For the function f[x] defined in [1] below, evaluation [5] below is inconsistent with evaluations [2] to [4] below. Note the change in sign preceding y in evaluation [5] below.
In[1]:= f[x_] := DiracDelta'[x - 1]
In[2]:= MellinConvolve[f[x], h[x], x, y, GenerateConditions -> True]
Out[2]= h[y] + y Derivative[1][h][y]
In[3]:= MellinConvolve[h[x], f[x], x, y, GenerateConditions -> True]
Out[3]= h[y] + y Derivative[1][h][y]
In[4]:= Integrate[f[x] h[y/x] 1/x, {x, 0, \[Infinity]},
GenerateConditions -> True]
Out[4]= h[y] + y Derivative[1][h][y]
In[5]:= Assuming[y > 0,
FullSimplify[
Integrate[h[x] f[y/x] 1/x, {x, 0, \[Infinity]},
GenerateConditions -> True]]]
Out[5]= h[y] - y Derivative[1][h][y]
Question 1: Does evaluation [5] above represent an error in the implementation of the Mathematica Integrate function?
For the function g[x] defined in [6] below, evaluations [7] and [8] below are inconsistent with evaluations [9] and [10] below. Note Mellin convolution is claimed to be commutative.
In[6]:= g[x_] := x DiracDelta'[x - 1]
In[7]:= MellinConvolve[g[x], h[x], x, y, GenerateConditions -> True]
Out[7]= y Derivative[1][h][y]
In[8]:= Integrate[g[x] h[y/x] 1/x, {x, 0, \[Infinity]},
GenerateConditions -> True]
Out[8]= y Derivative[1][h][y]
In[9]:= MellinConvolve[h[x], g[x], x, y, GenerateConditions -> True]
Out[9]= 2 h[y] - y Derivative[1][h][y]
In[10]:= Assuming[y > 0,
FullSimplify[
Integrate[h[x] g[y/x] 1/x, {x, 0, \[Infinity]},
GenerateConditions -> True]]]
Out[10]= 2 h[y] - y Derivative[1][h][y]
Question 2: Do the discrepancies between evaluations [7] to [10] above represent errors in the Mathematica MellinConvolve and Integrate functions or is Mellin convolution not always commutative with respect to evaluations involving distributions?