New Series Involving Binomial Coefficients

Preprint, arXiv:2307.
03086
NEW SERIES INVOLVING BINOMIAL COEFFICIENTS

arXiv:2307.03086v6 [math.NT] 16 Nov 2023
ZHI-WEI SUN
Abstract. In this paper, we deduce several new identities on infinite

series with denominators of summands containing both binomial coeffi-
cients and linear parts. For example, we evaluate the sums
∞ ∞ ∞
X xk0 X xk0 X xk0
3k
, 3k
and
(3k + 2) 3k

k=1
(2k − 1) k k=0
(3k + 1) k k=0 k
for any x0 ∈ (−27/4, 27/4). For any 1 < n 6 85/4, we obtain the
following fast converging series for log n:
(2(n2 + 6n + 1)2 k + n4 + 30n2 + 1)(n − 1)4k
X∞
(4k + 1)(−n)k (n + 1)2k 4k

k=0 2k
= 8n(n + 1)2 − 2n(n2 − 1) log n.

We also prove that
153965k2 − 96459k + 22786
∞
X 64
= (1634 − 45 log 2).
k 4k 5

k=0
(−8) k
In addition, we pose many new conjectural series identities involving

binomial coefficients; for example, we conjecture that
2k 3
∞
X X 1 25 5
k
9(42k + 5) + = π3.
4096k (2j + 1)4 (2k + 1)3 6
k=0 06j<k
1. Introduction
The curious identity
∞
X 25k − 3 π
3k
= , (1.1)
k
2 k 2
k=0
was first announced by R. W. Gosper in 1974, and later used by Bellard [6]
to find an algorithm for computing the nth decimal of π without computing
the earlier ones. Almkvist, Krattenthaler and Petersson [3] gave a proof of
the Gosper identity 1.1 by using the beta function
Z 1
B(a, b) = xa−1 (1 − x)b−1 dx (a > 0 and b > 0).
0
Key words and phrases. Binomial coefficients, combinatorial identities, infinite series,
congruences.
2020 Mathematics Subject Classification. Primary 05A19, 11B65; Secondary 11A07,
11B68, 33B15.
Supported by the Natural Science Foundation of China (grant no. 12371004).
1
2 ZHI-WEI SUN
A celebrated result of Euler states that

Γ(a)Γ(b)
B(a, b) = ,
Γ(a + b)
where the Gamma function is given by
Z +∞ x−1
t
Γ(x) = dt for x > 0.
0 et
Motivated by Gosper’s identity (1.1), the author [20] evaluated
∞ ∞
X xk0 X kxk0
3k and 3k
k=0 k k=0 k
for any x0 ∈ (−27/4, 27/4). As observed in [20], x3 = x0 (x − 1) for some

x ∈ (−3, c), where
!
√
q
3 3 1
c= 1+ 2− p √ = 0.8941 . . . . (1.2)
2 3
1+ 2
Combining the two parts of [20, Theorem 1.1], we find that
∞
X x3k 27(1 − x) 3x(x − 1)
= + log(1 − x)
1)k 3k (x + 3)(2x − 3)2 (2x − 3)3

k=0
(x − k
(1.3)
2x(x − 1)(x2 − 12x + 9)q(x)
+ p ,
(x + 3)(2x − 3)3 (1 − x)(x + 3)
where  q
x 3+x


 arctan x+2 1−x if − 2 < x < 1,
q(x) = − 2π (1.4)
if x = −2,
 q
arctan x
 3+x
x+2 1−x − π if − 3 < x < −2.
In view of the above, it is natural to ask whether for some a, b ∈ Z \ {0}
we can evaluate
∞
X x3k
,
k=0
(ak + b)(x − 1)k 3k
k
where x ∈ (−3, c). In this direction, we obtain the following new result.
Theorem 1.1. Let −3 < x < c with x 6= 0, where the constant c is given by
(1.2). Then
∞
X x3k x(1 − x) log(1 − x) 2x(x2 − 3)q(x)
3k
= − p .
(2k − 1)(x − 1)k k
2x − 3 3(2x − 3) (1 − x)(x + 3)
k=1
(1.5)
Also,
∞
x3k
r
X 3(1 − x) log(1 − x) (x − 3)q(x) 1−x
3k
= + (1.6)
(3k + 1)(x − 1)k
k
2x(2x − 3) x(2x − 3) x+3
k=0
NEW SERIES INVOLVING BINOMIAL COEFFICIENTS 3
and
∞
X x3k 3(3 − x)(1 − x) log(1 − x)
=
1)k 3k 2x2 (2x − 3)

k=0
(3k + 2)(x − k (1.7)
(x2 + 6x − 9)q(x) 1 − x
r
− ,
x2 (2x − 3) x+3
√
Taking x = − 3 in (1.5), we immediately obtain the following corollary.
Corollary 1.1. We have
√
∞
X ( 9−32 3 k
) √ √
3k = ( 3 − 1) log( 3 + 1). (1.8)
k=1
(2k − 1) k
Theorem 1.1 with x = −1 yields the following corollary.

Corollary 1.2. We have the following identities:
∞
X 1 2 π
= log 2 − , (1.9)
1)2k 3k 5 30

k=1
(2k − k
∞
X 1 π + 3 log 2
= , (1.10)
1)2k 3k 5

k=0
(3k + k
∞
X 1 7π − 24 log 2
= . (1.11)
2)2k 3k 10

k=0
(3k + k
Theorem 1.1 with x = −2 yields the following corollary.

Corollary 1.3. We have the following identities:
∞ √
X 8k 2 3 6
3k
= π + log 3, (1.12)
k
(2k − 1)3 k 63 7
k=1
∞ √
X 8k 5 3 π + 9 log 3
= , (1.13)
k=0
(3k + 1)3k 3kk
28
∞ √
X 8k 17 3 π − 45 log 3
= . (1.14)
(3k + 2)3k 3k
k=0 k
56
Now we state our second theorem.

Theorem 1.2. We have
∞
X 1 π 2 8π − 21 log 2
3k
= 3 log2 2 − + (1.15)
(k + 1)2k k
4 5
k=0
and √
∞
X 8k 9 log2 3 − 3π 2 3 9
3k
= + π − log 3. (1.16)
k
(k + 1)3 k 16 14 7
k=0
4 ZHI-WEI SUN
Motivated by Theorem 1.1 and 1.2, we formulate the following conjecture

based on our computation.
Conjecture 1.1. For any x ∈ (−3, c) with x 6= 0, we have
∞
X x3k 3(x − 1) 2(1 − x)
= log2 (1 − x) + q(x)2
1)k 3k 2x 3 x3

k=0
(k + 1)(x − k (1.17)
t(x)q(x)
+ s(x) log(1 − x) + p ,
(1 − x)(3 + x)
where s(x) and t(x) are suitable rational functions in x.
Remark 1.1. To illustrate Conjecture 1.1, for x = 1/2, 2/3 we find the
identities:
∞ √ √ √
X 1 33 2 2 7 41 7 7
3k
= log 2 − 6 log 2 + 8 arctan − arctan
k
(k + 1)(−4) k 4 5 14 5
k=0
(1.18)
and
∞ √ √
X (−8/9)k 18 27 2 9 2 11 51 √ 11
3k
= log 3 − log 3 + arctan − 11 arctan .
(k + 1) k
5 16 4 4 55 4
k=0
(1.19)
We also have the following conjecture.
Conjecture 1.2. If −3 < x < c with c given by (1.2), then
∞
X x3k 2 1
= q(x)2 − log2 (1 − x). (1.20)
k=1
k2 (x − 1)k 3k
k
3 2
Recall that the harmonic numbers are given by
X 1
Hn = (n = 0, 1, 2, . . .).
k
0<k6n
We also establish the following theorem.


∞
X H3k − Hk π2 π 2 log 2
3k
= − − (G + log2 2) + , (1.21)
36 18 15 3

k
(2k − 1)2 k
k=1
∞
X H2k − Hk (π − 2 log 2)(π − 6 log 2)
=− , (1.22)
1)2k 3k 80

k=1
(3k + k
∞
X H3k+1 − Hk π2 4 log2 2
3k
= + G− . (1.23)
(3k + 1)2k 24 5 5
k=0 k
∞
X H2k − Hk 3 π
log2 2 − 9 log 2 + (π + 7 log 2 + 30),
= (1.24)
2)2k 3k 5 20

k=1
(3k + k
∞ √ √
X (H2k − Hk )8k ( 3 π − log 3)( 3 π − 9 log 3)
=− , (1.25)
k=0
(3k + 1)3k 3k
k
112
∞ √
X (H2k − Hk )8k 3√ 27 15π 2 + 34 3 π log 3 + 45 log 2 3
= 3π − log 3 + ,
k=0
(3k + 2)3k 3k
k
8 8 224
(1.26)
Also,
∞ √ √
X H2k − Hk 3 2 5 √ 7 1 2 7
3k
= log 2 − (log 2) 7 arctan + arctan
(3k + 1)(−4)k k 16 28 5 4 5
k=0
(1.27)
and
∞ √
X H2k − Hk √ 7 15
3k = 9 log 2 − 6 7 arctan + log2 2
(3k + 2)(−4)k 5 16
k=0 k
√ √
23 √ 7 5 2 7
+ (log 2) 7 arctan + arctan .
28 5 4 5
(1.28)
The author [20] evaluated the series
∞ ∞
X xk0 X kxk0
4k
and 4k

k=0 2k k=0 2k
for any x0 ∈ (−16, 16). Clearly, if 0 < x0 < 16 then x0 = 1/x2 for√some
√ < x0 < 0 then x0 = 4/(x(1 − x)) for some x > (1 + 2)/2
x > 1/4, if −16
or x < (1 − 2)/2.
For x > 1 or x < 0, as in [20] we define
√ √
√ 1  x log √x+1 if x > 1,
2 x−1
R(x) := x arctanh √ = p 1 (1.29)
x  |x| arctan √ if x < 0,
|x|
where arctanh t is the inverse hyperbolic tangent function.

Now we state our fourth theorem.
6 ZHI-WEI SUN
Theorem 1.4. (i) Let x > 1/4. Then

∞ √
X 2(8x2 + 6x + 1)k − 4x2 − 6x − 1 24x2 arccot 4x − 1 2x + 1
= +
(2k − 1)x2k 4k (4x − 1)3/2 4x − 1

k=1 2k
(1.30)
and
∞
X 2(8x2 − 6x + 1)k − 4x2 + 6x − 1
(2k − 1)x2k 4k

k=1 2k (1.31)
√
12x2 4x + 1 + 1 2x − 1
= log √ + .
(4x + 1)3/2 4x + 1 − 1 4x + 1
Also,
∞ √
8x2

X 2(4x + 1)k + 2x + 1 arccot 4x − 1
= √ +1 , (1.32)
(4k + 1)x 2k 4k 4x − 1 4x − 1
k=0 2k
∞ √
X 2(4x − 1)k + 2x − 1 8x2 4x2 4x + 1 + 1
4k
= − 3/2
log √ , (1.33)
(4k + 1)x 2k
2k
4x + 1 (4x + 1) 4x + 1 − 1
k=0
∞
X 2(4x + 1)(8x + 1)k + 48x2 + 18x + 1
2k 4k

k=0
(4k + 3)x 2k (1.34)
2 8x
+1 24x2 √
= 8x + 3/2
arccot 4x − 1,
4x − 1 (4x − 1)
and
∞
X 2(4x − 1)(8x − 1)k + 48x2 − 18x + 1
(4k + 3)x2k 4k

k=0 2k
(1.35)
√
2 8x − 1 12x2 4x + 1 + 1
= 8x + 3/2
log √ .
4x + 1 (4x + 1) 4x + 1 − 1
√ √
(ii) If x > (1 + 2)/2 or x < (1 − 2)/2, then
∞
X (2(2x − 1)2 k − 2x2 − x + 2)4k R(x)
4k
= 3(1 − x) − 1, (1.36)
k k
(2k − 1)x (1 − x) 2k x
k=1
∞
X (2(2x − 1)2 k + 2x2 − 3x + 2)4k
= (1 − x)(R(x) − 2x) (1.37)
(4k + 1)(x(1 − x))k 4k

k=0 2k
and
∞
X (2(2x − 1)2 (4x − 3)k + 24x3 − 42x2 + 25x − 6)4k
k 4k

k=0
(4k + 3)(x(1 − x)) 2k (1.38)
= (1 − x)(3R(x) − 2x(4x − 3)).
√ √
Remark
√ 1.2. When x > (1√+ 2)/2 or x < (1 − 2)/2, we have 1 − x <
(1 − 2)/2 or 1 − x > (1 + 2)/2, thus all the three identities in Theorem
1.4(ii) with x replaced by 1 − x also hold. In view of this, Theorem 1.4

essentially determines the values of the series
∞ ∞ ∞
X kr xk0 X kr xk0 X kr xk0
4k
, 4k
and 4k
k=1
(2k − 1) 2k k=1
(4k + 1) 2k k=1
(4k + 3) 2k
for any r ∈ {0, 1} and x0 ∈ (−16, 16).

Corollary 1.4. We have
∞ √ √ √
X 1 3 5 1+ 5 3
4k
= log − π, (1.39)
(2k − 1) 2k
10 2 36
k=1
∞ √ √ √
X 1 2 5 1+ 5 3
4k
= log + π, (1.40)
(4k + 1) 2k
5 2 9
k=0
∞ √ √ √
X 1 7 3 18 5 1+ 5
= π− log . (1.41)
k=0
(4k + 3) 4k
2k
9 5 2
Proof. Putting x = 1 in (1.30) and(1.31), we get the following two identities:

∞
X 30k − 11 4π
4k
= √ + 1, (1.42)
k=1
(2k − 1) 2k
3 3
∞ √
X 6k + 1 24 1+ 5 1
4k
= √ log + . (1.43)
(2k − 1) 2k
5 5 2 5
k=1
Note that 5 × (1.43) − (1.30) yields the first identity in Corollary 1.4. Simi-
larly, the second and the third identities in Corollary 1.4 follow from the last
four identities in Theorem 1.4(i).
The following corollary provides fast converging series for log n with 1 <
n 6 85/4.
Corollary 1.5. For
√ q
2)/2 + 1(1 +
1<n< q √ = 21.2666866 . . . ,
(1 + 2)/2 − 1
we have the following formulas for log n:
∞
X (2(n2 + 6n + 1)2 k − n4 − 16n3 + 2n2 − 16n − 1)(n − 1)4k
k (n + 1)2k 4k

(2k − 1)(−n) 2k
k=1 (1.44)
6n(n − 1)3
= −(n − 1)4 − log n.
n+1
∞
X (2(n2 + 6n + 1)2 k + n4 + 30n2 + 1)(n − 1)4k
(4k + 1)(−n)k (n + 1)2k 4k

k=0 2k (1.45)
= 8n(n + 1)2 − 2n(n2 − 1) log n,
8 ZHI-WEI SUN
and
∞
X (2(n2 + 6n + 1)2 (n2 + 14n + 1)k + cn )(n − 1)4k
(4k + 3)(−n)k (n + 1)2k 4k

k=0 2k (1.46)
2 2 3
= 8n(n + 1) (n + 14n + 1) − 6n(n + 1)(n − 1) log n,
where
cn = n6 + 46n5 + 287n4 + 868n3 + 287n2 + 46n + 1.
√
Proof. Let x = (n + 1)2 /(n − 1)2 . Then x > (1 + 2)/2 > 1,
√ √
x(1 − x) n(n + 1)2 x x+1 n+1
=− 4
and R(x) = log √ = log n.
4 (n − 1) 2 x−1 2(n − 1)
Applying the identities (1.36), (1.37) and (1.38), we get the desired identities
(1.44), (1.45) and (1.46) respectively.
Similar to Theorem 1.2, we have the following result.
∞ √ √
X 4k 20π − 3π 2 7 3+1 3 2 3+1
4k
= − √ log √ + log √ . (1.47)
(k + 1) 2k 8 3 3−1 2 3−1
k=0
Similar to Conjecture 1.1, we formulate the following conjecture.

Conjecture 1.3. (i) For any x > 14 , we have
∞ √
X 1 arccot 4x − 1
= a(x) √ + b(x)R(4x + 1)
(k + 1)x 2k 4k 4x − 1
k=0 2k
√
+ c(x)arccot 2 4x + 1 + d(x)R(4x + 1)2 ,
where a(x), b(x), c(x),
√ d(x) are suitable
√ rational functions in x.
(ii) If x > (1 + 2)/2 or x < (1 − 2)/2, then
∞
X 4k
4k
k=0
(k + 1)(x(1 − x))k 2k
= a(x)R(x) + b(x)R(1 − x) + c(x)R(x)2 + d(x)R(1 − x)2 ,
where a(x), b(x), c(x), d(x) are suitable rational functions in x.
Conjectures 1.1-1.2 and part (i) of Conjecture 1.3 will be confirmed in a
forthcoming paper joint with Yajun Zhou.
If x > 1/4, then
∞ √
X 1 2
√ 2 4x + 1 + 1
4k
= 4arccot 4x − 1 − log √ ,
k=1
2 2k
k x 2k 4x + 1 − 1
which can be deduced by noting that
∞ ∞
1X 1 X 1 + (−1)k
4k
= .
2 2
k x 2k k 2 xk 2k
k=1 2k k=1 k
√ √
Conjecture 1.4. If x > (1 + 2)/2 or x < (1 − 2)/2, then
∞
4k R(x)2 R(1 − x)2
X
= −2 + . (1.48)
k=1
k2 (x(1 − x))k 4k
2k
x 1−x
Now we state our last theorem.

Theorem 1.6. We have the following identities:
∞
X 153965k2 − 96459k + 22786 64
4k
= (1634 − 45 log 2). (1.49)
5

k=0
(−8)k k
∞
X 26975k2 − 17111k + 2968
= −297 − 120 log 2, (1.50)
k 4k

k=1
k(−8) k
∞
X 524975k2 + 195959k − 32986
= −64(592 + 45 log 2), (1.51)
(2k + 1)(−8)k 4k

k=0 k
∞
X 18675k2 + 7627k + 670
= 64(10 − 3 log 2), (1.52)
k=0
(4k + 1)(−8)k 4k
k
∞
X 1819775k2 + 669431k − 521898
= −64(2818 + 45 log 2). (1.53)
(4k + 3)(−8)k 4k

k=0 k
We are going to prove Theorems 1.1-1.3 and Theorems 1.4-1.6 in Sections

2 and 3 respectively.
The classical Ramanujan-type series for 1/π has the form:
∞ 2k
√
k ck d
X
(ak + b) k
= ,
m π
k=0
where a, b, m are integers with am 6= 0, d ∈ Z+ = {1, 2, 3, . . .} and ck takes

one of the forms
2
2k 2k 3k 2k 4k 3k 6k
, , , .
k k k k 2k k 3k
For an excellent introduction to such series, one may consult Cooper [9,
Chapter 14].
In 2022, the author [21, Conjecture 3.12] guessed that each Ramanujan-
type series for 1/π has a variant with summands involving harmonic num-
bers. For example, motivated by the Ramanujan series
∞ 2k 3
X 16
(42k + 5) k k = (1.54)
4096 π
k=0
(cf. [16]), the author [21, Remark 3.19] conjectured that

∞ 2k 3
X
k 32 log 2
((42k + 5)(H2k − Hk ) + 7) = ,
4096k π
k=0
10 ZHI-WEI SUN
which was later confirmed in [23].

The generalized harmonic numbers are given by
X 1
Hn(m) = (m ∈ Z+ and n ∈ N).
km
0<k6n
Inspired by (1.54) and the work of Guo and Lian [12], Wei [24] and Wei and
Ruan [27], the author [21] posed many conjectures on series with summands
involving generalized harmonic numbers, and this has stimulated some fur-
ther research. For example, the author [21, (3.67)] conjectured the identity
∞ 2k3
X (2) 25 (2) 2π
(42k + 5) k k H2k − H = , (1.55)
4096 92 k 69
k=0
and this was later proved by Wei [25]. To obtain (1.55), Wei [25] deduced
the following two identities ([25, (2.5) and (2.7)]):
∞ 2k 3
X
k

(2) (2)
9 8
(42k + 5) 7H2k − 2Hk − =− π (1.56)
4096k 2k + 1 3
k=0
and
∞ 2k 3
X
k

(2) (2)
8 8
(42k + 5) 4H2k − Hk + = π (1.57)
4096k 2k + 1 3
k=0
Note that 7×(1.57)−4×(1.56) yields the identity

∞ 2k 3
X
k (2) 92 88
(42k + 5)Hk + = π. (1.58)
4096k 2k + 1 3
k=0
Motivated by (1.58), we make the following conjecture on congruences.
Conjecture 1.5. Let p > 3 be a prime. Then

(p−3)/2 2k 3
X (2) 11292 −1
k
(42k + 5)Hk + pBp−3 (mod p2 ),
≡
4096k 2k + 1
3 p
k=0
(1.59)
where ( p· ) denotes the Legendre symbol. Furthermore
(p−1)/2 2k 3
X
k (2) 92
p (42k + 5)Hk +
4096k 2k + 1
k=0 (1.60)
p5

−1
≡ 23 4 + Bp−5 (mod p6 ),
p 20
where B0 , B1 , B2 , . . . are the well known Bernoulli numbers.
In order to prove Conjecture 3.14 of the author [21], C. Li and W. Chu

[14, (11)] deduced the identity
∞ 2k 3

X
k (2) 8 8π
(6k + 1)Hk + = . (1.61)
256k 2k + 1 3
k=0
We conjecture that
(p−3)/2 2k 3
X
k (2) 8 −1
(6k + 1)Hk + ≡− 16qp (2) (mod p),
256k 2k + 1 p
k=0
(1.62)
where qp (a) denotes the Fermat quotient (ap−1 − 1)/p.
With the above background in mind, in Sections 4-6 we shall pose many
new series involving binomial coefficients and generalized harmonic numbers.
2. Proofs of Theorems 1.1-1.3

Lemma 2.1. If −3 < x < 1, then
x+1 x−1
arctan p + arctan p = q(x), (2.1)
(1 − x)(3 + x) (1 − x)(3 + x)
where q(x) is defined by (1.4).
This lemma is known, see [20, Lemma 3.1].
Proof of (1.6). For each k ∈ N, we have
1 k!(2k)! Γ(k + 1)Γ(2k + 1)

3k
= =
(3k + 1) (3k + 1)! Γ(3k + 2)
k (2.2)
Z 1
= B(k + 1, 2k + 1) = x2k (1 − x)k dx.
0
Thus
∞
X x3k
(3k + 1)(x − 1)k 3k

k=0 k
∞ Z 1 ∞ 3 2
Z 1X k
X x3k 2k k x t (1 − t)
= t (1 − t) dt = dt
(x − 1)k 0 0 x−1
k=0 k=0
and hence
∞ 1
x3k dt
X Z
3k
= . (2.3)
(3k + 1)(x − 1)k k 0 1 − x3 t2 (1 − t)/(x − 1)
k=0
12 ZHI-WEI SUN
Since
2x − 3 1
+
x3 t2 (t − 1) + x − 1 tx − 1
(t − 1)x + 2
= 2 2
t x + tx(1 − x) − x + 1
1 2tx2 + x(1 − x) 2(3 − x)
= · 2 2 + ,
2x t x + tx(1 − x) − x + 1 (2tx − x + 1)2 + (1 − x)(x + 3)
we have
1 1
dt dt
Z Z
(2x − 3) 3 2
+
0 x t (t − 1) + x − 1
0 tx − 1
Z 1 2
1 2tx + x(1 − x)
= 2 2
dt
2x 0 t x + tx(1 − x) − x + 1
Z 1
2(3 − x) dt
+ p
(1 − x)(x + 3) 0 ((2tx − x + 1)/ (1 − x)(x + 3))2 + 1
and hence
1 1
dt log |xt − 1|
Z
(2x − 3) +
0 x3 t2 (t − 1) + x − 1 x t=0
1
1
= log(t2 x2 + tx(1 − x) − x + 1)
2x t=0
1
p
2(3 − x) (1 − x)(x + 3) 2tx − x + 1
+ · arctan p
(1 − x)(x + 3) 2x (1 − x)(x + 3) t=0
1 (3 − x)Q(x)
= (log 1 − log(1 − x)) + p ,
2x x (1 − x)(x + 3)
where
x+1 1−x
Q(x) := arctan p − arctan p = q(x)
(1 − x)(x + 3) (1 − x)(x + 3)
by Lemma 2.1. Therefore

Z 1
dt 3 (3 − x)q(x)
(2x−3) 3 t2 (t − 1) + x − 1
= − log(1−x)+ p . (2.4)
0 x 2x x (1 − x)(x + 3)
Combining (2.3) with (2.4), we obtain

∞
!
X x3k x−1 3 log(1 − x) (3 − x)q(x)
3k
= − + p ,
k
(3k + 1)(x − 1) k 2x −3 2x x (1 − x)(x + 3)
k=0
which is equivalent to the desired (1.6).

Proof of (1.9). As (2.2) holds for any positive integer k, we have
∞
X x3k
(2k − 1)(3k + 1)(x − 1)k 3k

k=1 k
∞ Z 1 Z 1
X x3k 2k k
= t (1 − t) dt = S(x0 , t)dt,
(2k − 1)(x − 1)k 0 0
k=1
where x0 = x3 /(x − 1) and
∞ ∞
(t2 (1 − t)x0 )k (1 − t)x0 )2k−1
p
X p X (t
S(x0 , t) = = t (1 − t)x0 .
2k − 1 2k − 1
k=1 k=1
If x0 > 0, then
p p
S(x0 , t) = t (1 − t)x0 arctanh(t (1 − t)x0 ),
where arctanh t is the inverse hyperbolic tangent function. If x0 < 0, then
∞
(−1)k−1 (t (t − 1)x0 )2k−1
p
p X
S(x0 , t) = −t (t − 1)x0
2k − 1
k=1
p p
= −t (t − 1)x0 arctan(t (t − 1)x0 ).
For the function

2
f (t) = − (3t + 2)(1 − t)3/2
15
with 0 < x < 1, it is easy to verify that
√
f ′ (t) = t 1 − t.
If x0 > 0, then
1 1
√
Z Z
f ′ (t) arctanh(t
p
S(x0 , t)dt = x0 (1 − t)x0 )dt
0 0
1
√ p
= x0 f (t) arctanh(t (1 − t)x0 )
t=0
Z 1 √ ′ √
√ (t 1 − t) x0
− x0 f (t) dt
0 1 − t2 (1 − t)x0
Z 1 √
f (t)(t 1 − t)′
= −x0 2
dt.
0 1 − t (1 − t)x0
14 ZHI-WEI SUN
Similarly, if x0 < 0 then

Z 1 1
√
Z
f ′ (t) arctan(t
p
S(x0 , t)dt = − −x0 (t − 1)x0 )dt
0 0
1
√ p
= − −x0 f (t) arctan(t (t − 1)x0 )
t=0
Z 1 √ ′ √
√ (t 1 − t) −x0
+ −x0 f (t) dt
0 1 + t2 (t − 1)x0
Z 1 √
f (t)(t 1 − t)′
= −x0 2
dt.
0 1 − t (1 − t)x0
Note that
√ √ (1 − t)(9t2 − 4)

′ 2 3/2 t
f (t)(t 1 − t) = − (3t+2)(1−t) 1−t− √ = .
15 2 1−t 15
Therefore
1
x0 1 (1 − t)(4 − 9t2 )
Z Z
S(x0 , t) = dt
0 15 0 1 − t2 (1 − t)x0
x3 1 (1 − t)(4 − 9t2 )
Z
= dt.
15 0 (tx − 1)(t2 x2 + tx(1 − x) − x + 1)
Note that
x3 (1 − t)(4 − 9t2 )
(tx − 1)(t2 x2 + tx(1 − x) − x + 1)
3 − x − 2x2 2x2 + x − 3 2tx2 + x(1 − x)
= 9+ + · 2 2
tx − 1 2x t x + tx(1 − x) − x + 1
3 2
2(2x + 3x − 6x + 9)
− .
(2tx + 1 − x)2 + (1 − x)(x + 3)
Thus
Z 1 Z 1
15 S(x0 , t)dt − 9dt
0 0
!
log |tx − 1| 1 1 1
= (2x2 + x − 3) − + log(t2 x2 + tx(1 − x) − x + 1)
x t=0 2x t=0
3 2 Z 1
2(2x + 3x − 6x + 9) dt
− p
(1 − x)(x + 3) 2
0 ((2tx + 1 − x)/ (1 − x)(x + 3)) + 1

log(1 − x) log(1 − x)
= (2x2 + x − 3) − −
x 2x
1
2(2x3 + 3x2 − 6x + 9) 2tx − x + 1
− p arctan p
2x (1 − x)(x + 3) (1 − x)(x + 3) t=0
3(1 − x)(2x + 3) − 6x + 9 2x3 + 3x2
= log(1 − x) − p q(x)
2x x (1 − x)(x + 3)
with the aid of Lemma 2.1.

Note that
5 2 3
= − .
(2k − 1)(3k + 1) 2k − 1 3k + 1
By the above two paragraphs,
∞
x3k

X 2 3
− 3k
2k − 1 3k + 1 (x − 1)k k
k=1
∞ 1
x3k
X Z
=5 3k
=5 S(x0 , t)dt
k=1
(2k − 1)(3k + 1)(x − 1)k k 0
(1 − x)(2x + 3) 2x3 + 3x2 − 6x + 9

= 3+ log(1 − x) − p q(x)
2x 3x (1 − x)(x + 3)
Combining this with (1.6), we obtain
∞
X x3k
2
(2k − 1)(x − 1)k 3k

k=1 k

(1 − x) log(1 − x) 9
= + 2x + 3
2x 2x − 3

q(x) 9(x − 3)(1 − x) 3 2
+ p − 2x − 3x + 6x − 9
3x (1 − x)(x + 3) 2x + 3
2x(1 − x) log(1 − x) 4x(x2 − 3)
= − p
2x − 3 3(2x − 3) (1 − x)(x + 3)
and hence (1.5) follows.
Proof of (1.7). Set x0 = x3 /(x − 1).

Observe that
∞ ∞ ∞
X xk0 X 3xk+1
0
X xk0
3 3k
= 3k+3
= x 0 3k+2

k=1 k k=0 k+1 k=0 k
∞
X (2k + 1)(2k + 2)xk0
= x0
(3k + 1)(3k + 2) 3k

k=0 k
∞ k
2x0 X 2 1 x0
= 2+ + 3k
.
9 3k + 1 3k + 2 k
k=0
Therefore
∞ ∞ ∞
xk0 xk0 xk0
X
X X 27 27
3k
+2 3k
= −2 3k
− .
(3k + 2) k (3k + 1) k
2x0 k
2x0
k=0 k=0 k=0
Combining this with (1.3) and (1.6), we obtain the desired (1.7).
16 ZHI-WEI SUN
Proof of Theorem 1.2. Let −3 < x < c with x 6= 0. In view of (2.2), we have
∞ ∞ Z 1
X x3k X x3k
= t2k (1 − t)k dt
(k + 1)(3k + 1)(x − 1)k 3k (k + 1)(x − 1)k 0
k=0 k k=0
Z 1X ∞
(t2 (1 − t)x3 /(x − 1))k
= dt
0 k=0 k+1
Z 1
x3 t2 (1 − t)

x−1
=− 3 2
log 1 − dt
0 x t (1 − t) x−1
1 − x 1 log(1 + t2 (1 − t)x3 /(1 − x))
Z
= 3 dt.
x 0 t2 (1 − t)
and hence
∞
x3k

X 1 3
−
k + 1 3k + 1 (x − 1)k 3k

k=0 k (2.5)
2(x − 1) 1 log(1 + t2 (1 − t)x3 /(1 − x))
Z
= dt.
x3 0 t2 (1 − t)
Putting x = −1 in (2.5), we get
∞ ∞ Z 1
X 1 X 1 log(1 − t2 (1 − t)/2)
− 3 = 4 dt.
(k + 1)2k 3k (3k + 1)2k 3k t2 (1 − t)

k=0 k k=0 k 0
Using Mathematica 13 we find that

12 log 2 2 − 24 log 2 + 4π − π 2
Z 1
log(1 + t2 (t − 1)/2)
dt = .
0 t2 (1 − t) 16
(See also [15] for a manual proof of this identity.) Combining the last two
formulas with (1.10), we immediately get the desired (1.15).
Taking x = −2 in (2.5), we obtain
∞ ∞
8k 8k 3 1 log(1 − 83 t2 (1 − t))
X X Z
−3 = dt.
k=0
(k + 1)3k 3kk k=0
(3k + 1)3k 3k
k
4 0 t2 (1 − t)
Using Mathematica 13 we find that
log(1 + 38 t2 (t − 1))
Z 1
dt
0 t2 (1 − t)
π 11 log2 3 5 log2 2
= √ − π 2 − 3 log 3 + + − (log 2)(log 3)
3 36 4 2
√ √
1 1 2 3 − −3 3 + −3
+ Li2 − − Li2 + Li2 + Li2 ,
3 8 3 4 4
where the polylogarithm Li2 (x) is given by
∞
X xn
Li2 (x) = .
n2
n=1
By using Au’s Mathematica Package MultipleZetaValues available from

[4], one can simplify
√ √
1 1 2 3 − −3 3 + −3
Li2 − − Li2 + Li2 + Li2
3 8 3 4 4
as
log2 3 π 2 5 log2 2
+ − + (log 2)(log 3).
2 18 2
Thus
log(1 + 38 t2 (t − 1)) 3 log2 3 − π 2
Z 1
π
dt = √ − 3 log 3 +
0 t2 (1 − t) 3 4
and hence
∞ ∞
X 8k X 8k
− 3
(k + 1)3k 3k (3k + 1)3k 3k

k=0 k k=0 k
3 log2 3 − π 2

3 π
= √ − 3 log 3 + .
4 3 4
Combining this with (1.13), we immediately get the desired (1.16).
In view of the above, we have completed the proof of Theorem 1.2.
Remark 2.1. If we try to prove Conjecture 1.1 by using the method in the
proof of (1.15), then we have to face lots of polylogarithms.
For a > 0 and b > 0, we have
Z 1 Z 1
∂ ∂ a−1 b−1
B(a, b) = x (1 − x) dx = xa−1 (1 − x)b−1 log xdx;
∂a 0 ∂a 0
on the other hand,
∂ ∂ Γ(a)Γ(b) ∂ log Γ(a) Γ(a)Γ(b)
B(a, b) = = Γ(b) e Γ(a+b) = (ψ(a) − ψ(a + b)),
∂a ∂a Γ(a + b) ∂a Γ(a + b)
where ψ(x) = Γ′ (x)/Γ(x) is the well known polygamma function. Therefore
Z 1
Γ(a)Γ(b)
xa−1 (1 − x)b−1 log xdx = (ψ(a) − ψ(a + b)), (2.6)
0 Γ(a + b)
for any a > 0 and b > 0.
It is well known that ψ(n + 1) = Hn − γ for all n ∈ N, where γ = 0.577 . . .
is the Euler constant. In view of (2.6), for any k ∈ N we have
Z 1
Γ(k + 1)Γ(2k + 1)
xk (1 − x)2k log xdx = (ψ(k + 1) − ψ(3k + 2))
0 Γ(3k + 2)
k!(2k)!
= (ψ(k + 1) + γ − (ψ(3k + 2) + γ))
(3k + 1)!
and hence
1
Hk − H3k+1
Z
xk (1 − x)2k log xdx = . (2.7)
0 (3k + 1) 3k
k
18 ZHI-WEI SUN
Similarly,
1
H2k − H3k+1
Z
x2k (1 − x)k log xdx = . (2.8)
0 (3k + 1) 3k
k
Note that (2.8) minus (2.7) yields the identity
1
H2k − Hk x
Z
3k
= x2k (1 − x)k log dx, (2.9)
(3k + 1) 0 1−x
k
which is similar to the identity
1
x H2n−1 − Hn
Z
2n x2n−1 (1 − x)n log dx = 3n (2.10)
1−x

0 n
(with n ∈ Z+ ) observed by K. C. Au [5].
Proof of Theorem 1.3. In view of (2.9),
∞
X H2k − Hk
k 3k

k=0
(3k + 1)2 k
∞ Z 1
X 1 x
= (x2 (1 − x))k log dx
2k 0 1−x
k=0
Z 1X ∞ 2 k Z 1
x (1 − x) x log(x/(1 − x))
= log dx = 2
dx
0 k=0 2 1−x 0 1 − x (1 − x)/2
Z 1
log(x/(1 − x)) (π − 2 log 2)(π − 6 log 2)
=2 3 2
dx = − ,
0 x −x +2 80
where the last equality can be yielded by Mathematica. Similarly, with the
aid of (2.7), we have
∞
X Hk − H3k+1
k 3k

k=0
(3k + 1)2 k
∞ Z 1 Z 1X ∞ 2 k
X 1 2 k t (1 − t)
= (x(1 − x) ) log xdx = log(1 − t)
2k 0 0 k=0 2
k=0
24 log2 2 − 5π 2 − 96G
Z 1 Z 1
log(1 − t) log(1 − t)
= 2
dt = 2 3 2
dt = ,
0 1 − t (1 − t)/2 0 t −t +2 120
where the last equality can be yielded by Mathematica. So we have proved

(1.22) and (1.23).
In light of (1.9), (1.10) and (1.22), to prove (5.28) we only need to note
that
∞
X Hk − H3k+1
(3k + 1)(2k − 1)2k 3k

k=1 k
Z 1X ∞
(x(1 − x)2 /2)k
= log xdx
0 k=1 2k − 1
Z 1 r r
x x
= (1 − x) arctanh (1 − x) log xdx
0 2 2
960G + π(25π + 136) − 24(5 log 2 2 + 4 log 2 + 63)
=
1800
where the last equality can be yielded by Mathematica.
All the remaining identities in Theorem 1.3 can be proved similarly by
combining the ideas in Section 2 with the identities (2.9) and (2.10).
3. Proofs of Theorems 1.4-1.6

Lemma 3.1 ([20]). (i) For any x > 1/4, we have
∞
8x2 √

X 2(4x + 1)k − 2x + 1 3
= √ arccot 4x − 1 − 4x + 4
x2k 4k (4x − 1)2

k=0 2k
4x − 1
(3.1)
and
∞
8x2

X 2(4x − 1)k − 2x − 1 3R(4x + 1)
= − 4x − 4 . (3.2)
x2k 4k (4x + 1)2 4x + 1

k=0 2k
√ √
(ii) If x > (1 + 2)/2 or x < (1 − 2)/2, then
∞
X (2(2x − 1)2 (2x − 3)k − (4x3 − 16x2 + 7x + 6))4k
(x(1 − x))k 4k

k=0 2k (3.3)
= (1 − x) (3R(x) + 4x(x − 3)) .
Lemma 3.2. For any integer n > 0 and complex number z 6= 0, we have
the following identities:
n
X 2(1 − 16z)k2 + (32z + 1)k − 6z − 1 n+1
= 6z + , (3.4)
k=0
k 4k
(2k − 1)z 2k z n 4n
2n
n
X 2(1 − 16z)k2 + 3k + 2z + 1 (n + 1)(2n + 1)
= 2z + , (3.5)
4k n 4n

(4k + 1)z k (4n + 1)z
k=0 2k 2n
n
X 2(1 − 16z)k + 3k + 18z + 1 (n + 1)(2n + 1)
= 6z + . (3.6)
(4k + 3)z k 4k
(4n + 3)z n 4n
k=0 2k 2n
Proof. The three identities can be easily proved by induction on n.

20 ZHI-WEI SUN
Proof of Theorem 1.4(i). We just prove (1.31),(1.32) and (1.35) as the other
three identities in Theorem 1.4(i) can be proved similarly.
Since x > 1/4, by (3.4) with z = x2 we have
n
X 2(1 − 16x2 )k2 + (32x2 + 1)k − 6x2 − 1 n+1
= 6x2 +
4k
x2n 4n

2k
(2k − 1)x 2k
k=0 2n
for any integer n > 0. Letting n → +∞, we then obtain

∞
X 2(1 − 16x2 )k2 + (32x2 + 1)k − 6x2 − 1
= 6x2 .
2k 4k

k=0
(2k − 1)x 2k
It follows that
∞
X (2(4x − 1)k − 2x − 1)
12x2 = − (4x + 1)
2k 4k

k=0
x 2k
∞
X 2(2x − 1)(4x − 1)k − 4x2 + 6x − 1
+ .
(2k − 1)x2k 4k

k=0 2k
Combining this with (3.2) and noting that

√ √
4x + 1 4x + 1 + 1
R(4x + 1) = log √ , (3.7)
2 4x + 1 − 1
we get (1.31).
By (3.5) with z = x2 , we have
n
X 2(1 − 16x2 )k2 + 3k + 2x2 + 1 (n + 1)(2n + 1)
4k
= 2x2 + 4n

(4n + 1)x2n 2n

k=0
(4k + 1)x2k 2k
for any integer n > 0, and hence

∞
X 2(1 − 16x2 )k2 + 3k + 2x2 + 1
= 2x2 .
2k 4k

k=0
(4k + 1)x 2k
It follows that
∞
2
X (2(4x + 1)k − 2x + 1)
8x = (1 − 4x)
x2k 4k

k=0 2k
∞
X 2(4x + 1)k + 2x + 1
+3 .
k=0
(4k + 1)x2k 4k
2k
Combining this with (3.1), we immediately get (1.32).

By (3.6) with z = x2 , we have
n
X 2(1 − 16x2 )k2 + 3k + 18x2 + 1 (n + 1)(2n + 1)
= 6x2 +
4k 2n 4n

(4k + 3)x 2k (4n + 3)x
k=0 2k 2n

∞
X 2(1 − 16x2 )k2 + 3k + 18x2 + 1
4k
= 6x2 .
2k
(4k + 3)x 2k
k=0
It follows that
∞
X (2(4x − 1)k − 2x − 1)
24x2 = − (4x + 1)
x2k 4k

k=0 2k
∞
X 2(4x − 1)(8x − 1)k + 48x2 − 18x + 1
+ .
(4k + 3)x2k 4k

k=0 2k
Combining this with (3.2) and noting (3.7), we immediately get (1.35).
√ √
Proof of Theorem 1.4(ii). Assume that x > (1 + 2)/2 or x < (1 − 2)/2.
By (3.4) with z = x(1 − x)/4, we have
n
X 2(1 − 4x(1 − x))k + (8x(1 − x) + 1)k − 3x(1 − x)/2 − 1
k 4k

k=0
(2k − 1)(x(1 − x)/4) 2k
3 n+1
= x(1 − x) + 4n
2 (x(1 − x)/4)n 2n
∞
X 2(1 − 4x(1 − x))k + (8x(1 − x) + 1)k − 3x(1 − x)/2 − 1 3
4k
= x(1 − x).
k
(2k − 1)(x(1 − x)/4) 2k 2
k=0
It follows that
∞
X 2(2x − 1)2 (2x − 3)k − (4x3 − 16x2 + 7x + 6)
3x(1 − x)(2x − 3) =
(x(1 − x)/4)k 4k

k=0 2k
∞
X 2(2x − 1)2 k − 2x2 − x + 2
−x .
k=0
(2k − 1)(x(1 − x)/4)k 4k
2k
∞
X 2(2x − 1)2 k − 2x2 − x + 2
x
(2k − 1)(x(1 − x)/4)k 4k

k=1 2k
= (1 − x)3R(x) − x(1 − x)(2x + 3) − x(2x2 + x − 2)
= 3(1 − x)R(x) − x,
which is equivalent to (1.36).
Now we prove (1.38). By (3.6) with z = x(1 − x)/4, we have
n
X 2(1 − 4x(1 − x))k + 3k + 9x(1 − x)/2 + 1
k 4k

k=0
(4k + 3)(x(1 − x)/4) 2k
3 (n + 1)(2n + 1)
= x(1 − x) + 4n
2

(4n + 3)(x(1 − x)/4)n 2n
22 ZHI-WEI SUN

∞
X 4(2x − 1)2 k2 + 6k + 9x(1 − x) + 2
= 3x(1 − x).
(4k + 3)(x(1 − x)/4)k 4k

k=0 2k
It follows that
∞
X 2(2x − 1)2 (2x − 3)k − (4x3 − 16x2 + 7x + 6)
6x(1 − x)(2x − 3) =
k 4k

k=0
(x(1 − x)/4) 2k
∞
X 2(2x − 1)2 (4x − 3)k + 24x3 − 42x2 + 25x − 6
− .
k 4k

k=0
(4k + 3)(x(1 − x)/4) 2k

∞
X 2(2x − 1)2 (4x − 3)k + 24x3 − 42x2 + 25x − 6
(4k + 3)(x(1 − x)/4)k 4k

k=0 2k
= (1 − x)(3R(x) + 4x(x − 3)) − 6x(1 − x)(2x − 3)
= (1 − x)(3R(x) − 2x(4x − 3)).
This proves (1.38).

The identity (1.37) can be proved similarly by using (1.37) and (3.3).
Lemma 3.3. We have

√ ! √
√ 1− 3 π2 3+1 √
Li2 ( 3 − 1) − Li2 = − 3 log log(5 + 3 3). (3.8)
2 6 2
√
Proof. Let z = ( 3 + 1)/2. Then
√
1− 3 1 √
1−z = and = 3 − 1.
2 z
It is known that
π2
Li2 (z) + Li2 (1 − z) = − (log z) log(1 − z)
6
and
π 2 log2 (−z)

1
Li2 (z) + Li2 =− −
z 6 2
(cf. [28]). Therefore
π 2 log2 (−z)

1
Li2 − Li2 (1 − z) = − − + (log z) log(1 − z).
z 3 2
Hence
√ !
√ 1− 3
Li2 ( 3 − 1) − Li2
2
√ !2 √ ! √ !
π2 1 3+1 3+1 3−1
= − − iπ + log + log iπ + log
3 2 2 2 2
√ √
π2 1 3+1 √ 3+1
= − log2 − log( 3 + 1) log
6 2 √ 2 √ 2
π2 1 3+1 ( 3 + 1)3
= − log log .
6 2 2 2
This proves (3.8).
Proof of Theorem 1.5. For any positive integer k, we have

1 (2k)!(2k − 1)! k Γ(2k)2
4k
= = ·
(k + 1) 2(k + 1)(4k − 1)! k + 1 Γ(4k)
2k
Z 1
k k
= B(2k, 2k) = x2k−1 (1 − x)2k−1 dx.
k+1 k+1 0
Thus
∞ ∞
1X
4k k k
X Z
2k
= 4 (x(1 − x))2k−1 dx
(k + 1) k 0 k=1 k+1
k=1
1 ∞
1 X k
Z
= (4x2 (1 − x)2 )k dx
0 x(1 − x) k+1
k=0
1 ∞ ∞
(4x2 (1 − x)2 )k
X
1
Z X
2 2 k
= (4x (1 − x) ) − dx
0 x(1 − x) k+1
k=0 k=0
1
log(1 − 4x2 (1 − x)2 )

1 1
Z
= + dx.
0 x(1 − x) 1 − 4x2 (1 − x)2 4x2 (1 − x)2
Let I denote the last definite integration. Mathematica 13 yields that
√ √
24I = 60π − 3π 2 + 18 arccos2 2 + 56 3 log(2 − 3)
√ √ √
− 24 + 36 log 2 ( 3 − 1) − 72 log( 3 − 1) log( 3 + 1)
√ !
2
√ √ 1− 3
+ 54 log ( 3 + 1) − 36Li2 ( 3 − 1) + 36Li2 .
2
√
Note that arccos 2 = i log(2 + 3). Thus, from the above we can reduce
the desired (1.47) to the identity (3.8) in Lemma 3.3. This concludes our
proof.
24 ZHI-WEI SUN
Lemma 3.4. For any positive integer n, we have

n
X 2075k3 − 3045k2 + 1414k − 192 3(3n + 1)(3n + 2)
= − 6,
4k
(−8)n 4n

k
k(−8) k
k=1 n
n
X 2075k3 − 970k2 − 607k + 198 3(n + 1)(3n + 1)(3n + 2)
= + 192,
4k
(2n + 1)(−8)n 4n

k=0
(2k + 1)(−8)k k n
n
X 415k3 − 194k2 − 19k + 14 3(n + 1)(3n + 1)(3n + 2)
5 = + 64,
4k n 4n

(4k + 1)(−8)k (4n + 1)(−8)
k=0 k n
n
X 2075k3 − 970k2 − 1119k + 582 3(n + 1)(3n + 1)(3n + 2)
= + 192.
4k
(4n + 3)(−8)n 4n

k
(4k + 3)(−8) k
k=0 n
Proof. The four identities in Lemma 3.4 can be easily proved by induction
on n.
Proof of Theorem 1.6. We first prove that all the five identities in Theorem
1.6 are equivalent. Note that 5 × (1.49) − 24 × (1.50) gives the identity
∞
X 5k(153965k2 − 96459k + 22786) − 24(26975k2 − 17111k + 2968)
k(−8)k 4k

k=1 k
= −5 × 22786 + 64 × 1634 − (−297) × 24,
i.e.,
∞
X 2075k3 − 3045k2 + 1414k − 192
= −6. (3.9)
k(−8)k 4k

k=1 k
Letting n → ∞ in the first identity in Lemma 3.4, we immediately get (3.9).

So (1.49) and (1.50) are indeed equivalent. Similarly, by using Lemma 3.4,
we can deduce that each of (1.51), (1.52) and (1.53) is equivalent to (1.49).
So, the five identities in Theorem 1.6 are equivalent.
Now it suffices to prove (1.52).
For |z| < 1, we clearly have
∞ ∞
X 1 X d 1
zk = , kz k−1 = (1 − z)−1 = ,
1−z dz (1 − z)2
k=0 k=1
and
∞
X d 2
k(k − 1)z k−2 = (1 − z)−2 = .
dz (1 − z)3
k=2
Set
P (k) = 18675k2 + 7627k + 670 = 18675k(k − 1) + 26302k + 670.

Then, for |z| < 1, we have
∞
X ∞
X ∞
X ∞
X
P (k)z k = 18675z 2 k(k − 1)z k−2 + 26302z kz k−1 + 670 zk
k=0 k=2 k=1 k=0
2z 2 z 670
= 18675 × + 26302 × +
(1 − z)3 (1 − z)2 1−z
2(5859z 2 + 12481z + 335)
= .
(1 − z)3
In view of the above, we have
∞
X P (k)
4k
k=0
(4k + 1)(−8)k k
∞ ∞
X P (k)Γ(k + 1)Γ(3k + 1) X P( k)
= = B(k + 1, 3k + 1)
(−8)k Γ(4k + 2) (−8)k
k=0 k=0
∞ ∞ k
P (k) 1 3k
Z 1X 3
x (1 − x)
X Z
k
= x (1 − x) dx = P (k) dx
(−8)k 0 0 k=0 −8
k=0
Z 1
5859x6 (1 − x)2 /64 − 12481x3 (1 − x)/8 + 335
=2 dx
0 (1 + x3 (1 − x)/8)3
Z 1
5859x6 (x − 1)2 + 99848x3 (x − 1) + 21440
= −16 dx.
0 (x3 (x − 1) − 8)3
Observe that x3 (x − 1) − 8 = (x − 2)(x3 + x2 + 2x + 4) and
5859x6 (x − 1)2 + 99848x3 (x − 1) + 21440

(x3 (x − 1) − 8)3
747 403 9 Q(x)
= 3
+ 2
− + ,
5(x − 2) 5(x − 2) x − 2 5(x + x2 + 2x + 4)3
3
where
Q(x) = 45x8 − 178x7 − 2713x6 + 12108x5 + 34656x4

+ 54960x3 − 16506x2 − 21172x − 15312.
26 ZHI-WEI SUN
Therefore
∞ 1
P (k) Q(x)
16
X Z
4k
+ dx
(4k + 1)(−8) k 0 + x2 + 2x + 4)3
5 (x3
k=0 k
16 1
Z
747 403 45
= − + − dx
5 0 (x − 2)3 (x − 2)2 x − 2
!
1 1 1
16 747 −2 403 −1
= − (x − 2) + (x − 2) − 45 log(2 − x)
5 −2 0 −1 0 0

16 747 1 1
= − 1− − 403 −1 + − 45(0 − log 2)
5 −2 4 2
1258
= − 144 log 2.
5
So we have reduced (1.52) to the identity
Z 1
Q(x) 971
3 2 3
dx = 15 log 2 − . (3.10)
0 (x + x + 2x + 4) 8
Note that
x3 + x2 + 2x + 4 = (x − α)(x2 + (α + 1)x + α2 + α + 2),
where
1 3 √ √
q q
3
α= 3 249 − 46 − 3 249 + 46 − 1 ≈ −1.48.
3
However, it is rather complicated to prove (3.10) via writing the rational
function Q(x)/(x3 + x2 + 2x + 4)3 as a sum of partial fractions. Instead, we
define f (x) as
298x5 + 1663x4 − 3190x3 − 4451x2 − 4878x − 930
15 log(x3 + x2 + 2x + 4) + .
(x3 + x2 + 2x + 4)2
It is easy to verify that
d Q(x)
f (x) = 3 .
dx (x + x2 + 2x + 4)3
Therefore
Z 1
Q(x) 11488 930
dx = f (1) − f (0) = 15 log 8 − − 15 log 4 + 2
0 (x3 2
+ x + 2x + 4)3 82 4
and hence (3.10) holds.
In view of the above, we have completed our proof of Theorem 1.6.
4. Conjectural series with summands involving two binomial

coefficients
In this section, we propose some new conjectures on series whose sum-
mands involve two binomial coefficients and generalized harmonic numbers.
Conjecture 4.1 (2023-10-15). (i) We have

∞
X 10k − 3 K
= (4.1)
1)k2 3k 2k
3k
(2k − k k
2
k=1
and
∞
X (4k − 1)3k
= 2K, (4.2)
(2k − 1)k2 2k 3k

k=1 k k
where
∞ ∞
( k3 ) X
X
−3 1 1
K := L 2, = = − .
· k2 (3j + 1)2 (3j + 2)2
k=1 j=0
Also,
∞
X (10k − 3)(3H3k−1 + 4H2k−1 − 6Hk−1 ) − 12 4π 3
= √ (4.3)
(2k − 1)k2 3k 2k 3k

k=1 k k
27 3
and
∞
X 3k ((4k − 1)(3H3k−1 − 3H2k−1 − Hk−1 ) + 6k/(2k − 1)) 8π 3
= √ . (4.4)
(2k − 1)k2 2k 3k

k=1 k k
27 3
(ii) Let p be a prime. For any n ∈ Z+ , we have

pn−1
X (10k + 3)3k 2k 3k
n−1
X (10k + 3)3k 2k
3k
1 k k p k k
− ∈ Zp
(pn)2 2k + 1 3 2k + 1
k=0 k=0
where Zp is the ring of p-adic integers. When p > 3, for any n ∈ Z+ we also
have
pn−1
X (4k + 1) 2k 3k
n−1 2k 3k

1 k k p X (4k + 1) k k
− ∈ Zp .
(pn)2 (2k + 1)3k 3 (2k + 1)3k
k=0 k=0
Moreover, provided p > 3 we have the following congruences:

p−1
(10k + 3)3k 2k
3k p
X
2 1
k k
≡3 + p Bp−2 (mod p3 ),
2k + 1 3 3
k=0
p−1
(4k + 1) 2k
3k
X p 4 2 1
k
k
k
≡ + p Bp−2 (mod p3 ),
(2k + 1)3 3 9 3
k=0
p−1 k 2k 3k
X 3 k k
p ((10k + 3)(3H3k + 4H2k − 6Hk ) + 12)
2k + 1
k=0
p 17
1
≡ 12 + p2 Bp−2 (mod p3 ),
3 2 3
28 ZHI-WEI SUN
and
p−1 2k 3k
X
k k 6k
(4k + 1)(3H3k − 3H2k − Hk ) −
(2k + 1)3k 2k + 1
k=0

2 1
≡ pBp−2 (mod p3 ).
3 3
Remark 4.1. The series in (4.1) with converging rate 1/81 seems to be the
fastest series for the constant K. Note also that
3k
k 3k 3k
= −2 ∈Z for all k ∈ N.
2k + 1 k 2k + 1
∞
(−1)k π 2 log 2 11

X 7k − 2 1
H k−1 − = − ζ(3), (4.5)
k2 2k 3k 2k − 1 6k 6 12

k=1 k k
∞
(−1)k π 2 log 2 29

X 7k − 2 5
H 2k−1 + = − ζ(3), (4.6)
k2 2k 3k 2k − 1 12k 12 24

k=1k k
∞ (2)
X (−1)k ((7k − 2)Hk−1 − (6k − 2)/(2k − 1)2 ) π4
= , (4.7)
2 2k 3k 144

k=1
(2k − 1)k k k
∞
(−1)k π2

X 7k − 2 (3) 17 45
2k 3k
H k−1 − 3
= ζ(3) − ζ(5). (4.8)
2k − 1 32k 2 8

k=1
k2 k k
(ii) Let p > 3 be an odd prime. Then
(p−1)/2
k 2k 3k 7k + 2 1
X
(−1) Hk + ≡ −4qp (2) − 4p qp (2)2 (mod p2 ),
k k 2k + 1 6k
k=1

k 2k 3k 7k + 2 1
X
(−1) Hk + ≡ 6p qp (2)2 (mod p2 ),
k k 2k + 1 6k
p/2<k<p
(p−1)/2

X
k 3k 2k
7k + 2 5
(−1) H2k − ≡ qp (2) − 2p qp (2)2 (mod p2 ),
k k
2k + 1 12k
k=1

k 2k 3k 7k + 2 5
X
(−1) H2k − ≡ −3qp (2) + 3p qp (2)2 (mod p2 ).
k k 2k + 1 12k
p/2<k<p
Also,
(p−3)/2
X (−1)k 2k 3k

(2) 6k + 2 p
k k
(7k + 2)Hk + ≡ −2− Bp−3 (mod p2 )
2k + 1 (2k + 1)2 12
k=1
and
p−1
X 2k 3k 7k + 2 (3) 17 45Hp−1
(−1)k Hk + ≡ +9p2 Bp−5 (mod p3 ).
k k 2k + 1 32k3 4p2
k=1
Remark 4.2. Chu and Zhang [7, Example 24] observed that
∞
X (−1)k (7k − 2) π2
= − .
2 2k 3k 12

k=1
(2k − 1)k k k
Conjecture 4.3 (2023-09-11). We have

∞
X (−1)k ((28k2 − 41k + 9)(2H2k−1 + 5Hk−1 ) − 21k + 3)
2k 3k

(2k − 1)k
k=1 k k (4.9)
21 3 π2
= ζ(3) − π 2 log 2 +
2 2 4
and
∞
X (−1)k (6(28k2 − 41k + 9)(H3k−1 − 6Hk−1 ) − 142k + 125)
(2k − 1)k 2k
3k
k=1 k k (4.10)
15 147
= π 2 log 2 − ζ(3).
2 4
Remark 4.3. In view of the identity in Remark 4.2, we have
∞
X (−1)k (28k2 − 41k + 9) π2 1
= − ,
(2k − 1)k 2k 3k 8 2

k=1 k k
because 2k(28k2 − 41k + 9) + 3(7k − 2) = (2k − 1)(28k 2 − 27k + 6), and

n
X (−1)k (28k2 − 27k + 6) (−1)n
= −1 + 2n 3n
2 2k 3k

k=1
k k k n n
tends to −1 as n → +∞.
∞
X (−1)k (12(21k2 − 23k + 4)(2H2k−1 + 5Hk−1 ) + 35k − 121)
2k 3k
k=1
(2k − 1) k k (4.11)
63 9 7
= ζ(3) − π 2 log 2 + π 2 − 4
2 2 4
and
∞
X (−1)k (108(21k2 − 23k + 4)(H3k−1 − 6Hk−1 ) − 3247k + 2537)
(2k − 1) 2k
3k
k=1 k k (4.12)
1323 135 2 95 95
=− ζ(3) + π log 2 − π 2 + .
8 4 16 4
Remark 4.4. In view of the first identity in Remark 4.3, we have
∞
X (−1)k (21k2 − 23k + 4) π2 1
2k 3k
= − ,
32 2

k=1
(2k − 1) k k
30 ZHI-WEI SUN
because 4k(21k2 − 23k + 4) − (28k2 − 41k + 9) = 3(2k − 1)2 (7k − 3), and
n
X (−1)k (2k − 1)(7k − 3) 1 (−1)n (n + 1)
2k 3k
=− + 2n 3n
2

k=1
k k k
2 n n
tends to −1/2 as n → +∞.

∞
(−1)k−1 56k2 − 32k + 5

X (2) (2) 7
(2H2k−1 − Hk−1 ) − 2 = 9ζ(5). (4.13)
k3 2k
3k
k k
(2k − 1)2 4k
k=1
(ii) Let p > 3 be a prime. Then

p−1
(−1)k 2k 56k2 + 32k + 5

X 3k (2) (2) 7
(2H2k − Hk ) +
k k k (2k + 1)2 4k2
k=1
Hp−1 1608 2
≡ 66 − p Bp−5 (mod p3 ).
p2 5
Provided p > 5, we have
p−1
(−1)k 2k 56k2 + 32k + 5

X 3k (2) (2) 7
(2H2k − Hk ) + 2
k k k (2k + 1)2 4k
k=1
Hp−1
≡ 18 − 28p (mod p2 ).
p2
Remark 4.5. Chu and Zhang [7, Example 21] proved that
∞
X (−1)k−1 (56k2 − 32k + 5)
= 4ζ(3).
2 k 3 2k 3k

k=1
(2k − 1) k k

∞
(−1)k−1 56k2 − 32k + 5 (3) π6

X 1
H 2k−1 + = + 3ζ(3)2 (4.14)
k3 2k 3k (2k − 1)2 k3 1512

k=1 k k
and
∞
(−1)k−1 56k2 − 32k + 5 (4)

X (4)
15
4H2k−1 − Hk−1 − 4
3 2k 3k (2k − 1)2 4k

k
k=1 k k (4.15)
62 π4
= 205ζ(7) − π 2 ζ(5) + ζ(3).
3 5
(p−3)/2
X (−1)k 2k 3k 56k2 + 32k + 5 (3) 1

H 2k − ≡ 0 (mod p)
k k k (2k + 1)2 k3
k=1
and
p−1
(−1)k 2k 56k2 + 32k + 5

X 3k (3) 1
p H2k − 3
k k k (2k + 1)2 k
k=1
Hp−1 858 2
≡6 − p Bp−5 (mod p3 ).
p2 5
Also,
(p−3)/2
X (−1)k 2k 3k 56k2 + 32k + 5 (4) (4) 15

(4H2k − Hk ) + 4
k k k (2k + 1)2 4k
k=1
≡ 36Bp−5 (mod p)
and
p−1
(−1)k 2k 56k2 + 32k + 5

2
X 3k (4) (4) 15
p (4H2k − Hk ) + 4
k k k (2k + 1)2 4k
k=1
Hp−1 3996 2
≡ 24 − p Bp−5 (mod p3 ).
p2 5
Remark 4.6. This conjecture looks quite challenging.
∞
X (−1)k−1 (28k2 − 18k + 3) π4
= (4.16)
(2k − 1)3 k4 2k 3k 45

k=1 k k
and
∞ (2)
X (−1)k ((28k2 − 18k + 3)Hk−1 − F (k)) π6
2k 3k
= , (4.17)
378

(2k − 1)3 k4
k=1 k k
where
208k4 − 240k3 + 120k2 − 30k + 3
F (k) := .
4k2 (2k − 1)2
(ii) For any prime p > 3, we have
(p−1)/2
X (−1)k−1 (28k2 − 18k + 3) Hp−1 27
p 2k 3k ≡ 12 2 + p2 Bp−5 (mod p3 )
(2k − 1)3 k4 k k p 5
k=1
and
p−1
(−1)k 2k
3k
X
2 2
p (28k + 18k + 3) 2 k k
≡ −8 + 8p2 − 24p5 Bp−5 (mod p6 ).
k (2k + 1)3
k=1
Remark 4.7. It seems that (4.16) does not follow from any identity in [7].
∞
(−16)k

X 5k − 1 1 21
2k 4k
H k−1 − = π 2 log 2 − ζ(3), (4.18)
2k − 1 8k 4

2
k k 2k
k=1
32 ZHI-WEI SUN
∞
(−16)k π2

X 5k − 1 1 35
2k 4k
H 2k−1 + = log 2 − ζ(3), (4.19)
2k − 1 12k 3 6

2
k k 2k
k=1
and
∞
(−16)k π2

X 46k − 5 77
2k 4k
(5k − 1)H4k−1 − = log 2 − ζ(3).
(2k − 1)k2 24k 6 12
k=1 k 2k
(4.20)
(ii) Let p be any odd prime. Then
(p−1)/2 2k 4k
X 5k + 1 1 3 5
k 2k
k
Hk + ≡ − qp (2) − p qp (2)2 (mod p2 )
(−16) 2k + 1 8k 2 4
k=1
and
2k 4k

X 5k + 1 1
k 2k
Hk + ≡ 2p qp (2)2 (mod p2 ).
(−16)k 2k + 1 8k
p/2<k<p
(p−1)/2 2k 4k
X 5k + 1 1 p
k 2k
k
H2k − ≡ − qp (2)2 (mod p2 )
(−16) 2k + 1 12k 3
k=1
and
2k 4k
X 5k + 1 1 5
k 2k
H2k − ≡ −qp (2) + p qp (2)2 (mod p2 ).
(−16)k 2k + 1 12k 6
p/2<k<p
Remark 4.8. Theorem 9 of [7] with a = e = 1 and b = c = d = 1/2 yields

the identity
∞
X (−16)k (5k − 1) π2
2k 4k = − .
(2k − 1)k2 k 2k 2
k=1

∞ (2)
X (−16)k ((5k − 1)Hk−1 − 4(4k − 1)/(2k − 1)2 ) π4
= (4.21)
(2k − 1)k2 2k 4k 6

k=1 k 2k
and
∞ (3)
X (−16)k ((5k − 1)Hk−1 + 12(4k − 1)/(2k − 1)3 )
2k 4k
= −4π 2 ζ(3). (4.22)
2
(2k − 1)k k 2k
k=1
(ii) For any odd prime p, we have
(p−3)/2 2k 4k
X
k 2k (2) 4(4k + 1) p
k
(5k + 1)Hk + 2
≡ − Bp−3 (mod p)
(2k + 1)(−16) (2k + 1) 6
k=0
and
(p−3)/2 2k 4k
X
k 2k (3) 12(4k + 1)
(5k + 1)Hk − ≡ −6Bp−3 (mod p).
(2k + 1)(−16)k (2k + 1)3
k=0
Remark 4.9. We have not find any identity involving the fourth harmonic
numbers similar to (4.21) and (4.22).
∞
X (−16)k ((40k2 − 52k + 7)(3H2k−1 + 2Hk−1 ) − 20k − 11)
(2k − 1)k 2k
4k
k=1 k 2k (4.23)
9
= π 2 − 9π 2 log 2 + 84ζ(3)
2
and
∞
X (−16)k ((40k2 − 52k + 7)(9H4k−1 − 3H2k−1 − 17Hk−1 ) − 298k + 233)
(2k − 1)k 2k
4k
k=1 k 2k
99 2
= π log 2 − 147ζ(3).
2
(4.24)
∞
X (40k2 − 52k + 7)(−16)k 3
2k 4k
= π 2 − 4,
2

(2k − 1)k k 2k
k=1
because k(40k2 − 52k + 7) + 3(5k − 1) = (2k − 1)2 (10k − 3), and
n
X (−16)k (2k − 1)(10k − 3) (−1)n 42n+1
= −4 + 2n 4n
k2 2k 4k

k=1 k 2k n 2n
∞
X (−16)k (7(140k2 − 94k − 15)(3H2k−1 + 2Hk−1 ) + 15(226k − 249)
2k 4k

k=1
(2k − 1) k 2k
189 2 771 2
= 882ζ(3) − π log 2 + π − 388
2 4
(4.25)
and
∞
X (−16)k (7(140k2 − 94k − 15)(2H4k−1 + 5H2k−1 ) + 48(19k − 45))
(2k − 1) 2k
4k
k=1 k 2k
= 1323ζ(3) − 63π 2 log 2 + 219π 2 − 346.
(4.26)
Remark 4.11. In view of the first identity in Remark 4.10, we have
∞
X (140k2 − 94k − 15)(−16)k 9
2k 4k
= π 2 − 20,
(2k − 1) k 2k 4
k=1
because
2k(140k2 − 94k − 15) − 3(40k2 − 52k + 7) = 7(2k − 1)(20k 2 − 12k + 3),
34 ZHI-WEI SUN
and
n
X (−16)k (20k2 − 12k + 3) (−1)n (n + 1)42n+1
2k 4k
= −4 + 2n 4n
k=1
k k 2k n 2n
∞ (2)
X (−16)k ((20k2 − 8k + 1)Hk−1 − 16k(4k − 1)/(2k − 1)2 )
= 62ζ(5),
(2k − 1)2 k3 2k
4k
k=1 k 2k
(4.27)
∞ (2)
X (−16)k ((20k2 − 8k + 1)H2k−1 − 3k(4k − 1)/(2k − 1)2 )
2 k 3 2k 4k

k=1
(2k − 1) k 2k (4.28)
31 7
= ζ(5) − π 2 ζ(3),
2 4
and
∞
X (−16)k (20k2 − 8k + 1) (2) (2)

2
16H 2k−1 − 3H k−1 = 62ζ(5) − 28π ζ(3).
2 k 3 2k 4k

k=1
(2k − 1) k 2k
(4.29)
(p−3)/2 2k 4k 2
2k (20k + 8k + 1)

(2) (2)
X
k
16H 2k − 3H k ≡ −24Bp−3 (mod p)
(2k + 1)2 k(−16)k
k=1
and
p−1 2k 4k 2
2k (20k + 8k + 1)
(2) (2)
X
k
16H2k − 3Hk ≡ −52Bp−3 (mod p).
(2k + 1)2 k(−16)k
k=1

∞
X (−16)k (20k2 − 8k + 1)
= −14ζ(3).
(2k − 1)2 k 3 2k 4k
k=1 k 2k

∞
X 16k ((6k2 − 11k + 1)(4H2k−1 − Hk−1 ) − 18k − 1) 3
2k 4k
= −24G− π 3 (4.30)
(2k − 1)k k 2k 4
k=1
and
∞
X 16k ((24k2 − 34k − 7)(4H2k−1 − Hk−1 ) − 6k − 115) 9
2k 4k
= 11 − 138G − π 3 ,
(2k − 1) k 2k 4
k=1
(4.31)
where G denotes the Catalan constant ∞ k /(2k + 1)2 .
P
k=0 (−1)
Remark 4.13. By Chu and Zhang [7, Exmaple 84],

∞
X (6k − 1)16k
= 8G.
(2k − 1)k 2 2k 4k
k=1 k 2k
This implies the identity

∞
X (6k2 − 11k + 1)16k
2k 4k
= 1 − 6G,
k=1
(2k − 1)k k 2k
because 4k(6k2 − 11k + 1) + 3(6k − 1) = (2k − 1)(12k 2 − 16k + 3), and

n
X 16k (12k2 − 16k + 3) 42n+1
= 4 − 2n 4n
k2 2k 4k

k=1 k 2k n 2n
tends to 4 as n → +∞. Similarly, we have

∞
X (24k2 − 34k − 7)16k
2k 4k
= 7 − 18G,
k=1
(2k − 1) k 2k
because k(24k2 − 34k − 7) − 3(6k2 − 11k + 1) = (2k − 1)(2k − 3)(6k − 1), and
n
X 16k (2k − 3)(6k − 1)) 42n+1 (n + 1)
= 4 − 2n 4n
k 2k 4k

k=1 k 2k n 2n
tends to 4 as n → +∞.
∞
48k 2π 3

X 6
2k 4k
6H4k−1 − 9H2k−1 + 2Hk−1 + = √ .
k(2k − 1) k 2k 2k − 1 3
k=1
(4.32)
(p−1)/2
k 2k
4k
X
k 2k 6
6H4k − 9H2k + 2Hk − ≡ 0 (mod p)
(2k + 1)48k 2k + 1
k=1
and
p−1
k 2k
4k
X 6 5p 1
k 2k
k
6H4k − 9H2k + 2Hk − ≡ Bp−2 (mod p2 ).
(2k + 1)48 2k + 1 12 3
k=1
Remark 4.14. The author’s conjectural identity

∞ ∞
X 48k 15 X ( k3 )
2k 4k
=
k(2k − 1) k 2k
2 k2
k=1 k=1
(cf. [19]) was confirmed by Au [2].

36 ZHI-WEI SUN

∞
X (−16)k ((112k2 − 32k + 3)ak − 1536k − 120(2k + 1)/(2k − 1))
3k 6k
= π3 ,
2
k (4k − 1)(4k − 3) k 3k
k=1
(4.33)
where
ak := 6H6k−1 − 3H3k−1 + 18H2k−1 − 13Hk−1 .
Remark 4.15. By [7, Example 29], we have the identity
∞
X (−16)k (112k2 − 32k + 3)
6k = −8G.
k=1
k2 (4k − 1)(4k − 3) 3k
k 3k

∞
256k−1

X (2) 328(6k − 1)
3(22k − 1)Hk−1 − = −28β(4),
(2k − 1)k 2 3k 6k (2k − 1)2
k=1 k 3k
(4.34)
∞
256k−1

X (2) 29(6k − 1)
(22k − 1)H2k−1 −
3k 6k
= −2β(4) (4.35)
2
(2k − 1)k k 3k (2k − 1)2
k=1
and
∞
X 256k−1 (22k − 1) (2) (2)

328H 2k−1 − 87H k−1 = 156β(4), (4.36)
(2k − 1)k2 3k 6k

k=1 k 3k
where β(4) = ∞ k 4
P
k=0 (−1) /(2k + 1) .
p−1
(22k + 1) 3k
6k
X −1
k 3k
≡ + 15p2 Ep−3 (mod p3 ),
(2k + 1)256k p
k=0
(p−1)/2
X (22k + 1) 3k 6k

−1 3 5
k 3k
≡ 1 + p B p−3 (mod p6 ),
(2k + 1)256k p 80
k=0
(p−3)/2 3k
6k

X
k 3k (2) 328(6k + 1)
3(22k + 1)Hk + ≡ 192Ep−3 (mod p),
(2k + 1)256k (2k + 1)2
k=0
(p−3)/2 3k 6k
X
k 3k (2) 29(6k + 1)
(22k + 1)H2k + ≡ 16Ep−3 (mod p),
(2k + 1)256k (2k + 1)2
k=0
and
p−1 3k 6k
X (22k + 1) k 3k

(2) (2)

328H2k − 87Hk ≡ 4600Ep−3 (mod p).
(2k + 1)256k
k=0
∞
X 256k (22k − 1)
6k = 128G.
k=1
(2k − 1)k2 3k
k 3k

∞
X 256k ((22k2 − 65k − 48)H(k) + 258k + 3970)
3k 6k
= 20(3π 3 + 1216G) (4.37)
k=1
(2k − 1)k k 3k
and
∞
X 256k (24(1408k2 − 6318k + 3557)H(k) + 145906k + 533701)
(2k − 1) 3k
6k
k=1 k 3k (4.38)
3
= 16(159030G − 450π − 11381),
where
H(k) := 10H6k−1 − 5H3k−1 − 16H2k−1 + 5Hk−1 .
Remark 4.17. Based on the identity in Remark 4.16, we have
∞
X 256k (22k2 − 65k − 48)
6k = 16(1 − 30G)
k=1
(2k − 1)k 3kk 3k
and
∞
X 256k (1408k2 − 6318k + 3557)
= 16(59 + 150G)
(2k − 1) 3k
6k
k=1 k 3k
because of the following identities:
4k(22k2 − 65k − 48) + 15(22k − 1) = (2k − 1)(44k 2 − 108k + 15),
n
X 256k (44k2 − 108k + 15) 44n+3
= 64 − 3n 6n ,
2 3k 6k

k=1
k k 3k n 3n
k(1408k2 − 6318k + 3557) + 5(22k2 − 65k − 48)
= 16(2k − 1)(44k2 − 172k + 15),
n
X 256k (44k2 − 172k + 15) 44n+3 (n + 1)
= 64 − 3n 6n .
k 3k 6k

k=1 k 3k n 3n
2k 3k 6k
5. Conjectural series with summands containing k /( k 3k )
In 2013, the author [18] proved that

2k 3k 6k
2(2k + 1) for all k ∈ N.
k k 3k
In this section we pose some conjectures on series whose summands involve
2k 3k 6k
k /( k 3k ) (k ∈ N).
38 ZHI-WEI SUN

∞
(−4)k 2k

k ((7k − 1)Hk−1 − (6k − 1)/(4k − 2))
X
= π log 2 − 2G, (5.1)
k(2k − 1) 3k
6k
k=1 k 3k
∞
(−4)k 2k

k ((7k − 1)H2k−1 − 9(6k − 1)/(8k − 4)) 3
X
3k 6k
= π log 2 − G, (5.2)
4

k=1
k(2k − 1) k 3k
and
∞ 2k
(−4)k

X
k 34k − 9
3k 6k
(7k − 1)(2H6k−1 − H3k−1 ) −
4k − 2

k=1
k(2k − 1) k 3k (5.3)
π
= log 2 − 2G.
2
(ii) Let p be an odd prime. Then
(p−3)/2 3k 6k
3k ((7k + 1)Hk + (6k + 1)/(4k + 2))

X
k −1
≡ qp (2) (mod p)
(2k + 1)(−4)k 2k p

k=0 k
and
(p−3)/2 3k 6k
3k ((7k + 1)H2k + 9(6k + 1)/(8k + 4))
X
k
(2k + 1)(−4)k 2k

k=0 k

−1 3
qp (2) − p qp (2)2 (mod p2 ).

≡
p 4
(iii) Let p be any odd prime. Then
(p−1)/2 3k 6k 7( 7p ) + ( −1
p )

X
k 3k
2k
≡ (mod p),
(2k + 1)(−4)k k
8
k=0
p−1 3k 6k 7( 7p ) − 5( −1
p )

X
k 3k
≡
(mod p),
2k
(2k + 1)(−4)k k 2
k=0
(p−1)/2
X (7k + 1) 3k 6k

−1 3
k 3k
≡ 1 + p qp (2) (mod p3 ),
(2k + 1)(−4) k 2k p 2
k=0 k
p−1
(7k + 1) 3k
6k
X −1
k 3k
2k
≡ − 15p2 Ep−3 (mod p3 ).
(2k + 1)(−4) k p
k=0 k
∞
(7k − 1)(−4)k 2k

X
k π
=− .
3k 6k
(2k − 1)k k 3k 4
k=1
Suitable linear combinations of the three identities in Conjecture 5.1(i) yield

the identities
∞
(−4)k 2k

k ((7k − 1)(30H6k−1 − 15H3k−1 − 14H2k−1 + 3Hk−1 ) − 75/2)
X
(2k − 1)k 3k
6k
k=1 k 3k
= −22G
(5.4)
and
∞ 2k
X (−4)k k ((7k − 1)(16H6k−1 − 8H3k−1 − 6H2k−1 − 5Hk−1 ) − 20)
(2k − 1)k 3k
6k
k=1 k 3k
11
=− π log 2.
2
(5.5)
∞ (2) (2)
(−4)k 2k 2

X
k ((7k − 1)(6H2k−1 − Hk−1 ) − 9(6k − 1)/(2k − 1) ) π3
=
(2k − 1)k 3k 6k 24

k=1 k 3k
(5.6)
and
∞ (3) (3)
(−4)k 2k 3

k ((7k − 1)(8H2k−1 − Hk−1 ) − 12(6k − 1)/(2k − 1) )
X
(2k − 1)k 3k 6k

k=1 k 3k (5.7)
9 48
= πζ(3) − β(4).
4 7
(p−3)/2 3k 6k
X
k 3k (2) (2) 9(6k + 1)
(7k + 1)(6H2k − Hk ) +
(2k + 1)(−4) k 2k (2k + 1)2
k=0 k
≡ Ep−3 (mod p).
For each odd prime p 6= 7, we have
(p−3)/2 3k 6k

X
k 3k (3) (3) 12(6k + 1)
(7k + 1)(8H2k − Hk ) +
(2k + 1)(−4) k 2k
k
(2k + 1)3
k=0

−1 3
≡ Bp−3 (mod p).
p 4
Remark 5.2. Conjecture 5.2(i) looks challenging.
∞
(−4)k 2k

k ((280k − 51)H(k) − 1352)
X
3k 6k
= 18π − 624G (5.8)
k=1
k k 3k
40 ZHI-WEI SUN
and
∞ 2k
(−4)k

X
k (17(952k − 201)H(k) − 50924)
3k 6k
= −452 − 1587π − 26520G.
k=1 k 3k
(5.9)
where
H(k) := 30H6k−1 − 15H3k−1 − 22H2k−1 + 9Hk−1 .
∞
(280k − 51)(−4)k 2k

X
k
3k 6k
= −6π − 10,
k=1
k k 3k
because (2k − 1)(280k − 51) − 24(7k − 1) = 5(112k 2 − 110k + 15), and

n
(112k2 − 110k + 15)(−4)k 2k (−1)n 22n+1 2n

X
k n
= −2 + 3n 6n
(2k − 1)k 3k
6k
k=1 k 3k n 3n
tends to −2 as n → +∞. Similarly, we have

∞
(952k − 201)(−4)k 2k

X
k
3k 6k
= −15π − 42,
k=1 k 3k
because 2k(952k − 201) − 5(280k − 51) = 17(112k2 − 106k + 15), and

n
(112k2 − 106k + 15)(−4)k 2k (−1)n (2n + 1)22n+1 2n

X
k n
= −2 + 3n 6n
k 3k
6k
k=1 k 3k n 3n
∞
(350k − 17)8k 2k √

X
k
3k 6k
= 15 2π + 27. (5.10)
k=0 k 3k
Also,
∞
8k 2k

X
k
3k 6k
(21(350k − 17)(2H6k−1 − H3k−1 − Hk−1 ) + 4850)
k=1 k 3k (5.11)
√ √
= 976 + 1020 2π + 945π 2 log 2
and
∞
8k 2k

X
k
3k 6k
(7(350k − 17)(H2k−1 − Hk−1 ) + 2225)
k=1 k 3k (5.12)
493 315
= 276 + √ π + √ π log 2 − 420L,
2 2
where
∞ ∞
( −8
n ) (−1)k(k−1)/2
X
−8 X
L := L 2, = = .
· n2 (2k + 1)2
n=1 k=0
(ii) Let p be any odd prime. Then

p−1 3k 6k
X
k 3k −2 2
2k
≡2 − (mod p),
(2k + 1)8 kk p p
k=0
(p−1)/2 3k 6k −1
!
1 + ( )

X 2 p 3
k 3k
≡ − p qp (2) (mod p2 )
(2k + 1)8 k 2k p 2 4
k=0 k
(p−1)/2
X k(350k + 17) 3k 6k

−2 2 93
k
2k
3k
≡ 15 − p (mod p2 ).
p p 2

k=1
(2k + 1)8k k
Remark 5.4. The series in (5.10) has converging rate 2/27.

∞ 2k k
X
k 8 ((50k − 7)(H2k−1 − Hk−1 ) + 5)
√
3k 6k
= 3 2π(1 + log 2) − 8L. (5.13)
k=1
k k 3k
and
∞ 2k k
X
k 8 ((50k − 7)(2H6k−1 − H3k−1 − Hk−1 ) − 10) √
= 2π(4 + 6 log 2).
k 3k
6k
k=1 k 3k
(5.14)
(ii) We have
∞ 2k k
k 8 ((5k − 1)(H2k−1 − Hk−1 ) − 3(6k − 1)/(8k − 4))
X
3k 6k
k=1
(2k − 1)k k 3k (5.15)
3 √
= 2 π log 2 − L,
8
∞ 2k k
k 8 ((5k − 1)(2H6k−1 − H3k−1 − Hk−1 ) − (6k − 2)/(2k − 1))
X
(2k − 1)k 3k
6k
k=1 k 3k (5.16)
3√
= 2 π log 2,
4
and
∞ 2k k
k 8 ((5k − 1)(12H6k−1 − 6H3k−1 − 4H2k−1 − 2Hk−1 ) − 9)
X
(2k − 1)k 3k
6k
k=1 k 3k (5.17)
√
= 3 2 π log 2 + 4L.
(iii) We have the identity
∞ 2k k (2) (2) 2
X
k 8 ((5k − 1)(16H2k−1 − 3Hk−1 ) − 12(6k − 1)/(2k − 1) ) π3
= √ .
(2k − 1)k 3k 6k

k=1 k 3k
12 2
(5.18)
42 ZHI-WEI SUN
Also,
(p−3)/2 3k 6k (2) (2)
X
k + 1)(16H2k − 3Hk ) + 12(6k + 1)/(2k + 1)2 )
3k ((5k
(2k − 1)k 3k
6k
k=0 k 3k

1 1
≡ − Ep−3 (mod p)
4 4
for any prime p > 3.
Remark 5.5. In the spirit of the arguments in Remark 5.3, (5.10) implies
that
∞ ∞
(50k − 7)8k 2k √ k 2k

X
k
X (5k − 1)8 k π
3k 6k
= 4 + 2 2π and 3k 6k
= √ . (5.19)
k=1
k k 3k k=1
k(2k − 1) k 3k 2 2

∞ 2k
X (130k − 21) k 2π
=− √ . (5.20)
1)(−3)k 3k
6k
k=1
k(2k − k 3k
3 3
Also,
∞ 2k
k ((130k − 21)(H2k−1 + Hk−1 ) − 26(6k − 1)/(2k − 1))
X
k(2k − 1)(−3)k 3k
6k
k=1 k 3k (5.21)
2π
= 2K − √ log 3
3 3
and
∞ 2k
k ((130k − 21)(2H6k−1 − H3k−1 − Hk−1 ) − 16(13k − 4)/(2k − 1))
X
k(2k − 1)(−3)k 3k
6k
k=1 k 3k
2π
= K + √ log 3.
3 3
(5.22)
Remark 5.6. The series in (5.20) has converging rate −1/324. A linear
combination of the last two formulae yields the identity
∞ 2k
k ((130k − 21)(26H6k−1 − 13H3k−1 − 10H2k−1 − 3Hk−1 ) − 572)
X
k(2k − 1)(−3)k 3k
6k
k=1 k 3k
2π
= −33K − √ log 3.
3
∞
(10k − 1)(−27)k 2k

X
k 4π
= −√
3k 6k
(5.23)
k=1
k(2k − 1) k 3k 3
. Also,
∞ 2k k
X
k (−27) ((10k − 1)Hk−1 + 2(6k − 1)/(6k − 3)) √
3k 6k
= 2π 3 log 3 − 18K,
k
k(2k − 1)(−3) k 3k
k=1
(5.24)
∞ 2k k
X
k (−27) ((10k − 1)H2k−1 − 8(6k − 1)/(6k − 3)) √
= 2π 3 log 3 − 9K,
k(2k − 1)(−3)k 3k 6k

k=1 k 3k
(5.25)
and
∞ 2k k
k (−27) ((10k − 1)(2H6k−1 − H3k−1 ) − 2(102k − 29)/(18k − 9))
X
k(2k − 1)(−3)k 3k
6k
k=1 k 3k
2
= 15K − √ π log 3.
3
(5.26)
Remark 5.7. The series in (5.23) has converging rate −1/4. Suitable com-
binations of the last three formulae yield that
∞ 2k k
X
k (−27) (10k − 1)(H2k−1 + 4Hk−1 )
√
3k 6k
= 10π 3 log 3 − 81K
k=1
k(2k − 1) k 3k
and
∞ 2k k
k (−27) ((10k − 1)(6H6k−1 − 3H3k−1 + 11Hk−1 ) − 12)
X
k(2k − 1) 3k
6k
k=1 k 3k
√
= 24π 3 log 3 − 243K.

∞ 2k k 2
k 16 (46k − 11k + 1)
X
3k 6k
= 2π 2 , (5.27)
2 2
k (2k − 1) k 3k
k=1
∞ 2k
k
X
k 16 ((46k2 − 11k + 1)H2k−1 + 5k(6k − 1)/(2k − 1))
= 28ζ(3),
k2 (2k − 1)2 3k
6k
k=1 k 3k
(5.28)
and

2k k 8k(19k−2)
∞
X k 16 (46k2 − 11k + 1)(2H6k−1 − H3k−1 + Hk−1 ) + 2k−1
k2 (2k − 1)2 3k
6k
k=1 k 3k
= 112ζ(3),
(5.29)
44 ZHI-WEI SUN
Also,
∞ 2k k 2
X
k 16 ((46k − 11k + 1)(21H2k−1 − 5Hk−1 ) + 5(2k − 1)2 /(2k))
k2 (2k − 1)2 3k
6k
k=1 k 3k
= 238ζ(3) + 20π 2 log 2
(5.30)
and
∞ 2k k 2 (2) (2)
k 16 (46k − 11k + 1)(292H2k−1 − 77Hk−1 ) 178 4
X
3k
6k = π . (5.31)
k2 (2k − 1)2 k 3k 3
k=1
Remark 5.8. Note that (5.29)−6×(5.28) yields the identity

∞ 2k k 2
k 16 ((46k − 11k + 1)(2H6k−1 − H3k−1 − 6H2k−1 + Hk−1 ) − 14k)
X
k2 (2k − 1)2 3k
6k
k=1 k 3k
= −56ζ(3).
∞
(22k2 − 7k + 1)64k 2k

X
= 4π 2 .
3k 6k
k
(5.32)
2 2
k (2k − 1) k 3k
k=1
Also,
∞ 2k k 2
k 64 ((22k − 7k + 1)Hk−1 + P (k)/(5k(2k − 1)))
X
k2 (2k − 1)2 3k
6k
k=1 k 3k (5.33)
24 2
= (π log 2 + 14ζ(3))
5
and
∞ 2k
k
X
k 64 ((22k2 − 7k + 1)H2k−1 + Q(k)/(5k(2k − 1)))
k2 (2k − 1)2 3k
6k
k=1 k 3k (5.34)
16
= (3π 2 log 2 + 7ζ(3)),
5
where
P (k) = 296k3 − 60k2 + 6k − 1 and Q(k) = 142k 3 − 45k2 + 12k − 2.
Moreover,
∞ 2k
64k ((22k2 − 7k + 1)H(k) − 8k(3k − 1)/(2k − 1))

X
k
= 0, (5.35)
k2 (2k − 1)2 3k
6k
k=1 k 3k
where H(k) = 2H6k−1 − H3k−1 − 2H2k−1 .

(ii) We have

∞ 2k 64k (22k 2 − 7k + 1)(H (2) − 3 H (2) ) − 3k(6k−1)

X k 2k−1 16 k−1 (2k−1)2 π4
= . (5.36)
k2 (2k − 1)2 3k 6k 12

k=1 k 3k

∞ 2k
k
k 16 f (k)
X
3k 6k
= π2, (5.37)
k=1
k2 (2k − 1)2 (6k − 1)(6k − 5) k 3k
where f (k) := 276k3 − 248k2 + 69k − 5. Also,

∞ 2k k
k 16 (f (k)(8H6k−1 − 4H3k−1 + H2k−1 − Hk−1 ) − f1 (k))
X
k2 (2k − 1)2 (6k − 1)(6k − 5) 3k
6k
k=1 k 3k (5.38)
9
= π 2 + 10π 2 log 2
2
and
∞ 2k k
k 16 (f (k)(2H6k−1 − H3k−1 − H2k−1 + Hk−1 ) − f2 (k))
X
= 3π 2 ,
k2 (2k − 1)2 (6k − 1)(6k − 5) 3k
6k
k=1 k 3k
(5.39)
where
235 25 30
f1 (k) := 54k2 + 222k − + and f2 (k) = 186k2 − 227k − 15 − .
2 2k 2k − 1
Conjecture 5.11 (2023-08-23). Let P (k) = 828k 3 − 888k2 + 207k − 11.
(i) We have
∞ 2k k
k 16 P (k) 3
X
3k 6k = π 2 , (5.40)
k(2k − 1)2 (6k − 1)(6k − 5) 2
k=1 k 3k
∞ 2k k 2
k 16 (P (k)(H2k−1 − Hk−1 ) − (6k − 1) (138k − 109)/(4(2k − 1)))
X
2 (6k − 1)(6k − 5) 3k 6k

k=1
k(2k − 1) k 3k
= 3π 2 log 2 − 21ζ(3),
(5.41)
and
∞ 2k k
X
k 16 (P (k)(2H6k−1 − H3k−1 ) − (6k − 1)(84k2 − 4k − 79)/(4(2k − 1)))
k(2k − 1)2 (6k − 1)(6k − 5) 3k
6k
k=1 k 3k
= 3π 2 log 2 + 21ζ(3).
(5.42)
(ii) We have
∞ 2k k (2) (2) 2 2
k 16 (P (k)(5H2k−1 − Hk−1 ) − (6k − 1) (231k − 185)/(2k − 1) )
X
k(2k − 1)2 (6k − 1)(6k − 5) 3k 6k

k=1 k 3k
π4
=−
2
(5.43)
46 ZHI-WEI SUN
and
∞ 2k
k (2) (2) (2) (2)
X
k 16 (P (k)(556H6k−1 − 139H3k−1 − 692H2k−1 + 194Hk−1 ) + 4g(k))
k(2k − 1)2 (6k − 1)(6k − 5) 3k
6k
k=1 k 3k
55 4
=− π ,
2
(5.44)
where
6090k2 + 4225k − 2217
g(k) := .
2k − 1
Conjecture 5.12. We have
∞ 2k k 3 2
k 16 (828k + 1320k − 745k + 65)
X
= 16π 2 . (5.45)
k 2 (2k − 1)(6k − 1)(6k − 5) 3k 6k
k=1 k 3k
Remark 5.12. We haven’t found any variant of this identity with summands
involving harmonic numbers.
6. Conjectural series with summands involving products of

three or more binomial coefficients
∞ (2)
X (21k − 8)Hk−1 + 2/k π4
3 = . (6.1)
k3 2k
360
k=1 k
p−1
X (2) 2 16
(21k + 8)Hk − ≡ p4 Bp−5 (mod p5 ).
k 5
k=1
Remark 6.1. In 1993, D. Zeilberger [29] established the identity

∞
X 21k − 8 π2
3 = .
3 2k 6
k=1 k k
The author [17] proved that
p−1 3
X 2k
(21k + 8) ≡ 8p + 16p4 Bp−3 (mod p5 )
k
k=0
for any prime p. The author’s conjectural identity
∞
47π 4

X 21k − 8 (2) 25 (2)
3 H 2k−1 − H k−1 =
3 2k
8 2880
k=1 k k
(cf. [21, Conjecture 3.1]) was confirmed by K.C. Au [2]. We will prove the
identity (6.1) in a forthcoming preprint joint with Q.-H. Hou.

∞ (3)
X (21k − 8)Hk−1 + 1/k 2 62 16
3 = ζ(5) − π 2 ζ(3). (6.2)
2k
k3 k 7 21
k=1

p−1 3
X 2k (3) 1 256 Hp−1 752 3
(21k + 8)Hk − 2 ≡ · + p Bp−5 (mod p4 ).
k k 7 p 35
k=1
Remark 6.2. The author [21, Conjecture 3.2] conjectured that

∞
X 21k − 8 (3) 43 (3) 711 29
3 H2k−1 + Hk−1 = ζ(5) − π 2 ζ(3),
3 2k
8 28 14
k=1 k k
which remains open.

∞ (4)
X 9(21k − 8)Hk−1 + 25/k3 13π 6
3 = . (6.3)
k3 2k 3780
k=1 k

(p−1)/2 3
X 2k (4) 25 Hp−1 96 2
9(21k + 8)Hk − 3 ≡ 96 + p Bp−5 (mod p3 ).
k k p2 5
k=1
Remark 6.3. This conjecture looks quite challenging.

∞
16k π4

X (2) 1
3 (6k − 2)H k−1 + = . (6.4)
3 2k
k 24
k=1 k k

p−1 2k 3
X
k (2) 1
(6k + 2)H k −
16k k
k=1
4
≡ − 4qp (2) + 2p qp (2)2 − p2 qp (2)3 + p3 qp (2)4 (mod p4 ).
3
∞
(3k − 1)16k π4

X (2) 5 (2)
H 2k−1 − H =
k3 2k
3 4 k−1 24
k=1 k
(cf. [21]) was confirmed by K.C. Au [2] as well as C. Wei [25]. We will prove
the identity (6.4) in a forthcoming preprint joint with Q.-H. Hou.
48 ZHI-WEI SUN

∞
8k π4

X (2) 4
2 (50k − 15)H k−1 + = . (6.5)
k 3 2k
3k k 24
k=1 k k

p−1 2k2 3k
X
k k (2) 4
(50k + 15)H k −
8k k
k=1
≡ − 12qp (2) + 6p qp (2)2 − 4p2 qp (2)3 + 3p3 qp (2)4 (mod p4 ).
∞
X 8k ((10k − 3)(H2k−1 − Hk−1 ) − 1) 7
2 = ζ(3)
k3 2k
3k 2
k=1 k k
(cf. [21, Conjecture 3.7]) remains open.

∞
(−27)k 4π 3

X 1
(15k − 4)(H 2k−1 − H k−1 ) + = − √ . (6.6)
3 2k 2 3k 2k − 1 3 3

k=1 k k k

p−1 2k 2 3k
X
k k 1
(15k + 4)(H2k − Hk ) +
(−27)k 2k + 1
k=0
p 4
1
≡ + p2 Bp−2 (mod p3 ).
3 3 3
Remark 6.6. As conjectured by the author [17] and confirmed by Kh.
Hessami Pilehrood and T. Hessami Pilehrood [13], we have
∞ ∞
X (15k − 4)(−27)k−1 X ( k3 )
2 3k = .
k 3 2k k2
k=1 k k k=1
The author [21, Conjecture 3.8] also conjectured that

∞
X (−27)k ((15k − 4)(3H3k−1 − Hk−1 ) − 9) 4π 3
2 3k = − √ ,
k3 2k 3

k=1 k k
which has been confirmed by Q.-H. Hou and the author.

∞
64k

X (2) 8 8
2 (55k − 15)Hk−1 + = π4 . (6.7)
3 2k
3k
k 3
k=1 k k k

p−1 2k 2 3k
X
k k (2) 8
(55k + 15)Hk −
64k k
k=1
∞
X 64k−1 ((11k − 3)(2H2k−1 + Hk−1 ) − 4) 7
2 3k = ζ(3)
k3 k2k 2
k=1 k

∞
81k

X (2) 27
(350k − 80)Hk−1 + = 4π 4 . (6.8)
3 2k 2 4k k
k=1 k k 2k

p−1 2k2 4k
X
k 2k (2) 27
(350k + 80)Hk −
81k k
k=1
∞
X 81k ((35k − 8)(H4k−1 − Hk−1 ) − 35/4)
2 3k = 12π 2 log 3 + 39ζ(3)
k 3 2k
k=1 k k

∞ (3)
X 17(145k2 − 104k + 18)Hk−1 + 28(2k − 1)/k 2
= 528ζ(5) − 46π 2 ζ(3).
2k 3k 2

3
k (2k − 1) k k
k=1
(6.9)
p−1 2k 3k 3

X
2 4572 6
(145k +104k+18) k k
≡ 18p−288p2 Hp−1 + p Bp−5 (mod p7 ).
2k + 1 5
k=0
Remark 6.9. This is related to the author’s conjectural identity

∞
X 145k2 − 104k + 18 π2
=
3 2k 3k 2 3

k=1 (2k − 1)k k k
(cf. [22]) confirmed by K.-C. Au.

50 ZHI-WEI SUN

∞
X (5535k3 − 4689k2 + 1110k − 80)(−27)k−1
6k2 = K. (6.10)
k=1 k3 (3k − 1)(3k − 2) 3k
k 3k
(p−1)/2 6k2
X P (k) 3k p
k 3k
≡ 4p (mod p2 ) (6.11)
(3k + 1)(3k + 2)(−27)k 3
k=0
and
p−1 6k2
P (k) 3k p 400
X
3 1
k 3k
≡ 40p + p B p−2 (mod p4 ),
(3k + 1)(3k + 2)(−27)k 3 3 3
k=0
(6.12)
3 2
where P (k) = 5535k + 4689k + 1110k + 80.
Remark 6.10. The series in (6.10) has converging rate −1/1024.
∞
(−1)k−1

X
2 (3) (3) 125
8(205k − 160k + 32)(H2k−1 + Hk−1 ) + = 16ζ(3)2 .
5 2k 5 k
k=1 k k
(6.13)
p−1 5
k 2k (3) (3) 125
X
2
(−1) 8(205k + 160k + 32)(H2k + Hk ) −
k k
k=1
≡ −3584p4 Bp−5 (mod p5 ).
Remark 6.11. In 1997, T. Amdeberhan and Zeilbeger [1] obtained the
identity
∞
X (−1)k (205k2 − 160k + 32)
5 = −2ζ(3).
k=1 k5 2k
k

∞
(−1)k−1

X
2 (4) (4) 195
5 16(205k − 160k + 32)(H2k−1 + 3Hk−1 ) − k2 = 32ζ(7).
k 5 2k
k=1 k
(6.14)
p−1 5
k 2k (4) (4) 195
X
2
(−1) 16(205k + 160k + 32)(H2k + 3Hk ) + 2
k k
k=1

Hp−1 34 3
≡ 1024 − p Bp−5 (mod p4 ).
p 5
Remark 6.12. Note that (6.14) gives a surprising series for ζ(7).

∞
(−256)k

X
2 (3) (3) 2
(10k − 6k + 1)(8H 2k−1 − H k−1 ) + = −196ζ(3)2 . (6.15)
5 2k 5 k
k=1 k k

p−1 2k 5
X
k 2 (3) (3) 2
(10k + 6k + 1)(8H 2k − H k ) −
(−256)k k
k=1
≡ −16qp (2) + 8p qp (2)2 (mod p2 ).
Remark 6.13. It is known that
∞
X (10k2 − 6k + 1)(−256)k
5 = −28ζ(3)
k=1 k5 2k
k
(cf. [11, Identity 8]).

∞
X (−256)k 2

(4) (4)

5 (10k − 6k + 1) 16H 2k−1 + 3H k−1 = −508ζ(7). (6.16)
k 5 2k
k=1 k

p−1 2k 5
X
2

(4) (4)
56
k
k
(10k + 6k + 1) 16H 2k + 3H k ≡ − pBp−3 (mod p2 ).
(−256) 3
k=1
Remark 6.14. The identity (6.18) looks interesting and challenging.

∞
(−64)k

X
2 (3) (3) 4
(28k − 18k + 3)(4H2k−1 + 3Hk−1 ) + = −49ζ(3)2 .
5 2k 4 3k k
k=1 k k k
(6.17)
p−1 2k 4 3k
X
k k 2 (3) (3) 4
(28k + 18k + 3)(4H 2k + 3H k ) −
(−64)k k
k=1
≡ −24qp (2) + 12p qp (2)2 (mod p2 ).
∞
X (28k2 − 18k + 3)(−64)k
4 3k = −14ζ(3)
k=1 k5 2k
k k
(cf. [18]) was confirmed by Au [2].

52 ZHI-WEI SUN

∞
256k 4 31π 8

(4) (4)
X
3 2
7 (21k − 22k + 8k − 1) 16H2k−1 + 7Hk−1 + k = 1440 .
7 2k
k=1 k k
(6.18)
p−1 2k 7
X
k 3 2

(4) (4)
4
(21k + 22k + 8k + 1) 16H2k + 7Hk −
256k k
k=1
32 2
≡ −32qp (2) + 16p qp (2)2 − p qp (2)3 + 8p3 qp (2)4 (mod p4 ).
3
Remark 6.16. This is inspired by the identity
∞
X (21k3 − 22k2 + 8k − 1)256k π4
7 =
k7 2k
8
k=1 k
conjectured by Guillera [10].

∞ 2k 3

X
k (3) 352 32 ζ(3)
(42k + 5)Hk − = 335 − 224G , (6.19)
4096k (2k + 1)2 7 π
k=0
where G is the Catalan constant.

(ii) For any prime p > 3 with p 6= 7, we have
(p−3)/2 2k 3
X
k (3) 352
(42k + 5)Hk −
4096k (2k + 1)2
k=0

−1
≡ − 32 Bp−3 + 16Ep−3 (mod p),
p
where E0 , E1 , E2 , . . . are the well known Euler numbers.
Remark 6.17. The author [21, (3.70)] conjectured that

∞ 2k 3
X
k (3) 43 (3) 555 ζ(3) 32
(42k + 5) H2k − H = · − G,
4096k 352 k 77 π 11
k=0
which is still open.

∞ 2k 3 (4)
! !
X (4) Hk 25 5
k
9(42k + 5) H2k − + = π3. (6.20)
4096k 16 (2k + 1)3 6
k=0
(ii) For any odd prime p 6= 5, we have

(p−3)/2 2k 3 (4)
! !
X
k (4) Hk 25
9(42k + 5) H2k − +
4096k 16 (2k + 1)3
k=0

−1
≡ −4 Bp−3 (mod p).
p
Remark 6.18. This was motivated by (1.54), (1.58) and Conjecture 6.17.
∞ 2k 2 4k (2) (2)
! !
X
k 2k (2) H H 1 π
k
(20k + 3) H4k − 2k − k + = . (6.21)
(−1024) 4 16 4k + 1 6
k=0

p−1 2k2 4k (2) (2)
! !
X
k 2k (2) H H 1
p (20k + 3) H4k − 2k − k +
(−1024)k 4 16 4k + 1
k=0

−1 10
≡ + p2 Ep−3 (mod p3 ).
p 3
Remark 6.19. This was motivated by the Ramanujan series
∞ 2k 2 4k

X
k 2k 8
(20k + 3) k
= .
(−1024) π
k=0

∞ √
(66k2 + 37k + 4) 2k
3k 4k

X
k k 2k 27 3
= . (6.22)
(2k + 1)729k 2π
k=0
Also,
∞ 2k 3k 4k √
X
k k 2k 81 3 log 3
(P (k)(2H4k − H2k − 3Hk ) + 60k + 26) =
(2k + 1)729k 2π
k=0
(6.23)
and
∞ 2k
3k
4k

108k2 + 92k + 23
X
k k 2k
P (k)(3H3k − 2H2k + 3Hk ) − = 0,
(2k + 1)729k 2k + 1
k=0
(6.24)
where P (k) = 66k2 + 37k + 4.
√
Remark 6.20. Note that the series in (6.22) for 3/π is not of the Ra-
manujan type.
Conjecture 6.21 (2023-11-23). Let P (k) = 360k 3 + 612k2 + 230k + 15.
54 ZHI-WEI SUN
(i) We have
∞ 6k2 √
X P (k) 3k
k 3k 32 2
= . (6.25)
(3k + 1)(3k + 2)215k π
k=0
Moreover, for any odd prime p we have
(p−1)/2 6k2
P (k) 3k

X −2 2 2
k 3k
≡ 12p − 18p (mod p3 ) (6.26)
(3k + 1)(3k + 2)215k p p
k=0
and
p−1 6k2
P (k) 3k

X 15 −2 225 3 1
k 3k
15k
≡ p − p Ep−3 (mod p3 ).
(3k + 1)(3k + 2)2 2 p 32 4
k=0
(6.27)
(ii) We have
∞ 3k 6k2 2
X
k 3k (P (k)(H2k − Hk ) − 180k + 36k + 23)
√ log 2
= 48 2 , (6.28)
(3k + 1)(3k + 2)215k π
k=0
and
∞ 3k 6k2 √ log 2
3k (P (k)(4H6k − 3H3k − Hk ) + f (k))
X
k
15k
= 192 2 , (6.29)
(3k + 1)(3k + 2)2 π
k=0
where f (k) = (1296k3 + 1980k2 + 912k + 139)/((3k + 1)(3k + 2)).
∞ 2k 5
X (28k2 + 10k + 1) k 3
3k 6k
= , (6.30)
(6k + k
1)(−64) k 3k π
k=0
∞ 2k 5
X ((28k2 + 10k + 1)(2H2k − 3Hk ) + 20k + 4) k 18 log 2
3k 6k
= , (6.31)
(6k + 1)(−64)k π
k=0 k 3k
and
∞ 5
X ((28k2 + 10k + 1)(2H6k − H3k − 3Hk ) + f (k)) 2k
k 30 log 2
3k 6k
= , (6.32)
π

(6k + 1)(−64)k
k=0 k 3k
where f (k) = 4(138k2 + 52k + 5)/(3(6k + 1)).

(ii) Let p be an odd prime. If p > 3, then
X (28k2 + 10k + 1) 2k 5
(p−1)/2
−1 7 4
k
3k 6k
≡ p + p B p−3 (mod p5 ).
k
(6k + 1)(−64) k 3k p 48
k=0
When p 6= 5, we have
p−1 2k 5
(28k2 + 10k + 1) p3

X −1
k
3k 6k
≡p + Ep−3 (mod p4 ).
(6k + k
1)(−64) k 3k p 5
k=0
Remark 6.22. The series in (6.30) has converging rate −1/27.

2k 5

(2) (2)
X∞ (28k2 + 10k + 1)(10H2k − 3Hk ) + 2 k π
= . (6.33)
(6k + 1)(−64)k 3k
6k
2
k=0 k 3k
5
(p−1)/2 (28k 2 + 10k + 1)(10H (2) − 3H (2) ) + 2 2k
X 2k k k
p 3k 6k
k=0
(6k + 1)(−64)k k 3k

−1 35 3
≡ 2 + p Bp−3 (mod p4 )
p 24
and
5
p−1 (28k 2 + 10k + 1)(10H (2) − 3H (2) ) + 2 2k
X 2k k k
p 3k 6k
k
(6k + 1)(−64) k 3k
k=0

−1
≡2 + 2p2 Ep−3 (mod p3 ).
p
Remark 6.23. We haven’t found identities similar to (6.33) involving har-
monic numbers of order greater than two.
∞ 2k 5
X
k 2 (3) 8 64ζ(3)
k
(20k + 8k + 1)H k + = . (6.34)
(−4096) 2k + 1 π2
k=0
Remark 6.24. This was motivated by the known identity

∞ 2k 5
X
2 k 8
(20k + 8k + 1) k
= 2
(−4096) π
k=0
(cf. [11, Identity 8]). The author [21] conjectured that
∞ 2k 5
8
(2) (2)
X
k 2
(20k + 8k + 1) 8H 2k − 3H k + 4 = ,
(−4096)k 3
k=0
which was later confirmed by Wei [26].
∞ 2k5
X
2

(4) (4)
56
k
k
(20k + 8k + 1) 64H 2k − 3H k = − π2 . (6.35)
(−4096) 45
k=0
p−1 2k 5
X
2

(4) (4)
224
k
(20k + 8k + 1) 64H 2k − 3H k ≡− pBp−3 (mod p2 ).
(−4096)k 3
k=0
56 ZHI-WEI SUN
Remark 6.25. We haven’t found similar results involving harmonic num-

bers of the fifth order.
∞ 2k 5
X
k 2

(3) (3)
125 ζ(3)
20 k
(820k + 180k + 13) 9H2k − Hk + = 1024 2 .
(−2 ) 2k + 1 π
k=0
(6.36)
(p−3)/2 2k 5
X
k 2

(3) (3)
125
(820k + 180k + 13) 9H2k − Hk +
(−220 )k 2k + 1
k=0
≡ −256Bp−3 (mod p).
Remark 6.26. This was motivated by the known identity
∞ 2k 5
X
2 128
(820k + 180k + 13) k20 k = 2
(−2 ) π
k=0
(cf. [11, Identity 9]). The author [21] conjectured that

∞ 2k 5 128
(2) (2)
X
k 2
(820k + 180k + 13) 11H 2k − 3H k + 43 = ,
(−220 )k 3
k=0
which was later confirmed by Wei [26].

∞ 2k 5
X
k 2

(4) (4)
195
(820k + 180k + 13) 49H 2k − 3H k −
(−220 )k (2k + 1)2 (6.37)
k=0
896 2
=− π
45
and
∞ 2k 5
X
k 2

(5) (5)
275
(820k + 180k + 13) 351H2k − 11Hk +
(−220 )k (2k + 1)3
k=0

ζ(5)
= 512 85 2 − 7ζ(3) .
π
(6.38)
(p−3)/2 2k 5

X
k 2

(4) (4)
195
(820k + 180k + 13) 49H2k − 3Hk −
(−220 )k (2k + 1)2
k=0
≡ 0 (mod p).
Remark 6.27. It seems that (6.38) is the first nontrivial series for ζ(5)/π 2 .

∞ 7
X (92k3 + 54k2 + 12k + 1) 2k 12
k = , (6.39)
(6k + 1)256k 3k 6k π2
k=0 k 3k
and
∞ 2k 7
(92k3 + 54k2 + 12k + 1)(2H6k − H3k + 15Hk ) − f (k)

X
k
(6k + 1)256k 3k
6k
k=0 k 3k (6.40)
log 2
= −288 2
π
with f (k) = 2(5064k3 + 2828k2 + 542k + 35)/(3(6k + 1)).
(ii) Let p be an odd prime. If p > 3, then
X (92k3 + 54k2 + 12k + 1) 2k 7

(p−1)/2
93
3k 6k
k
≡ p2 + p7 Bp−5 (mod p8 ).
80

k
(6k + 1)256 k 3k
k=0
When p 6= 5, we have
p−1 7
X (92k3 + 54k2 + 12k + 1) 2k 7 5
k ≡ p2 + p Bp−3 (mod p6 ).
(6k + 1)256k 3k 6k 10
k=0 k 3k
Remark 6.28. The hypergeometric series in (6.39) has converging rate

4/27.
Conjecture 6.29 (2023-11-15). (i) We have and
7 3 + 54k 2 + 12k + 1)(20H (2) − 7H (2) ) + 44k + 10

∞ 2k (92k
X k 2k k
3k 6k
= 4.
k
(6k + 1)256 k 3k
k=1
(6.41)
7 3 + 54k 2 + 12k + 1)(20H (2) − 7H (2) ) + 44k + 10

p−1 2k (92k
X k 2k k
k 3k 6k
k=1
(6k + 1)256 k 3k
≡ 14p3 Bp−3 (mod p4 ).
If p > 3, then
2k 7

(2) (2)
(p−1)/2 (92k3 + 54k2 + 12k + 1)(20H2k − 7Hk ) + 44k + 10

X k
(6k + 1)256k 3k
6k
k=1 k 3k
93 5
≡ p Bp−5 (mod p6 ).
4
Remark 6.29. We haven’t found identities similar to (6.41) with summands
involving harmonic numbers of order at least three.
58 ZHI-WEI SUN

∞ 2k7
X
k 3 2 (4) (4) 64 976
20k
(168k + 76k + 14k + 1)(128H2k − 7Hk ) + = π.
2 2k + 1 45
k=0
(6.42)
(ii) For any odd prime p 6= 5, we have
(p−3)/2 2k 7
X
k 3 2

(4) (4)
64
(168k + 76k + 14k + 1) 128H2k − 7Hk +
220k 2k + 1
k=0
≡ −256qp (2) (mod p).
Remark 6.30. This was motivated by the identity

∞ 2k 7

X
3 2 k 32
(168k + 76k + 14k + 1) 20k = 3
2 π
k=0
conjectured by B. Gourevich (cf. [8]). The author [21] conjectured that

∞ 2k 7
X
k 3 2

(2) (2)
80
(168k + 76k + 14k + 1) 16H2k − 5Hk + 8(6k + 1) = .
220k 3π
k=0
Conjecture 6.31 (2023-06-19). Set
P (x) = 4528x4 + 3180x3 + 972x2 + 147x + 9.

(i) We have
∞ 2k 7 3k 4k
X
k k 2k H2k 3 2
H4k + P (k) − 484k − 108k + 9k + 3 = 0,
(−224 )k 2
k=0
(6.43)
∞ 2k 7 3k 4k
512
(2) (2)
X
k k 2k
11H2k − 4Hk P (k) + 1780k2 + 633k + 63 = 2 ,
(−224 )k π
k=0
(6.44)
and
∞ 2k 7 3k 4k
916k2 + 65k − 9

(3) (3)
X
k k 2k
9H2k − 2Hk P (k) − = 0. (6.45)
(−224 )k 2k + 1
k=0

p−1 2k 7 3k 4k
X
k k 2k H2k 3 2
H4k + P (k) − 484k − 108k + 9k + 3
(−224 )k 2
k=0
1953 8
≡ 3p3 − p Bp−5 (mod p9 ),
10
p−1 2k 7 3k 4k
(2) (2)
X
k k 2k
11H2k − 4Hk P (k) + 1780k2 + 633k + 63
(−224 )k
k=0
9207 7
≡ 63p2 − p Bp−5 (mod p8 ),
10
and
(p−1)/2 2k 7 3k 4k
916k2 + 65k − 9

(3) (3)
X
k k 2k
9H2k − 2Hk P (k) −
(−224 )k 2k + 1
k=0
81 2 243 6
≡ 9p − p Hp−1 − p Bp−5 (mod p7 ).
2 8
Remark 6.31. This was motivated by the conjectural identity
∞ 2k 3 3k 4k
X 768
P (k) k k 2k
= 4
(−224 )k π
k=0
(cf. [8]).
Conjecture 6.32 (2023-06-19). Set
Q(x) = 43680k4 + 20632k3 + 4340k2 + 466k + 21
and
R(x) = 87360x3 + 30948x2 + 4340x + 233.
(i) We have
∞ 2k 8 4k
X
k 2k 215 log 2
((2(H4k + 3H2k − 4Hk )Q(k) + R(k)) = (6.46)
232k π4
k=0
and
∞ 2k 8 4k 2048
(2) (2)
X
k 2k
Q(k) + 3624k2 + 926k + 69 =
7H2k − 2Hk .
232k 3π 2
k=0
(6.47)
p−1 2k 8 4k
X
k 2k
((2(H4k + 3H2k − 4Hk )Q(k) + R(k))
232k
k=0
≡ 233p3 + 336p4 qp (2) − 168p5 qp (2)2 + 112p6 qp (2)3 − 84p7 qp (2)4 (mod p8 ).
p−1 2k 8 4k
(2) (2)
X
k 2k 2
7H 2k − 2H k Q(k) + 3624k + 926k + 69
232k
k=0
1953 7
≡ 69p2 + p Bp−5 (mod p8 ).
20
60 ZHI-WEI SUN
Remark 6.32. This was motivated by the conjectural identity

∞ 2k8 4k
X 2048
Q(k) k 32k2k = 4
2 π
k=0
(cf. [8]).
Acknowledgments. The author would like to thank Dr. Qing-Hu Hou,

Chen Wang, Ce Xu and Yajun Zhou for helpful comments.
References
[1] T. Amdeberhan and D. Zeilberger, Hypergeometric series acceleration via the WZ
method, Electron. J. Combin., 4(2) (1997), #R3.
[2] K. C. Au, Multiple zeta values, WZ-pairs and infinite sums computations,
arXiv:2212.02986, 2022.
[3] G. Almkvist, C. Krattenthaler and J. Petersson, Some new series for π, Experiment.
Math. 12 (2003), 441–456.
[4] K. C. Au, Mathematica Package MultipleZetaValue, 2022. Available from the website:
https://researchgate.net/publication/357601353
[5] K. C. Au, Iterated integrals and multiple polylogarithm at algebraic arguments,
arXiv:2201.01676, 2022.
[6] F. Bellard, π Formulas, Algorithms and Computations, https://bellard.org/pi.
[7] W. Chu and W. Zhang, Accelerating Dougall’s 5 F4 -sum and infinite series involving
π, Math. Comp. 83 (2014), 475–512.
[8] H. Cohen and J. Guillera, Rational hypergeometric identities for 1/π c : survey and
generalizations, arXiv:2101.12592, 2021.
[9] S. Cooper, Ramanujan’s Theta Functions, Springer, Cham, 2017.
[10] J. Guillera, About a new kind of Ramanujan-type series, Experiment. Math. 12 (2003),
507-510.
[11] J. Guillera, Hypergeometric identities for 10 extended Ramanujan-type series, Ra-
manujan J. 15 (2008), 219–234.
[12] V.J.W. Guo and X. Lian, Some q-congruenceson double basic hypergeometric sums,
J. Difference Equ. Appl. 27 (2021), 453–461.
[13] Kh. Hessami Pilehrood and T. Hessami Pilehrood, Bivariate identities for values
of the Hurwitz zeta function and supercongruences, Electron. J. Combin. 18 (2012),
#P35, 30pp.
[14] C. Li and W. Chu, Infinite series about harmonic numbers inspired by Ramanujan-like
formulae, Electron. Res. Arch. 31 (2023), 4611–4636.
[15] L. Pinelis, Answer to Question 450162 at MathOverflow, July 4, 2023.
https://mathoverflow.net/questions/450162/450164#450164
[16] S. Ramanujan, Modular equations and approximations to π, Quart. J. Math. (Oxford)
(2) 45 (1914), 350–372.
[17] Z.-W. Sun, Super congruences and Euler numbers, Sci. China Math. 54 (2011), 2509–
2535.
[18] Z.-W. Sun, Products and sums divisible by central binomial coefficients, Electron. J.
Combin. 20 (2013), no. 1, #P9, 1-14.
[19] Z.-W. Sun, New series for some special values of L-functions, Nanjing Univ. J. Math.
Biquarterly 32 (2015), 189–218.
k mk
P
[20] Z.-W. Sun, Series of the type ∞ k=0 (ak + b)x / nk , arXiv:2204.08275, 2022.
[21] Z.-W. Sun, Series with summands involving harmonic numbers, in: M. B. Nathanson
(ed.), Combinatorial and Additive Number Theory, Springer, to appear. See also
arXiv:2210.07238.
[22] Z.-W. Sun, Three conjectural series for π 2 and related identities, Question 456443 at
MathOverflow, Oct. 14, 2023. https://mathoverflow.net/questions/456443
[23] Q.-H. Hou, X. Wang and H. He, Some series involving harmonic numbers,
arXiv:2305.00498.
[24] C. Wei, On two double series for π and their q-analogues, Ramanujan J. 60 (2023),
615–625.
[25] C. Wei, On two conjectural series for π and their q-ananlogues, arXiv:2211.11484,
2022.
[26] C. Wei, On some conjectural series containing binomial coefficients and harmonic
numbers, arXiv:2306.02461, 2023.
[27] C. Wei and G. Ruan, Double series for π and their q-ananlogues, arXiv:2210.01331,
2022.
[28] D. Zagier, The dilogarithm function, in: P. Cartier, P. Moussa, B. Julia, P. Vanhove
(eds.). Frontiers in Number Theory, Physics and Geometry II, Springer, Berlin, 2007,
pp. 3-65.
[29] D. Zeilberger, Closed form (pun intended!), Contemporary Math. 143 (1993), 579–
607.
Department of Mathematics, Nanjing University, Nanjing 210093, People’s

Republic of China
Email address: zwsun@nju.edu.cn
Homepage: http://maths.nju.edu.cn/~ zwsun

New Series Involving Binomial Coefficients

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

New Series Involving Binomial Coefficients

Uploaded by

Copyright:

Available Formats

Preprint, arXiv:2307.

NEW SERIES INVOLVING BINOMIAL COEFFICIENTS

Abstract. In this paper, we deduce several new identities on infinite

(4k + 1)(−n)k (n + 1)2k 4k

= 8n(n + 1)2 − 2n(n2 − 1) log n.

In addition, we pose many new conjectural series identities involving

A celebrated result of Euler states that

for any x0 ∈ (−27/4, 27/4). As observed in [20], x3 = x0 (x − 1) for some

Theorem 1.1 with x = −1 yields the following corollary.

Theorem 1.1 with x = −2 yields the following corollary.

Now we state our second theorem.

Motivated by Theorem 1.1 and 1.2, we formulate the following conjecture

Conjecture 1.1. For any x ∈ (−3, c) with x 6= 0, we have

where s(x) and t(x) are suitable rational functions in x.

We also have the following conjecture.

Conjecture 1.2. If −3 < x < c with c given by (1.2), then

Recall that the harmonic numbers are given by

We also establish the following theorem.

Theorem 1.3. We have

where arctanh t is the inverse hyperbolic tangent function.

Theorem 1.4. (i) Let x > 1/4. Then

1.4(ii) with x replaced by 1 − x also hold. In view of this, Theorem 1.4

for any r ∈ {0, 1} and x0 ∈ (−16, 16).

Proof. Putting x = 1 in (1.30) and(1.31), we get the following two identities:

Similar to Conjecture 1.1, we formulate the following conjecture.

Now we state our last theorem.

We are going to prove Theorems 1.1-1.3 and Theorems 1.4-1.6 in Sections

where a, b, m are integers with am 6= 0, d ∈ Z+ = {1, 2, 3, . . .} and ck takes

(cf. [16]), the author [21, Remark 3.19] conjectured that

which was later confirmed in [23].

Note that 7×(1.57)−4×(1.56) yields the identity

Motivated by (1.58), we make the following conjecture on congruences.

Conjecture 1.5. Let p > 3 be a prime. Then

In order to prove Conjecture 3.14 of the author [21], C. Li and W. Chu

2. Proofs of Theorems 1.1-1.3

where q(x) is defined by (1.4).

This lemma is known, see [20, Lemma 3.1].

Proof of (1.6). For each k ∈ N, we have

1 k!(2k)! Γ(k + 1)Γ(2k + 1)

by Lemma 2.1. Therefore

Combining (2.3) with (2.4), we obtain

which is equivalent to the desired (1.6).

Proof of (1.9). As (2.2) holds for any positive integer k, we have

where x0 = x3 /(x − 1) and

where arctanh t is the inverse hyperbolic tangent function. If x0 < 0, then

For the function

Similarly, if x0 < 0 then

with the aid of Lemma 2.1.

(1 − x)(2x + 3) 2x3 + 3x2 − 6x + 9

Proof of (1.7). Set x0 = x3 /(x − 1).

Using Mathematica 13 we find that

By using Au’s Mathematica Package MultipleZetaValues available from

Note that (2.8) minus (2.7) yields the identity

which is similar to the identity

(with n ∈ Z+ ) observed by K. C. Au [5].

Proof of Theorem 1.3. In view of (2.9),

where the last equality can be yielded by Mathematica. So we have proved

3. Proofs of Theorems 1.4-1.6

Proof. The three identities can be easily proved by induction on n.

for any integer n > 0. Letting n → +∞, we then obtain

Combining this with (3.2) and noting that

for any integer n > 0, and hence