\mu )^{-m} \right] _n,\varphi (x) \rangle= & {} \frac{(-1)^mm!}{s!} \left[ 2c(\rho )+\phi (m-1)\right] \varphi ^{(s-1)}(0) \\&\,\,\, -\sum _{k=0}^{s-2}\frac{\varphi ^{(k)}(0)}{k!(s-k-1)} +\int _\eta ^1 \varphi ^{(s)}(\xi x) \,dx+ O(\eta ). \end{aligned}$$

Because \(\eta \) can be arbitrarily small, it follows that

$$\begin{aligned}&\mathop {\mathrm{N}-\mathrm{lim}}\limits _{n\rightarrow \infty }\langle \left[ (x_+^\mu )^{-m} \right] _n,\varphi (x) \rangle \nonumber \\&\quad =\frac{(-1)^mm!}{s!} \left[ 2c(\rho )+\phi (m-1)\right] \varphi ^{(s-1)}(0)-\sum _{k=0}^{s-2}\frac{\varphi ^{(k)}(0)}{k!(s-k-1)}\\&\qquad +\int _\eta ^1 \varphi ^{(s)}(\xi x) \,dx+ O(\eta ). \\&\quad =\int _0^1 x^{-s} \left[ \varphi (x)-\sum _{k=0}^{s-1} \frac{\varphi ^{(k)}(0)}{k!} x^k \right] \,dx-\sum _{k=0}^{s-2} \frac{\varphi ^{(k)}(0)}{k!(s-k-1)} \\&\qquad +\frac{(-1)^mm!}{s!} \left[ 2c(\rho )+\phi (m-1)\right] \varphi ^{(s-1)}(0) \\&\quad =\langle x_+^{-s}, \varphi (x) \rangle + \\&\qquad -(-1)^s\frac{(-1)^mm!\left[ 2c(\rho )+\phi (m-1)\right] +s\phi (s-1)}{s!}\langle \delta ^{(s-1)}(x),\varphi (x)\rangle \end{aligned}$$

on using Eq. (1). This proves Eq. (5) on the interval \([-1,1].\) However, Eq. (5) clearly holds on any interval not containing the origin, and the proof is complete. \(\square \)

Note that if \(\mu \in {{\mathbb {Z}}}^+,\) then Theorem 1 is in agreement with the Theorem 3 given in [7].

Corollary 2.3

The distribution \((x_-^\mu )^{-m}\) exists and

$$\begin{aligned} (x_-^\mu )^{-m}=x_-^{-s}+\frac{(-1)^mm!\left[ 2c(\rho )+\phi (m-1)\right] +s\phi (s-1)}{s!}\delta ^{(s-1)}(x) \end{aligned}$$
(32)

for \(\mu >0, \quad m=1,2,\ldots \) and \(\mu m=s\in {{\mathbb {Z}}}^+.\)

Proof

Equation (32) follows after replacing x by \(-x\) in Eq. (5). \(\square \)