Did you test this? At all? (These are rhetorical questions.)
Here are the relevant computations.
ee = HypergeometricPFQ[{1, 1, 1, 1}, {2, 2, 2, 2}, -x];
First the series from the second pod.
ss = Normal[Series[ee, {x, Infinity, 5}]]
(* Out[477]=
E^-x (-(6/x^5) + 1/x^4) + (1/(
12 x))(2 EulerGamma^3 + EulerGamma \[Pi]^2 +
6 EulerGamma^2 Log[x] + \[Pi]^2 Log[x] + 6 EulerGamma Log[x]^2 +
2 Log[x]^3 - 2 PolyGamma[2, 1]) *)
Quick numerical check.
{ee, ss} /. x -> 1111.1
(* Out[479]= {0.0715749986818, 0.0715749986818} *)
Good so far. Now for that third pod. I rewrote ever so slightly to use the more common positive exponent notation.
ss2 = Sum[(-1)^n*x^n/((n + 1)^4*Factorial[n]), {n, 0, Infinity}]
ss2 === ee
(* Out[480]= HypergeometricPFQ[{1, 1, 1, 1}, {2, 2, 2, 2}, -x]
Out[481]= True *)
To restate, it automatically evaluated to the input expression. We can check numerically that this is plausible.
terms = 4000;
NSum[(-1)^n*(1111 + 1/10)^n/((n + 1)^4*Factorial[n]), {n, 0, terms},
WorkingPrecision -> 1000, NSumTerms -> terms]
(* Out[504]=
0.07157499868179492615228113024556896152461872165957332195911424357595\
7027394013518883378928089038297513792617279360234433705402207531750966\
0717006728987049036451842823432573305827850163550084804675769957660934\
4824128631596694605973285589988236274789694902987106035520518673838341\
8403680202985945005798435970592342850033187925346634452756821088978508\
5259444658718487981866392080437473574473862047376793571126464969800787\
8803181449171740731887666924329532184773807981764974033314775480749947\
1096391441816719897059492637173806116477129407659923386426954912528103\
9169790296933292120323513441841312866076886262356996247624553397188136\
7527417883237598541289814556372772783972646774583249135592744712362526\
9777934478025389513686897128733758151270977948985716655740364364858294\
3580719151581187700654582635529859694967679171786536858445379972974013\
6089392197806995711217621078952538615859473022861414912694258074963677\
5180352839003018355521752110323676919920539509259744224721471269246962\
90624132618936092170922 *)
(Exercise: Show that 4000 terms suffices to get a good approximation.)
Or use Sum
with a numeric input for x
:
Sum[(-1)^n*(1111.1)^n/((n + 1)^4*Factorial[n]), {n, 0, Infinity}]
(* Out[495]= 0.0715749986818 *)
Since careful verification is the order of the day, I'll show the variant from the W|A third pod as well:
Sum[(-1)^(-n)*x^(-n)/((n - 1)^4*Factorial[-n]), {n, 0, -Infinity, -1}]
(* Out[505]= HypergeometricPFQ[{1, 1, 1, 1}, {2, 2, 2, 2}, -x] *)
Regarding the nature of series expansions: many functions have Puiseux expansions at infinity. This one does not. I agree the first pod either should not be present, or should state more clearly what is intended. But that does not make it wrong, just a bit confusing when taken in tandem with the next two.