Zhi Wei SUN CONJECTURES

To appear in: M. B. Nathanson (ed.), Combinatorial and Additive Number Theory, Springer.
SERIES WITH SUMMANDS

INVOLVING HARMONIC NUMBERS
arXiv:2210.07238v8 [math.NT] 2 Jun 2023
ZHI-WEI SUN
Abstract. For each positive integer m, the mth order harmonic num-
bers are given by
X 1
Hn(m) = (n = 0, 1, 2, . . .).
km
0<k6n
We discover exact values of some series involving harmonic numbers of

order not exceeding three. For example, we conjecture that
2k 3
∞
X
k (3) 7 (3) 25ζ(3)
(6k + 1) H 2k − H = − G,
256k 64 k 8π
k=0
k 2
P
where G denotes the Catalan constant ∞ k=0 (−1) /(2k+1) . This paper
contains 66 conjectures posed by the author since October 2022.
1. Introduction
The usual harmonic numbers are those rational numbers
X 1
Hn = (n = 0, 1, 2, . . .).
k
0<k6n
For each m ∈ Z+ = {1, 2, 3, . . .}, the harmonic numbers of order m are

defined by
X 1
Hn(m) = (n ∈ N = {0, 1, 2, . . .}).
km
0<k6n
For any m, n ∈ Z+ , we clearly have
n 2n n
X 1 X 1 X 1 (m) 1
= − = H2n − m Hn(m) .
(2k − 1)m km (2j)m 2
k=1 k=1 j=1
J. Wolstenholme [46] established two fundamental congruences for harmonic

numbers:
(2)
Hp−1 ≡ 0 (mod p2 ) and Hp−1 ≡ 0 (mod p)
for any prime p > 3. For series and congruences involving harmonic numbers,
one may consult [28, 35, 40], [39, Section 10.5], and the recent preprint [4]
solving various conjectures of the author.
Key words and phrases. Harmonic numbers, series for π, Dirichlet L-functions, combi-
natorial identities, congruences.
2020 Mathematics Subject Classification. Primary 11B65, 05A19; Secondary 11A07,
11B68.
Supported by the Natural Science Foundation of China (grant no. 11971222).
1
2 ZHI-WEI SUN
In 2012, K. N. Boyadzhiev [7] proved that

∞ √
X 2k k 2 1 + 1 − 4x 1 1
Hk x = √ log √ for x ∈ − , . (1.1)
k 1 − 4x 2 1 − 4x 4 4
k=0
In 2016, H. Chen [8] deduced that
∞ √
X 2k k 1 1 + 1 − 4x 1 1
H2k x = √ log for x ∈ − , . (1.2)
k 1 − 4x 2(1 − 4x) 4 4
k=0
It is well known that
2k
x X k x2k+1
∞
2 arcsin = for |x| 6 2;
2 (2k + 1)16k
k=0
in particular,
2k 2k
∞ ∞
X
k π X
k π
k
= and k
= √ .
(2k + 1)16 3 (2k + 1)8 2 2
k=0 k=0
The author [26, Theorem 1.1(ii)] determined
(p−3)/2 2k
p−1 2k

X X
k k
and
(2k + 1)16k (2k + 1)16k
k=0 k=(p+1)/2
modulo p2 for any prime p > 3. By [6], we have

∞ 2k 2k+1
x 3 k x 1
X X
arcsin =3 (1.3)
2 (2k + 1)16k (2j + 1)2
k=0 06j<k
and
(2)
x 4 3 X Hk−1 x2k
∞
arcsin = (1.4)
2 2 k2 2k

k k=1
for |x| 6 2. In particular,
2k

π3
∞
X
k(2) 1 (2)
H2k − Hk = ,
(2k + 1)16k 4 648
k=0
∞ 2k √ 3
X
k (2) 1 (2) 2π
H2k − Hk = .
(2k + 1)8k 4 384
k=0
and
(2)
π4
∞
X Hk−1
= .
2 2k 1944

k=1
k k
In view of (1.3), we have the following result.
Theorem 1.1. If |x| < 2, then
2k 2k
(arcsin(x/2))2
∞
k x

X (2) 1 (2)
√ = H2k − Hk . (1.5)
4 − x2 16k 4
k=1
SERIES WITH SUMMANDS INVOLVING HARMONIC NUMBERS 3
Proof. By taking derivatives of both sides of (1.3), we get

∞ 2k 2k X
x 2 1/2 k x 1
X
3 arcsin ×p =3 k
2 1 − (x/2)2 16 (2j + 1)2
k=0 06j<k
and hence
∞ 2k k Xk
(arcsin(x/2))2 X
k x 1
√ = k
,
4−x 2 16 (2j − 1)2
k=1 j=1
which is equivalent to (1.5).
Motivated by the Ramanujan series
∞ 2k 3
√
X
k 2 2
(6k + 1) = (1.6)
(−512)k π
k=0
(cf. [24]), L. Long [22] conjectured the congruence
(p−1)/2 2k 3 Xk
X 1 1
(6k + 1) k k − ≡ 0 (mod p) (1.7)
(−512) (2j − 1)2 16j 2
k=0 j=1
for any odd prime p, which was confirmed by H. Swisher [41] in 2015. Note
that (1.7) can be rewritten as
(p−1)/2 2k 3

X
k (2) 5 (2)
(6k + 1) H2k − Hk ≡ 0 (mod p).
(−512)k 16
k=0
In 2022 C. Wei [44] deduced the two identities
∞ 2k 3 √
X
k (2) 5 (2) 2
(6k + 1) k
H2k − Hk =− π
(−512) 16 48
k=0
and
∞ 2k 3
X (2) 5 (2) π
(6k + 1) k k H2k − Hk =
256 16 12
k=0
conjectured by V.J.W. Guo and X. Lian [18], as well as their q-analogues.
Motivated by Bauer’s series
∞ 2k 3
X 2
(4k + 1) k k = (1.8)
(−64) π
k=0
and Ramanujan’s series
∞ 2k 2 4k √
X
k 2k 2 3
(8k + 1) = , (1.9)
482k π
k=0
Wei and G. Ruan [45] proved the two new identities
2k3 X2k
(−1)j
∞
X π
(4k + 1) k k 2
=
(−64) j 12
k=1 j=1
4 ZHI-WEI SUN
and
∞ 2k2 4k k √
X
k 2k
X 1 1 3π
(8k + 1) − = ,
482k (2j − 1)2 36j 2 54
k=1 j=1
i.e.,
∞ 2k 3
X
k (2) 1 (2) π
(4k + 1) H2k − Hk =− (1.10)
(−64)k 2 12
k=0
and
∞ 2k 2 4k √
X
k 2k (2) 5 (2) 3π
(8k + 1) H2k − Hk = . (1.11)
482k 18 54
k=0
In 1997 van Hamme [42] thought that series for powers of π = Γ(1/2)2
should have their p-adic analogues involving the p-adic Gamma function
Γp (x), where p is an odd prime. Note that for any odd prime p we have
2
1 (p+1)/2 −1
Γp = (−1) =− ,
2 p
where ( p· ) denotes the Legendre symbol. However, van Hamme’s philosophy
fails for some Ramanujan-type series for 1/π. For example, T. Huber, D.
Schultz and D. Ye [21] used modular forms to obtain that
∞
X ak 16
(6k + 1) k = ,
16 π
k=0
where a0 = 1, a1 = 4, a2 = 20 and
(n+1)3 an+1 = 4(2n+1)(2n2 +2n+1)an −16n(4n2 +1)an−1 +8(2n−1)3 an−2
for all n = 2,
P3, . . .; but for a general odd prime p we even cannot find any
pattern for p−1k=0 (6k + 1)a k /16k modulo p.
The Bernoulli numbers B0 , B1 , B2 , . . . are defined by

xn
∞
x X
= B n (0 < |x| < 2π) .
ex − 1 n!
n=0
Equivalently,
n
X n+1
B0 = 1, and Bk = 0 for n = 1, 2, 3, . . . .
k
k=0
In 1900 J.W.L. Glaiser [12] proved that

p2 (2) 2
Hp−1 ≡ − Bp−3 (mod p3 ) and Hp−1 ≡ pBp−3 (mod p2 )
3 3
for any prime p > 3. The Bernoulli polynomials are given by
n
X n
Bn (x) = Bk xn−k (n ∈ N).
k
k=0
The Euler numbers E0 , E1 , E2 , . . . are defined by

xn
∞
2 X π
= En |x| < .
ex + e−x n=0 n! 2
Clearly E2n+1 = 0 for all n ∈ N. It is also known that

n
X 2n
E2k = 0 for each n ∈ N.
2k
k=0
The Euler polynomials are given by

n
1 n−k

X n Ek
En (x) = x− (n ∈ N).
k 2k 2
k=0
The author [27, 31] first observed that Ramanujan-type series have corre-
sponding congruences involving Bernoulli or Euler polynomials.
Now we introduce some notations throughout this paper. The Riemann
zeta function is defined by
∞
X 1
ζ(s) = with ℜ(s) > 1.
ns
n=1
The Dirichlet beta function is given by

(−1)k
∞
X
β(m) = (m = 1, 2, 3, . . .).
(2k + 1)m
k=0
Note that G = β(2) is the Catalan constant. We also adopt the notation
k
X ∞
−3 3
K = L 2, =
· k2
n=1
· ) the Kronecker symbol. For a prime p and an integer a 6≡ 0 (mod p),

with ( −3
we use qp (a) to denote the Fermat quotient (ap−1 − 1)/p. Many congruences
in later sections involve Fermat quotients.
In Sections 2–4, we will propose 66 new conjectures on series and related
congruences with summands involving not only harmonic numbers of order
at most three, but also products of several binomial coefficients. All the
conjectures have been checked via Mathematica.
2. Series with summands containing

one or two binomial coefficients
Conjecture 2.1 (2022-10-12). (i) We have
(−1)k−1
∞
X (2) 123 (2) 451 14 2
2k
H 2k−1 − H k−1 = 40 ζ(5) − 15 π ζ(3). (2.1)
16

3
k k
k=1
6 ZHI-WEI SUN
(ii) For any prime p > 5, we have

(p−1)/2
X (−1)k (2) (2)

16H 2k−1 − 123H k−1 ≡ −542Bp−5 (mod p) (2.2)
k3 2k

k=1 k
and
p−1
(−1)k 2k

X (2) (2)
Hp−1
p 16H2k − 123Hk ≡ 192 2 (mod p2 ). (2.3)
k2 k p
k=1
Remark 2.1. In 1979 R. Apéry [2] proved the irrationality of ζ(3) =

3
P ∞
n=1 1/n via the identity
(−1)k−1
∞
X 2
2k
= ζ(3).
3
k k 5
k=1
In 2014 the author [32] proved the congruence
(p−1)/2
X (−1)k−1
≡ 2Bp−3 (mod p)
k 3 2k
k=1 k
for any prime p > 5. The author’s conjectural identity (cf. [33])
(−1)k−1 2π 4
X∞
(H 2k + 4H k ) =
k3 2k 75

k=1 k
was proved by W. Chu [9] as well as K. C. Au [3, Prop. 7.14]. After seeing
an earlier arXiv version of this paper, Au [4, Corollary 2.9] confirmed the
author’s conjectural identity (2.1).
Conjecture 2.2 (2022-11-14). We have the identity
∞ 2k √ √
X
k (3) 1 (3) 35 2 2
k
H2k − Hk = ζ(3) − πG. (2.4)
8 8 64 8
k=0
Remark 2.2. Applying (1.1) and (1.2) with x = 1/8, we see that
∞ 2k ∞ 2k √
X
k
√ √ X
k log(3/2 + 2)
Hk = − 2 log(12 − 8 2) and H2k = √ .
8k 8k 2
k=0 k=0
In contrast with (2.4), we have
∞ 2k
π2

X
k (2) 1 (2)
H2k − Hk = √
8k 4 16 2
k=0
√
by applying (1.5) with x = 2.
∞ 2k
X
k (3) 1 (3) 2ζ(3) πK
H 2k − H = √ − . (2.5)
16k 8 k 3 3 8
k=0
Remark 2.3. Applying (1.1) and (1.2) with x = 1/8, we see that
∞ 2k ∞ 2k √
X
k 2 √ X
k log((7 + 4 3)/9)
Hk = − √ log(84 − 48 3) and H2k = √ .
16k 3 16k 3
k=0 k=0
In contrast with (2.5), we have

∞ 2k
π2

X
k (2) 1 (2)
H2k − Hk = √
16k 4 36 3
k=0
by applying (1.5) with x = 1.

π2
∞
X H3k − Hk 2 2
= (G + log 2) − (2.6)
k2k 3k 5 24

k=1 k
and
π2
∞
X H3k − Hk 11
= ζ(3) − log 2 − πG. (2.7)
2 2k 3k 4 24

k=1
k k
(ii) For any prime p > 5 with p ≡ 1 (mod 4), we have
p−1
X H3k − Hk 7
p 3k
≡ qp (2) (mod p). (2.8)
k
k2 k 10
k=1
Also, for each odd prime p we have
p−1
X H3k − Hk qp (2)
p2 3k
≡− (mod p). (2.9)
k 2 2k
k
4
k=1
Remark 2.4. The author’s conjectural identities

π2
∞
X H2k − Hk 3 2 π
= log 2 + log 2 −
k2k 3k 10 20 60

k=1 k
and
π2
∞
X H2k − Hk 33 πG
3k
= ζ(3) + log 2 −
32 24 2

k=1
k2 2k k
(cf. [39, Conjecture 10.61]) was confirmed by Au [3] in 2022.
∞
X 25k − 3
(H3k − 8H2k + 7Hk ) = 2G − 2(π + 9) log 2. (2.10)
k=1
2k 3k
k
(ii) For any odd prime p, we have the congruence

∞
X 25k − 3 −1 9
p2 3k
(H 3k − 8H 2k + 7H k ) ≡ − (mod p). (2.11)
p 4

2k
k=1 k
8 ZHI-WEI SUN
Remark 2.5. In 1974 R. W. Gosper announced the identity

∞
X 25k − 3 π
3k = ,
2k k 2
k=0
an elegant proof of which can be found in [1].

√ !
2k 3k √ √
∞
X
k 3 3
H 2k = (1 + 3) log 1 + − 3 log 2 (2.12)
27k 2 3
k=0
and
3k √
2k √ √
∞
X
k 1+ 3 log 3
H3k = 2 log(1 + 3) − − 3 log 2. (2.13)
27k 2 2
k=0

p−1
1 − ( p3 )
k
X 3k 2
≡ (mod p). (2.14)
k 27 3
k=(p+1)/2
Remark 2.6. For any positive integer n, we clearly have

n−1 n−1
XZ 1 Z 1 n−1 Z 1
X 1
k
X
k 1 − tn
Hn = = t dt = t dt = dt.
k+1 0 0 0 1−t
k=0 k=0 k=0
Using this trick we can deduce that

2k 3k √ √ √ √
∞
X
k 3
k
H k = (1 − 3) log 4 − (1 + 3) log 3 + 2 3 log(1 + 3).
27 4
k=0
Conjecture 2.7 (2023-05-28). We have

∞ √ !k
X 3k 3+ 5
(H3k − H2k ) = φ(log 3 − 2 log φ), (2.15)
k 54
k=0
√
where φ denotes the golden ratio (1 + 5)/2 ≈ 1.618 . . ..
Remark 2.7. Mathematica yields that
√
4x k
∞
X 3k 1 arcsin x
=√ cos
k 27 1−x 3
k=0
√
for any x ∈ (−1, 1). Applying this with x = ((1 + 5)/4)2 we obtain that
√ !k
q √
∞
X 3k
3+ 5 cos(π/10) (5 + 5)/8
=q √ =q √ = φ.
k 54
k=0 (5 − 5)/8 (5 − 5)/8
√
Conjecture 2.8 (2023-05-30). If (1 − 2)/2 < x < 1/2, then
√
x(1 − x) k
∞
X 4k 1−x
(2H4k − 3H2k + Hk ) = log(1 − x). (2.16)
2k 4 2x − 1
k=0
Remark 2.8. For any x ∈ (−1, 1), we have
∞
s √
X 4k x k 1+ 1−x
=
2k 16 2(1 − x)
k=0
which can be proved directly or via Mathematica. In particular,
3 k √
∞
X 4k
= 3.
2k 64
k=0
For x ∈ (−1, 1), we obviously have
∞ ∞
X 2k x k x k X 4k x 2k
Hk + − =2 H2k
k 4 4 2k 4
k=0 k=0
and
∞ ∞
X 2k x k x k X 4k x 2k
H2k + − =2 H4k ,
k 4 4 2k 4
k=0 k=0
and hence we may find closed formulas for the two series
∞ ∞
X 4k x 2k X 2k x 2k
H2k and H4k
2k 4 k 4
k=0 k=0
by using (1.1) and (1.2). In particular,
√
3 k √ √ √
∞
X 4k 3−1
H2k = 2(1+ 3) log(1+ 3)−(1+3 3) log 2+ log 3
2k 64 2
k=0
and
√
3 k √ √ √
∞
X 4k 3−1
H4k = (2 + 3) log(1 + 3) − (1 + 3) log 2 + log 3.
2k 64 4
k=0
10 ZHI-WEI SUN
With the aid of Mathematica, we obtain that
3 k 3 k 1 1 − tk
∞ ∞ Z
X 4k X 4k
Hk = dt
2k 64 2k 64 0 1−t
k=0 k=1
3 k
Z 1 ∞
1 X 4k
= (1 − tk )dt
0 1 − t 2k 64
k=1
p √ !
Z 1
1 √ 2 + 4 − 3t
= 3− √ dt
0 1−t 4 − 3t
√ √
2+ 3 √ 7+4 3
= log + 3 log
3
√ √ 8 √
= (2 + 4 3) log(1 + 3) − log 3 − (1 + 5 3) log 2.

2k 2
(log 2)Γ(1/4)2
∞
X
k
(2H 2k − H k ) = − √ (2.17)
(−16)k 4π 2π
k=0
and
2k 2
(log 2)Γ(1/4)2
∞
X
k √
(2H2k − Hk ) = , (2.18)
32k 4π π
k=0
where Γ(x) is the well-known Gamma function.

Remark 2.9. Mathematica yields the identities
2k 2 ∞ 2k2
Γ(1/4)2 Γ(1/4)2
X∞ X
k √ k √ .
= and =
(−16)k 2π 2π 32k 2π π
k=0 k=0
For any prime p ≡ 3 (mod 4), the author [30] proved that
p−1 2k 2
X (−1)(p−3)/4 2p
k
≡ (mod p2 ),
(−16)k (p+1)/2
k=0 (p+1)/4
and Z.-H. Sun [25] confirmed the author’s conjecture (cf. [26, Conjecture
5.5])
p−1 2k 2
X
k
≡ 0 (mod p2 ).
32k
k=0
For any prime p ≡ 1 (mod 4) with p = x2 +y 2 (x, y ∈ Z and x ≡ 1 (mod 4)),

Z.-H. Sun [25] confirmed the author’s conjecture (cf. [26, Conjecture 5.5])
p−1 2k 2 p−1 2k 2
(p−1)/4
X X p
(−1) k
≡ k
≡ 2x − (mod p2 ).
(−16)k 32k 2x
k=0 k=0

∞ 2k 2 2 2
X
k (2) 1 (2) 1 π − 8G
H2k − H =Γ √ . (2.19)
32k 4 k 4 32π π
k=0
Remark 2.10. In contrast with (2.19), Mathematica yields that

∞ 2k2
X
k Γ(1/4) Γ(1/4)2
= √ = √ .
32k 2πΓ(3/4) 2π π
k=0

∞ 2k 3k
X
k k (3 log 2)Γ(1/3)3
(3H 3k − H k ) = √ . (2.20)
54k 4π 2 3 2
k=0
Remark 2.11. Mathematica yields the identity

∞ 2k 3k
X
k k 3Γ(1/3)3
= √ .
54k 4π 2 3 2
k=0
By [30, Corollary 1.3], for any prime p > 3 with p ≡ 2 (mod 3), we have
p−1 2k 3k
X
k k
≡ 0 (mod p2 ).
54k
k=0

∞ 2k 3k ∞ 2k 3k
X
k k 8 X k k
(3H3k − Hk ) = log . (2.21)
(−216)k 9 (−216)k
k=0 k=0
Remark 2.12. For any prime p > 3, we have

p−1 2k 3k
X p−1 2k 3k
p X
k k
≡ k k
(mod p2 )
(−216)k 3 24k
k=0 k=0
by [30, Corollary 1.4], and

p−1 2k 3k
( (2p−2)/3
X
k k (p−1)/3
(mod p2 ) if p ≡ 1 (mod 3),
k
≡ (2p+2)/3 2
24 p/ (p+1)/3 (mod p ) if p ≡ 2 (mod 3),
k=0
as conjectured by the author [27, Conjecture 5.13] and proved by C. Wang

and Sun [43, Theorem 1.2].
∞ 4k 2k √
X
2k k (log 2) π
(2H 4k − H 2k ) = . (2.22)
128k 2Γ(5/8)Γ(7/8)
k=0
12 ZHI-WEI SUN

∞ 4k 2k √
X
2k k π
k
= .
128 Γ(5/8)Γ(7/8)
k=0
By [30, Corollary 1.3], for any prime p ≡ 5, 7 (mod 8) we have

p−1 4k 2k
X
2k k
≡ 0 (mod p2 ).
128k
k=0

X∞ 4k 2k ∞
X 4k 2k
2k k 2k k
(2H4k − H2k ) = (log 3) (2.23)
72k 72k
k=0 k=0
and
4k 2k 4k 2k
∞ ∞
X
2k k 1 9 X 2k k
(2H4k − H2k ) = log . (2.24)
576k 2 8 576k
k=0 k=0
Remark 2.14. Let p > 3 be a prime. By [30, Corollary 1.4],

p−1 4k 2k X p−1 4k 2k
X −2
2k k
≡ 2k k
(mod p2 ).
576k p 72k
k=0 k=0
We also have
X p−1 4k 2k
6 2k k
p 72k
 k=0
2x − p (mod p2 ) if p = x2 + y 2 (x, y ∈ Z & 4 | x − 1),
2x
≡ 2p 2
 3((p+1)/2) (mod p ) if p ≡ 3 (mod 4),
(p+1)/4
as conjectured by the author [27, Conjecture 5.14(iii)] and proved by Wang

and Sun [43, Theorem 5.2 and Remark 5.2].
∞ 4k 2k ∞ 4k 2k
X
2k k 1 3 X 2k k
(2H4k − H2k ) = log . (2.25)
(−192)k 2 4 (−192)k
k=0 k=0
Remark 2.15. Let p > 3 be a prime. By [30, Corollary 1.4], we have

p−1 4k 2k X p−1 4k 2k
X −2
2k k
≡ 2k k
(mod p2 ).
(−192)k p 48k
k=0 k=0
If p = x2 + 3y 2 with x, y ∈ Z and x ≡ 1 (mod 3), then

p−1 4k 2k
X p
2k k
≡ 2x − (mod p2 ),
48k 2x
k=0
as conjectured by the author [27, Conjecture 5.14] and confirmed by G.-S.

Mao and H. Pan [23]. If p ≡ 2 (mod 3), then
p−1 4k 2k
X 3p
2k
k
k
≡ (p+1)/2 (mod p2 )
48 2
k=0 (p+1)/6
as conjectured by the author [27, Conjecture 5.14] and confirmed by Wang

and Sun [43].
∞ 4k 2k ∞ 4k 2k
X
2k k 1 63 X 2k k
(2H4k − H2k ) = log . (2.26)
(−4032)k 2 64 (−4032)k
k=0 k=0
Remark 2.16. Let p > 3 be a prime with p 6= 7. By [30, Corollary 1.4],

p−1 4k 2k X p−1 4k 2k
X −2
2k k
≡ 2k k
(mod p2 ).
(−4032)k p 63k
k=0 k=0
The author [27, Conjecture 5.14(ii)] conjectured that

p−1 4k 2k
X
2k k
63k
k=0
(
p
( p3 )(2x − 2x ) (mod p2 ) if p = x2 + 7y 2 with x, y ∈ Z and ( x7 ) = 1,
≡
0 (mod p) if ( p7 ) = −1.

∞ 6k 3k √
X
3k k (log 2) π
(6H 6k − 3H 3k − 2H 2k + H k ) = . (2.27)
864k Γ(7/12)Γ(11/12)
k=0

∞ 6k 3k √
X
3k k π
k
= .
864 Γ(7/12)Γ(11/12)
k=0
By [30, Corollary 1.3], for any prime p > 3 with p ≡ 3 (mod 4) we have
p−1 6k 3k
X
3k k
≡ 0 (mod p2 ).
864k
k=0
3. Series and congruences with summands

containing 3 or 4 binomial coefficients
In 1993 D. Zeilberger [47] used the WZ method to establish the identity
π2
∞
X 21k − 8
3 = .
k 3 2k 6
k=1 k
14 ZHI-WEI SUN
The author [34] proved that

(p−1)/2
X 21k − 8 −1
≡ − 4Ep−3 (mod p)
2k 3 p

k 3
k=1 k
for any prime p > 3.
Conjecture 3.1 (2022-10-11). (i) We have the identity
47π 4
∞
X 21k − 8 (2) 25 (2)
H 2k−1 − H = . (3.1)
3 2k
3 8 k−1 2880
k=1 k k
(p−1)/2 3
X 2k (2) 25 (2) −1
(21k + 8) H2k − Hk ≡ 32p Ep−3 (mod p2 )
k 8 p
k=1
(3.2)
and
p−1 3
X 2k (2) 25 (2) 246 4
(21k + 8) H2k − Hk ≡ −48Hp−1 + p Bp−5 (mod p5 ).
k 8 5
k=0
(3.3)
(iii) For each prime p > 5, we have
p−1 3
X 2k
(2(21k + 8)(H2k − Hk ) + 7) ≡ −112pHp−1 (mod p5 ). (3.4)
k
k=1
Remark 3.1. After seeing an earlier arXiv version of this paper, K. C. Au [4,
Corollary 2.3] confirmed the author’s conjectural identity (3.1), and proved
an identity (after the proof of [4, Theorem 2.2]) which has the equivalent
form
∞
X (21k − 8)(H2k−1 − Hk−1 ) − 7/2
3 = ζ(3).
k=1 k3 2k
k
∞
X 21k − 8 (3) 43 (3) 711 29
3 H2k−1 + Hk−1 = ζ(5) − π 2 ζ(3). (3.5)
3 2k
8 28 14
k=1 k k
(p−1)/2 3
X 2k (3) 43 (3) −1
(21k + 8) H2k + Hk ≡ 32 Ep−3 (mod p) (3.6)
k 8 p
k=0
and
p−1 3
X 2k (3) 43 (3) 120
(21k + 8) H2k + Hk ≡− pBp−3 (mod p2 ). (3.7)
k 8 7
k=0
Remark 3.2. The identity (3.5) looks quite challenging.


(3k − 1)(−8)k
∞
X (2) 5 (2)
3 H2k−1 − Hk−1 = −2β(4). (3.8)
k3 2k 4
k=1 k
(p−1)/2 2k 3
X (2) 5 (2) 2 p 1
(3k +1) k
H2k − Hk ≡ Ep−3 (mod p2 ) (3.9)
(−8)k 4 p 4 4
k=0
and
p−1 2k 3

X (2) 5 (2)
(3k + 1) k
H2k − Hk ≡ pEp−3 (mod p2 ). (3.10)
(−8)k 4
k=0
(iii) Let p be an odd prime. Then

(p−1)/2 2k 3
X −1
k
k
(2(3k + 1)(H2k − Hk ) + 1) ≡ 2p−1 (mod p2 ) (3.11)
(−8) p
k=0
and
p−1 2k 3
X
k (3) 7 (3)
(3k + 1) H2k + Hk ≡ 0 (mod p). (3.12)
(−8)k 8
k=0
Remark 3.3. In 2008, J. Guillera [14] used the WZ method to find the
identity
(3k − 1)(−8)k
X∞
3 = −2G.
k=1 k3 2k
k
The identity (3.8) provides a fast converging series for computing the con-
stant β(4). We are unable to find the exact values of the series
(−8)k
X∞
3 (2(3k − 1)(H2k−1 − Hk−1 ) − 1)

k 3 2k
k=1 k
and
(3k − 1)(−8)k
∞
X (3) 7 (3)
3 H2k−1 + Hk−1 .
k3 2k 8
k=1 k

(3k − 1)16k π4
∞
X (2) 5 (2)
H 2k−1 − H = . (3.13)
k3 2k
3 4 k−1 24
k=1 k
(ii) Let p > 3 be a prime. Then
(p−1)/2 2k 3
X (2) 5 (2) −1
(3k + 1) k k H2k − Hk ≡ 2p Ep−3 (mod p2 ) (3.14)
16 4 p
k=0
16 ZHI-WEI SUN
and
p−1 2k 3
X (2) 5 (2) 7 2
(3k + 1) k
H2k − Hk ≡ p Bp−3 (mod p3 ). (3.15)
16k 4 6
k=0
(iii) Let p be an odd prime. Then

p−1 2k 3
X 4 2
k
(2(3k + 1)(H2k − Hk ) + 1) ≡ p qp (2) − p2 qp (2)2 (mod p3 ).
16k 3 3
k=1
(3.16)
If p ≡ 2 (mod 3), then
(p−1)/2 2k 3
X
k (2) 5 (2)
H2k − Hk ≡ 0 (mod p). (3.17)
16k 4
k=0
Remark 3.4. In 2008 J. Guillera [14, Identity 1] used the WZ method to

establish the identity
(3k − 1)16k π2
X∞
3 = .
k3 2k
2
k=1 k
Two q-analogues of this identity were given by Q.-H. Hou, C. Krattenthaler
and Z.-W. Sun [20]. Guo and Lian [18] proved that the two sides of (3.15)
are congruent modulo p2 for any prime p > 3. After seeing an earlier arXiv
version of this paper, Au [4] confirmed the author’s conjectural identity
(3.13), and proved an identity (after the proof of [4, Corollary 2.3]) which
has the equation form
16k π2
∞
X 1 7
3 (3k − 1)(H 2k−1 − H k−1 ) − = log 2 + ζ(3).
3 2k
2 3 6
k=1 k k
(3k − 1)16k π2
∞
X (3) 7 (3)
H 2k−1 + H = ζ(3). (3.18)
k 3 2k
3 8 k−1 2
k=1 k
(ii) For any odd prime p, we have

(p−1)/2 2k 3

X
k (3) 7 (3) −1
(3k + 1) H2k + Hk ≡2 Ep−3 (mod p) (3.19)
16k 8 p
k=1
and
p−1 2k 3
X (3) 7 (3) 3
(3k + 1) k
H2k + Hk ≡ pBp−3 (mod p2 ). (3.20)
16k 8 2
k=0
Remark 3.5. Conjecture 3.5 looks more challenging than Conjecture 3.4.
(4k − 1)(−64)k
∞
X (2) 1 (2)
3 H2k−1 − Hk−1 = −16β(4). (3.21)
k3 2k 2
k=1 k

(p−1)/2 2k 3
X (2) 1 (2)
(4k + 1) k k H2k − Hk ≡ pEp−3 (mod p2 ), (3.22)
(−64) 2
k=0
p−1 2k 3
X (2) 1 (2)
(4k + 1) k k H2k − Hk ≡ pEp−3 (mod p2 ). (3.23)
(−64) 2
k=0
Remark 3.6. J. Guillera [14] proved that
(4k − 1)(−64)k
X∞
3 = −16G.
k=1 k3 2k
k
The congruence (3.22) is motivated by (1.10). After seeing an earlier arXiv

version of this paper, Au [4, Corollary 2.11] confirmed the author’s conjec-
tural identity (3.21). It seems that for any m, n ∈ Z+ and odd prime p, we
have
(p−1)/2 2k n (2m)
!
X (2m) H
(4k + 1) k kn H2k − 2m−1 k
≡ 0 (mod p). (3.24)
(−4) 2
k=0
8k ((10k − 3)(H2k−1 − Hk−1 ) − 1)

∞
X 7
2 3k = ζ(3) (3.25)
2k
k3 k 2
k=1 k
and
8k ((10k − 3)(H3k−1 − Hk−1 ) − 8/3) 2π 2 log 2 + 7ζ(3)
X∞
2 3k = . (3.26)
k3 2k 4
k=1 k k

p−1 2k 2 3k
X 63
k k
((10k + 3)(H2k − Hk ) + 1) ≡ p3 Bp−3 (mod p4 ) (3.27)
8k 8
k=1
and
p−1 2k 2 3k
X 9
k k
(3(10k + 3)(H3k − Hk ) + 8) ≡ 9p qp (2) − p2 qp (2) (mod p3 ).
8k 2
k=1
(3.28)
18 ZHI-WEI SUN
Remark 3.7. As conjectured by the author [27] and confirmed by J. Guillera

and M. Rogers [17], we have
(10k − 3)8k π2
X∞
2 = .
k 3 2k
3k 2
k=1 k k

(−27)k ((15k − 4)(3H3k−1 − Hk−1 ) − 9) 4π 3
X∞
2 3k = − √ . (3.29)
3 2k 3

k=1 k k k

p−1 2k2 3k p
X
2 1
k k
k
((15k + 4)(3H3k − Hk ) + 9) ≡ 9 + 6p Bp−2 (mod p3 ).
(−27) 3 3
k=0
(3.30)
Remark 3.8. As conjectured by the author [27] and confirmed by Kh.
Hessami Pilehrood and T. Hessami Pilehrood [19], we have
(15k − 4)(−27)k−1
X∞
2 3k = K.
k=1 k3 2k
k k

64k−1 ((11k − 3)(2H2k−1 + Hk−1 ) − 4)
∞
X 7
2 3k = ζ(3) (3.31)
k3 k2k 2
k=1 k
and
64k−1 ((11k − 3)(3H3k−1 − 6Hk−1 ) − 7) 6π 2 log 2 − 21ζ(3)
X∞
2 3k = . (3.32)
k3 2k
8
k=1 k k

Xp−1 2k 2 3k
k k
((11k + 3)(2H2k + Hk ) + 4) ≡ 21p3 Bp−3 (mod p4 ) (3.33)
64k
k=1
and
p−1
X 2k2 3k
k k
((11k + 3)(3H3k − 6Hk ) + 7) ≡ 18p qp (2) + 9p2 qp (2)2 (mod p3 ).
64k
k=1
(3.34)
Remark 3.9. As conjectured by the author [27] and confirmed by J. Guillera
[16], we have
(11k − 3)64k
X∞
2
2k 2 3k
= 8π .
3
k=1 k k k

81k ((35k − 8)(H4k−1 − Hk−1 ) − 35/4)
X∞
= 12π 2 log 3 + 39ζ(3). (3.35)
3 2k 2 4k
k=1 k k 2k
p−1 2k 2 4k
X
k 2k
(4(35k + 8)(H4k − Hk ) + 35)
81k (3.36)
k=1
p−1 p−1 2 3
≡ 32(3 − 1) − 16(3 − 1) (mod p ).
Remark 3.10. As conjectured by the author [27] and confirmed in [17], we
have
(35k − 8)81k
X∞
= 12π 2 .
3 2k2 4k
k=1 k k 2k
Conjecture 3.11 (2023-01-14). For any prime p > 3, we have

p−1 2k 2 4k
X
k 2k
(4(5k + 1)(H4k − Hk ) + 5)
(−144)k (3.37)
k=1
p
≡ (5 + 2p(2qp (2) + qp (3))) (mod p2 ).
3
Remark 3.11. As conjectured by the author [27] and confirmed in [17], we
have
(5k − 1)(−144)k
∞
X 45
2 = − K.
k3 2k
4k 2
k=1 k 2k
We are unable to find the exact value of the series
(−144)k
X∞
2 4k (4(5k − 1)(H4k−1 − Hk−1 ) − 5).

k 3 2k
k=1 k 2k
The classical rational Ramanujan-type series for 1/π have the following
four forms:
∞ 2k 3
√
X c d
(ak + b) k k = , (3.38)
m π
k=0
∞ 2k 2 3k √
X c d
(ak + b) k k k = , (3.39)
m π
k=0
∞ 2k 2 4k √
X c d
(ak + b) k k 2k = , (3.40)
m π
k=0
∞ 2k 3k 6k
√
X
k k 3k c d
(ak + b) = , (3.41)
mk π
k=0
20 ZHI-WEI SUN
where a, b, m ∈ Z, am 6= 0, c ∈ Q \ {0}, and d is a positive squarefree

integer. It is known that there are totally 36 such series, see, e.g., S. Cooper
[11, Chapter 14].
For a positive integer m, can we find similar series for (log m)/π? Moti-
vated by Ramanujan-type series of the forms (3.38)-(3.41), the author for-
mulated the following general conjecture.
Conjecture 3.12 (General Conjecture, 2022-12-08). (i) If we have an iden-
tity (3.38) with a, b, m ∈ Z, am 6= 0, c ∈ Q \ {0}, and d ∈ Z+ squarefree,
then
∞ 2k 3 √
X
k c d
(6(ak + b)(H2k − Hk ) + a) = log |m|, (3.42)
mk π
k=0
and
p−1 2k 3
X −d
k
(6(ak + b)(H2k − Hk ) + a) ≡ (a + b(mp−1 − 1)) (mod p2 )
mk p
k=0
(3.43)
for any prime p ∤ dm.
(ii) If we have an identity (3.39) with a, b, m ∈ Z, am 6= 0, c ∈ Q \ {0},
and d ∈ Z+ squarefeee, then
∞ 2k 2 3k √
X
k k c d
((ak + b)(3H3k + 2H2k − 5Hk ) + a) = log |m|, (3.44)
mk π
k=0
and
p−1
X 2k 2 3k
k k
((ak + b)(3H3k + 2H2k − 5Hk ) + a)
mk
k=0
(3.45)
−d
≡ (a + b(mp−1 − 1)) (mod p2 )
p
for any odd prime p ∤ dm.
(iii) If we have an identity (3.40) with a, b, m ∈ Z, am 6= 0, c ∈ Q \ {0},
and d ∈ Z+ squarefree, then
∞ 2k 2 4k √
X
k 2k c d
(4(ak + b)(H4k − Hk ) + a) = log |m|, (3.46)
mk π
k=0
and
p−1
X 2k 2 4k
k 2k
(4(ak + b)(H4k − Hk ) + a)
mk
k=0
(3.47)
−d
a + b(mp−1 − 1) (mod p2 )

≡
p
(iv) If we have an identity (3.41) with a, b, m ∈ Z, am 6= 0, c ∈ Q \ {0},

and d ∈ Z+ squarefree, then
∞ 2k 3k 6k √
X
k k 3k c d
(3(ak + b)(2H6k − H3k − Hk ) + a) = log |m|, (3.48)
mk π
k=0
and
p−1
X 2k 3k 6k
k k 3k
(3(ak + b)(2H6k − H3k − Hk ) + a)
mk
k=0
(3.49)
−d
a + b(mp−1 − 1) (mod p2 )

≡
p
Remark 3.12. Ramanujan [24] found the irrational series
∞ 2k 3
X 31 16
k+ √ √k = √ .
270 + 48 5 (220 /( 5 − 1)8 )k (15 + 21 5)π
k=0
In the spirit of part (i) of our general conjecture, we guess that

2k 3
∞
X
k 31
√ 6 k+ √ (H2k − Hk ) + 1
k=0
(220 /( 5 − 1)8 )k 270 + 48 5
16 220
= √ × log √ ,
(15 + 21 5)π ( 5 − 1)8
which can be easily checked via Mathematica.
Example 3.1. In view of the Ramanujan series (cf. [11, Chapter 14] and
[24])
∞ 2k 2 3k
√
X
k k 4 3
(5k + 1) =
(−192)k π
k=0
and the known identity
∞ 2k2 3k √
X
k k 4 π
= √
(−192)k 3 3 Γ(5/6)3
k=0
(cf. (14.29) of [11, p. 624]), by part (ii) of Conjecture 3.12 we should have
∞ 2k 2 3k √
X
k k 4 log 192 20 π
(5k + 1)(3H3k + 2H2k − 5Hk ) = √ − √ .
(−192)k 3π 3 3 Γ(5/6)3
k=0

∞ 2k 3
X (3) 15ζ(3)
(4k + 1) k k H2k = − 2G. (3.50)
(−64) 4π
k=0
22 ZHI-WEI SUN
Remark 3.13. For any m, n ∈ Z+ and odd prime p not dividing 22m−1 − 1,
(2m−1)
we have Hp−1 ≡ 0 (mod p) since
p−1 p−1
X 1 X 1
≡ (mod p),
(2j)2m−1 k2m−1
j=1 k=1
thus
(p−1)/2
(p − 1)/2 n (2m−1)
X
(4k + 1) H2k
k
k=0
(p−1)/2
(p − 1)/2 n (2m−1)

X p−1
= 4 −k +1 Hp−1−2k
2 k
k=0
(p−1)/2 n
X (p − 1)/2 (2m−1)
≡ − (4k + 1) H2k (mod p)
k
k=0
and hence
(p−1)/2 2kn
(2m−1)
X
(4k + 1) k kn H2k ≡ 0 (mod p). (3.51)
(−4)
k=0
−1/2 2k k
(Note that k = k /(−4) for any k ∈ N.)
∞ 2k 3 √ √
X
k 4 32 π 8 log 2
(6k + 1) Hk = · − (3.52)
256k 3 Γ(5/6)3 π
k=0
and
∞ 2k 3
√ √
X
k 2 32 π 8 log 2
(6k + 1) H2k = · − . (3.53)
256k 3 Γ(5/6)3 3π
k=0
(ii) Let p be an odd prime. Then
(p−1)/2 2k 3
X
k
((6k + 1)(3H2k − Hk ) − 1) ≡ (−1)(p+1)/2 (mod p4 ). (3.54)
256k
k=0
If p > 3, then
(p−1)/2
X 2k 3
k
((6k + 1)(H2k − Hk ) + 1)
256k

k=0
(3.55)
−1 4 2
≡ 1 + p qp (2) − p2 qp (2)2 (mod p3 ).
p 3 3
Remark 3.14. It is known (cf. (14.27) of [11, p. 623]) that
∞ 2k 3 √ √
X
k 2 32 π
= · .
256k 3 Γ(5/6)3
k=0
In view of this, (3.52) and (3.53) together implies the identities

X∞ 2k 3
k
((6k + 1)(3H2k − Hk ) − 1) = 0
256k
k=0
and
2k 3
∞
X
k 16 log 2
((6k + 1)(H2k − Hk ) + 1) = .
256k 3π
k=0
The last identity is also implied by Conjecture 3.12(i).
Conjecture 3.15 (2022-10-12). Let p be any odd prime.
(i) We have
(p−1)/2 2k 3

X (2) 5 (2) 7 −1
(6k + 1) k
H2k − Hk ≡ p2 Bp−3 (mod p3 ).
256k 16 24 p
k=0
(3.56)
If p > 3, then
p−1 2k 3
X (2) 5 (2)
(6k + 1) k k H2k − Hk ≡ −pEp−3 (mod p2 ). (3.57)
256 16
k=0
(ii) If p ≡ 2 (mod 3), then

(p−1)/2 2k 3
X
k (2) 5 (2)
H2k − Hk ≡ 0 (mod p). (3.58)
256k 16
k=0
Remark 3.15. For any prime p > 3, Guo and Lian [18] proved that the two
sides of (3.56) are congruent modulo p2 .
∞ 2k 3
X
k (3) 7 (3) 25ζ(3)
(6k + 1) k
H2k − Hk = − G. (3.59)
256 64 8π
k=0

(p−1)/2 2k 3
X
k (3) 7 (3) 1
(6k + 1) k
H2k − Hk ≡ − Ep−3 (mod p) (3.60)
256 64 2
k=0
and
p−1 2k 3
X (3) 7 (3) 3
(6k + 1) k k H2k − Hk ≡ − Ep−3 (mod p). (3.61)
256 64 2
k=0
Remark 3.16. For any k ∈ Z+ , it is easy to see that

k
(3) 7 (3) X 1 1
H2k − Hk = + .
64 (2j − 1)3 (4j)3
j=1
24 ZHI-WEI SUN
Conjecture 3.17 (2022-10-12). Let p > 3 be a prime. Then

(p−1)/2 2k 3

X (2) 5 (2) 2 p
(6k + 1) k
k
H2k − Hk ≡ Ep−3 (mod p2 )
(−512) 16 p 4
k=0
(3.62)
and
p−1 2k 3

X (2) 5 (2) p 1
(6k + 1) k
k
H2k − Hk ≡ Ep−3 (mod p2 ).
(−512) 16 16 4
k=0
(3.63)
Remark 3.17. Note that (3.62) is stronger than (1.7).
∞ 2k3
X
k (3) 7 (3) 57 ζ(3)
(6k + 1) k
H2k − Hk = ·√ − L, (3.64)
(−512) 64 16 2π
k=0
where
(−1)k(k−1)/2
∞ ∞
( −8
n )
X
−8 X
L = L 2, = =
· n=1
n2 (2k + 1)2
k=0
with ( −8
· )the Kronecker symbol.
p−1 2k 3
X
k (3) 7 (3)
(6k + 1) H2k − Hk ≡ 0 (mod p). (3.65)
(−512)k 64
k=0
Remark 3.18. For any odd prime p, we are unable to find a closed form
for the left-hand side of the congruence (3.65) modulo p2 .
(p−1)/2 2k 3
X
k
((42k + 5)(H2k − Hk ) + 7)
4096k
k=0
(3.66)
−1
7 + 10p qp (2) − 5p2 qp (2)2 (mod p3 ).

≡
p
Remark 3.19. In view of the Ramanujan series (cf. [24])
2k 3
∞
X
k 16
(42k + 5) k
= ,
4096 π
k=0
by part (i) of Conjecture 3.12 we should have

∞ 2k 3
X
k 32 log 2
((42k + 5)(H2k − Hk ) + 7) = ,
4096k π
k=0
and
p−1
X 2k 3
k
(6(42k + 5)(H2k − Hk ) + 42)
4096k
k=0

−1 12 −1
(42 + 60p qp (2)) (mod p2 )

≡ 42 + 5p qp (2 ) ≡
p p
for any odd prime p.
2k 3
∞
X
k (2) 25 (2) 2π
(42k + 5) k
H2k − Hk = . (3.67)
4096 92 69
k=0
(ii) Let p > 3 be a prime. If p 6= 23, then
(p−1)/2 2k 3

X
k (2) 25 (2)
(42k + 5) H2k − Hk
4096k 92
k=0
(3.68)
−1 3 4
p Bp−5 − 5Hp−1 (mod p5 ).

≡
p 20
Also,
p−1 2k 3

X (2) 25 (2)
(42k + 5) k
H2k − Hk ≡ −pEp−3 (mod p2 ). (3.69)
4096k 92
k=0
Remark 3.20. It is interesting to compare this conjecture with Remark

3.19.
∞ 2k 3
X
k (3) 43 (3) 555 ζ(3) 32
(42k + 5) k
H2k − Hk = · − G. (3.70)
4096 352 77 π 11
k=0
(p−1)/2 2k 3
X
k (3) 43 (3)
(42k + 5) 11H2k − Hk ≡ −16Ep−3 (mod p) (3.71)
4096k 32
k=0
and
p−1 2k 3

X
k (3) 43 (3)
(42k + 5) 11H2k − Hk ≡ −27Ep−3 (mod p). (3.72)
4096k 32
k=0
Remark 3.21. Conjecture 3.21 looks quite challenging.

∞ 2k 2 3k √
X
k k 9 3 log 3
((6k + 1)(H2k − 2Hk ) + 3) = (3.73)
216k 2π
k=0
26 ZHI-WEI SUN
and
∞ 2k2 3k √
X
k k 9 3 log 2
(6k + 1)(3H3k − Hk ) = . (3.74)
216k π
k=0
p−1 2k 2 3k
X p 3p + 3
k k
((6k + 1)(H 2k − 2H k ) + 3) ≡ (mod p2 ) (3.75)
216k 3 2
k=0
and
p−1
X 2k 2 3k p
k k
(6k + 1)(3H3k − Hk ) ≡ 3 p qp (2) (mod p2 ). (3.76)
216k 3
k=0
Remark 3.22. This is motivated by the Ramanujan series (cf. [11, Chapter
14] and [24])
∞ 2k 2 3k
√
X
k k 3 3
(6k + 1) = .
216k π
k=0

2k 2 4k
∞
X
k 2k 56 log 2
((20k + 3)(H2k − 3Hk ) + 12) = . (3.77)
(−1024)k π
k=0

p−1 2k 2 4k
X −1
k 2k
k
((20k + 3)(H2k − 3Hk ) + 12) ≡ (12+21p qp (2)) (mod p2 ).
(−1024) p
k=0
(3.78)
Remark 3.23. This is motivated by the Ramanujan series (cf. [11, Chapter
14] and [24])
∞ 2k2 4k
X 8
(20k + 3) k 2k
= .
(−1024)k π
k=0

∞ 2k 2 4k √
X
k 2k 16 3 log 2
((8k + 1)(3H2k − 4Hk ) + 6) = . (3.79)
482k π
k=0

p−1 2k2 4k
X p
k 2k
((8k + 1)(3H 2k − 4H k ) + 6) ≡ (6 + 8p qp (2)) (mod p2 ).
482k 3
k=0
(3.80)
Remark 3.24. This is motivated by the Ramanujan series (1.9).

(p−1)/2 2k 2 4k
X (2) 5 (2) p 1
(8k + 1) k
2k
2k
H2k − Hk ≡ Bp−2 (mod p2 ),
48 18 36 3
k=0
(3.81)
p−1 2k 2 4k
X (2) 5 (2) 5 1
(8k + 1) k 2k2k H2k − Hk ≡ − pBp−2 (mod p2 ).
48 18 24 3
k=0
(3.82)
Remark 3.25. The congruence (3.81) is motivated by (1.11).
Conjecture 3.26 (2022-10-19). (i) For any prime p > 3, we have
(p−1)/2 2k 4
X
k (2) 1 (2)
(4k + 1) H2k − Hk
256k 2
k=0
(3.83)
p−1 2k 4
X (2) 1 (2) 7
≡ (4k + 1) k
k
H2k − Hk ≡ p2 Bp−3 (mod p3 ).
256 2 6
k=0

(p−1)/2 2k 4 p−1 2k 4
X (3)
X (3) 3
(4k + 1) k
H ≡ (4k + 1) k k H2k ≡ pBp−3 (mod p2 ).
256k 2k 256 2
k=0 k=0
(3.84)
Remark 3.26. Guo and Lian [18, (1.7)] proved that for any prime p > 3
we have
(p−1)/2 2k4
X (2) 1 (2)
(4k + 1) k
H2k − Hk ≡ 0 (mod p2 ).
256k 2
k=0
4. Series and congruences with summands

containing at least five binomial coefficients
The following two conjectures are motivated by the identity
(−1)k (205k2 − 160k + 32)
X∞
5 = −2ζ(3)
k=1 k5 2k
k
established by T. Amdeberhan and D. Zeilberger [5] in 1997 via the WZ
method.
(−1)k−1 π4
X∞
2
5 (205k − 160k + 32)(H 2k−1 − H k−1 ) − 41k + 16 = .
5 2k 60
k=1 k k
(4.1)
28 ZHI-WEI SUN

p−1 5
k 2k
X
(205k2 + 160k + 32)(H2k − Hk ) + 41k + 16

(−1)
k (4.2)
k=0
≡ 16p + 64p2 Hp−1 (mod p6 ).
(2) (2)
X∞
(−1)k ((205k2 − 160k + 32)(4H2k−1 − 12Hk−1 ) − 43)
5 = −8ζ(5). (4.3)
k=1 k5 2k
k
(p−1)/2 (2) (2)
X (−1)k ((205k2 − 160k + 32)(4H2k−1 − 12Hk−1 ) − 43)
5 (4.4)
k=1 k5 2k
k
≡ −200Bp−5 (mod p),
and
p−1 5
k 2k (2) (2)
X
(−1) ((205k2 + 160k + 32)(4H2k − 12Hk ) + 43)
k (4.5)
k=1
≡ 256pHp−1 (mod p5 ).
(10k2 − 6k + 1)(−256)k
X∞
5 = −28ζ(3)
k=1 k5 2k
k
(cf. [14, Identity 8]).
(−256)k π4
X∞
2

(10k − 6k + 1)(2H 2k−1 − H k−1 ) − 3k + 1 = − . (4.6)
5 2k 5 2

k=1 k k
(p−1)/2
X 2k 5
k
(10k2 + 6k + 1)(2H2k − Hk ) + 3k + 1

(−256) k
k=0 (4.7)
14
≡ p + p4 Bp−3 (mod p5 ).
3
Remark 4.1. For any prime p > 3, the author [36, Conjecture 31(ii)] con-
jectured the congruence
p−1 2k 5

X 7
k
k
(10k2 + 6k + 1) ≡ − p5 Bp−3 (mod p6 )
(−256) 2
k=(p+1)/2
which implies that

p−1
X 2k 5
k
(10k2 + 6k + 1)(2H2k − Hk ) + 3k + 1

(−256)k
k=(p+1)/2
p−1 2k 5

X 2
≡ k
· ≡ −7p4 Bp−3 (mod p5 ).
(−256)k p
k=(p+1)/2

(2) (2)
X∞
(−256)k ((10k2 − 6k + 1)(4H2k−1 − 3Hk−1 ) − 2)
5 = −124ζ(5). (4.8)
k 5 2k
k=1 k

(p−1)/2
X (−256)k (2) (2)
2
((10k − 6k + 1)(4H2k−1 − 3Hk−1 ) − 2) ≡ −124Bp−5 (mod p)
2k 5
5
k k
k=1
(4.9)
and
(p−1)/2 2k 5

X (2) (2) 28 3
k
k
((10k2 + 6k + 1)(4H2k − 3Hk ) + 2) ≡ p Bp−3 (mod p4 ).
(−256) 3
k=1
(4.10)
(28k2 − 18k + 3)(−64)k
X∞
4 3k = −14ζ(3)
k=1 k5 2k
k k
conjectured by the author [29] and confirmed recently by Au [4].

(−64)k π4
∞
X
2
4 (28k − 18k + 3)(4H 2k−1 − 3H k−1 ) − 20k + 6 = .
k5 2k 3k 2
k=1 k k
(4.11)
p−1 2k 4 3k
X
k k
(28k2 + 18k + 3)(4H2k − 3Hk ) + 20k + 6

(−64) k
k=0
(4.12)
≡ 6p − 14p4 Bp−3 (mod p5 ).
(2) (2)
X∞
(−64)k ((28k2 − 18k + 3)(2H2k−1 − 3Hk−1 ) − 2)
4 3k = −31ζ(5). (4.13)
k 5 2k
k=1 k k
30 ZHI-WEI SUN

p−1 2k 4 3k
(2) (2)
X
k k 2
(28k + 18k + 3)(2H 2k − 3H k ) + 2
(−64)k (4.14)
k=1
≡ −7p3 Bp−3 (mod p4 ).
∞ 2k 5
X
2 k 8
(20k + 8k + 1) = 2
(−4096)k π
k=0

2k 5
∞
X 16 log 2
k
k
((20k2 + 8k + 1)Hk − 6k − 1) = − (4.15)
(−4096) π2
k=0
and
∞ 2k 5
X 32 log 2
k
(5(20k2 + 8k + 1)H2k − 10k − 1) = − . (4.16)
(−4096)k π2
k=0

(p−1)/2 2k 5

X
k
((20k2 + 8k + 1)Hk − 6k − 1)
(−4096)k (4.17)
k=0
2
≡ − p − 2p2 qp (2) + p3 qp (2)2 − p4 qp (2)3 (mod p5 ),
3
and
(p−1)/2
X 2k 5
k
(5(20k2 + 8k + 1)Hk − 10k − 1)
(−4096)k (4.18)
k=0
≡ − p − 4p2 qp (2) + 2p3 qp (2)2 (mod p4 ).

∞ 2k 5
X (2) (2) 4
k
k
((20k2 + 8k + 1)(8H2k − 3Hk ) + 4) = − . (4.19)
(−4096) 3
k=1

(p−1)/2 2k 5
X (2) (2) 14 3
k
((20k2 + 8k + 1)(8H2k − 3Hk ) + 4) ≡ p Bp−3 (mod p4 )
(−4096)k 3
k=1
(4.20)
and
p−1 2k 5

X (2) (2) 28 3
k
((20k2 + 8k + 1)(8H2k − 3Hk ) + 4) ≡ − p Bp−3 (mod p4 ).
(−4096)k 3
k=1
(4.21)
2k 5
∞
X
2 k 128
(820k + 180k + 13) 20 k
= 2
(−2 ) π
k=0
2k 5
∞
X 256 log 2
k
20 k
(820k2 + 180k + 13)(H2k − Hk ) + 164k + 18 = .
(−2 ) π2
k=0
(4.22)
(p−1)/2
X 2k 5
k 2

(820k + 180k + 13)(H 2k − H k ) + 164k + 18
(−220 )k (4.23)
k=0
≡ 18p + 26p2 qp (2) − 13p3 qp (2)2 (mod p4 ).
2k 5
∞
X (2) (2) 1
k
((820k2 + 180k + 13)(11H2k − 3Hk ) + 43) = − . (4.24)
(−220 )k 3
k=1
p−1 2k 5
(2) (2)
X
k
20 k
((820k2 + 180k + 13)(11H2k − 3Hk ) + 43)
(−2 ) (4.25)
k=1
77
≡ − p3 Bp−3 (mod p4 ),
6
and
(p−1)/2 2k5
(2) (2)
X
k
((820k2 + 180k + 13)(11H2k − 3Hk ) + 43)
(−220 )k (4.26)
k=1
11
≡ − pHp−1 (mod p5 )
4
if p > 5.
The following two conjectures are motivated by the known identity
∞ 2k 4 3k
X
2 48
(74k + 27k + 3) k k
k
= 2
4096 π
k=0
(cf. [15]).
32 ZHI-WEI SUN

X∞ 2k 4 3k
k k
(74k2 + 27k + 3)H2k − 17k − 3 = 0

k
(4.27)
4096
k=0
and
2k 4 3k
∞
X
k k
((74k2 + 27k + 3)(51H3k + 250H2k − 153Hk ) + 15)
4096k
k=0 (4.28)
9792 log 2
= .
π2
p−1 2k 4 3k
X
k k
((74k2 + 27k + 3)H2k − 17k − 3) ≡ −3p + 7p4 Bp−3 (mod p5 ),
4096k
k=0
(4.29)
and
p−1 2k 4 3k
X
k k
((74k2 + 27k + 3)(51H3k + 250H2k − 153Hk ) + 15)
4096k (4.30)
k=0
≡ 15p + 612p2 qp (2) − 306p3 qp (2)2 (mod p4 ).
∞ 2k 4 3k
(2) (2)
X
k k
((74k2 + 27k + 3)(92H2k − 33Hk ) + 112) = 160. (4.31)
4096k
k=1
p−1 2k 4 3k
(2) (2)
X
k k 2
(74k + 27k + 3)(92H 2k − 33H k ) + 112
4096k (4.32)
k=1
≡ 644p3 Bp−3 (mod p4 ).
∞ 2k 4 4k
X
2 32
(120k + 34k + 3) k 16k2k = 2
2 π
k=0
X∞ 2k 4 4k
k 2k
(2(120k2 + 34k + 3)H4k − 16k − 1) = 0 (4.33)
216k
k=0
and
∞ 2k 4 4k
X 128 log 2
k 2k
((120k2 + 34k + 3)(H2k − 2Hk ) + 68k + 9) = . (4.34)
216k π2
k=0

p−1 2k4 4k
X
k 2k
((120k2 + 34k + 3)(H2k − 2Hk ) + 68k + 9)
216k (4.35)
k=0
≡ 9p + 12p2 qp (2) − 6p3 qp (2)2 (mod p4 ),
and
p−1 2k4 4k
X 77 4
k 2k
(2(120k2 + 34k + 3)H4k − 16k − 1) ≡ −p + p Bp−3 (mod p5 )
216k 6
k=0
(4.36)
if p > 3.
∞ 2k 4 4k
X (2) (2) 16
k
16k
2k
((120k2 + 34k + 3)(23H2k − 7Hk ) + 24) = . (4.37)
2 3
k=1

p−1 2k4 4k
X (2) (2) 161 3
k
16k
2k
((120k2 +34k+3)(23H2k −7Hk )+24) ≡ p Bp−3 (mod p4 ),
2 2
k=1
(4.38)
and
(p−1)/2 2k4 4k
(2) (2)
X
k 2k
((120k2 + 34k + 3)(23H2k − 7Hk ) + 24)
216k (4.39)
k=1
≡ −23pHp−1 (mod p5 )
if p 6= 5.
X∞ 2k3 3k 4k
k k 2k
(252k2 + 63k + 5)(4H4k + 3H3k − 7Hk ) + 504k + 63

4
(−24 ) k
k=0
192 log 24
= .
π2
(4.40)
p−1 2k3 3k 4k
X
k k 2k 2

(252k + 63k + 5)(4H 4k + 3H 3k − 7H k ) + 504k + 63
(−244 )k
k=0
5
≡ 63p + 5p2 qp (244 ) − p3 qp (244 )2 (mod p4 ).
2
(4.41)
34 ZHI-WEI SUN
Remark 4.2. Conjecture 4.15 is motivated by the identity

2k 3 3k 4k
∞
X
2 k k 2k 48
(252k + 63k + 5) 4 k
= 2
(−24 ) π
k=0
(cf. [10]).
∞ 2k 2 3k 2 6k
X
k k 3k
3(532k2 + 126k + 9)(H6k − Hk ) + 532k + 63

10 6k
k=0 (4.42)
1125 log 10
= .
4π 2
(ii) For any odd prime p 6= 5, we have
p−1 2k 2 3k 2 6k
X
k k 3k 2

3(532k + 126k + 9)(H 6k − H k ) + 532k + 63
106k (4.43)
k=0
9 9
≡ 63p + p2 qp (106 ) − p3 qp (106 )2 (mod p4 ).
2 4
2k 2 3k 2 6k
∞
X
2 k k 3k 375
(532k + 126k + 9) 6k
= 2
10 4π
k=0
(cf. [10]).
X∞ 2k 2 3k2 6k
k k 3k 2

6(1930k + 549k + 45)(H 6k − H k ) + 3860k + 549
(−218 )k
k=0
6912 log 2
= .
π2
(4.44)
p−1 2k 2 3k2 6k
X
k k 3k
6(1930k2 + 549k + 45)(H6k − Hk ) + 3860k + 549

18
(−2 ) k
k=0
45 3
≡ 549p + 45p2 qp (218 ) − p qp (218 )2 (mod p4 ).
2
(4.45)
∞ 2k 2 3k2 6k
X
2 384
(1930k + 549k + 45) k k
18 k
3k
= 2
(−2 ) π
k=0
(cf. [10]).

X∞ 2k 2 3k2 6k
k k 3k
2(5418k2 + 693k + 29)(H6k − Hk ) + 3612k + 231

18 6
(−2 3 5 ) 3 k
k=0
√
128 5
= log(26 32 5).
π2
(4.46)
p−1 2k 2 3k2 6k
X
k k 3k
2(5418k2 + 693k + 29)(H6k − Hk ) + 3612k + 231

18 6
(−2 3 5 ) 3 k
k=0

5 29 2 18 6 3 29 3 18 6 3 2
≡ 231p + p qp (2 3 5 ) − p qp (2 3 5 ) (mod p4 ).
p 3 6
(4.47)
∞ 2k 2 3k 2 6k √
X
2 k k 3k 128 5
(5418k + 693k + 29) =
(−218 36 53 )k π2
k=0
(cf. [10]).
Conjecture 4.19 (2023-01-17). For k ∈ N, set
H(k) := 6H6k + 4H4k − 3H3k − 2H2k − 5Hk .
(i) We have
X∞ 2k2 3k 4k 6k
k k 2k 3k
(1640k2 + 278k + 15)H(k) + 3280k + 278

(−222 33 )k
k=0 (4.48)
256
=√ log(222 33 ).
3π 2
p−1 2k2 3k 4k 6k
X
k k 2k 3k
(1640k2 + 278k + 15)H(k) + 3280k + 278

22
(−2 3 )3 k
k=0
(4.49)
3 15
≡ 278p + 15p2 qp (222 33 ) − p3 qp (222 33 )2 (mod p4 ).
p 2
∞ 2k 2 3k 4k 6k
X
2 256
(1640k + 278k + 15) k k
22
2k 3k
3 k
=√
(−2 3 ) 3π 2
k=0
(cf. [10]).
36 ZHI-WEI SUN
Conjecture 4.20 (2023-01-17). For k ∈ N, set

H(k) := 4H8k − 2H4k + H2k − 3Hk .
(i) We have
X∞ 2k 3 4k 8k

k 2k 4k
(1920k2 + 304k + 15)H(k) + 1920k + 152

(218 74 )k
k=0
√ (4.50)
56 7
= (9 log 2 + 2 log 7).
π2
(ii) For any odd prime p 6= 7, we have
p−1 2k 3 4k 8k
X
k 2k 4k
(1920k2 + 304k + 15)H(k) + 1920k + 152

18
(2 7 )4 k
k=0
(4.51)
7 15 2 18 4 15 3 18 4 2
≡ 152p + p qp (2 7 ) − p qp (2 7 ) (mod p4 ).
p 2 4
∞ 2k 3 4k 8k √
X
2 k 2k 4k 56 7
(1920k + 304k + 15) =
(218 74 )k π2
k=0
(cf. [10]).
256k π6
∞
(2) (2)
X
3 2
7 (21k − 22k + 8k − 1)(4H 2k−1 − 5H k−1 ) − 6k + 2 = .
k 7 2k 24
k=1 k
(4.52)
(p−1)/2 2k 7
(2) (2)
X
k 3 2
(21k + 22k + 8k + 1)(4H 2k − 5H k ) + 6k + 2
256k (4.53)
k=0
≡ 2p (mod p5 ).
Remark 4.8. This is motivated by the identity
(21k3 − 22k2 + 8k − 1)256k π4
X∞
7 =
k7 2k
8
k=1 k
conjectured by Guillera [13].
∞ 2k 7
X
3 2 32
(168k + 76k + 14k + 1) k20k = 3 (4.54)
2 π
k=0
conjectured by B. Gourevich (cf. [10]).


X∞ 2k7
k
7(168k3 + 76k2 + 14k + 1)(H2k − Hk ) + 252k2 + 76k + 7

220k
k=0
320 log 2
= .
π3
(4.55)
(p−1)/2 2k 7
X
k 3 2 2

7(168k + 76k + 14k + 1)(H 2k − H k ) + 252k + 76k + 7
220k
k=0

−1 2 3 4 2 10 5 3 5 6 4
≡ 7p + 10p qp (2) − 5p qp (2) + p qp (2) − p qp (2)
p 3 2
(mod p7 ).
(4.56)
∞ 2k 7
X
k 3 2 (2) (2)
80
20k
(168k + 76k + 14k + 1)(16H 2k − 5H k ) + 8(6k + 1) = .
2 3π
k=0
(4.57)
(p−1)/2 2k 7
(2) (2)
X
k 3 2
(168k + 76k + 14k + 1)(16H 2k − 5H k ) + 8(6k + 1)
220k
k=0

−1
≡ 8p (mod p6 ).
p
(4.58)
References
[1] G. Almkvist, C. Krattenthaler and J. Petersson, Some new series for π, Experiment.
Math. 12 (2003), 441–456.
[2] R. Apéry, Irrationalité de ζ(2) et ζ(3), Astérisque 61 (1979), 11–13.
[3] K. C. Au, Iterated integrals and multiple polylogarithm at algebraic arguments,
arXiv:2201.01676, 2022.
[4] K. C. Au, Colored multiple zeta values, WZ-pairs and infinite sums, arXiv:2212.02986,
2022.
[5] T. Amdeberhan and D. Zeilberger, Hypergeometric series acceleration via the WZ
method, Electron. J. Combin., 4(2) (1997), #R3.
[6] J. M. Borwein and M. Chamberland, Integer powers of arcsin, Int. J. Math. Math.
Sci. 2007 (2007), Article ID 19381, 10pp.
[7] K. N. Boyadzhiev, Series with central binomial coefficients, Catalan numbers, and
harmonic numbers, J. Integer Seq. 15 (2012), Article 12.1.7, 11pp.
[8] H. Chen, Interesting series associated with central binomial coefficients, Catalan num-
bers and harmonic numbers, J. Integer Seq. 19 (2016), Article 16.1.5, 11pp.
[9] W. Chu, Alternating series of Apéry-type for the Riemann zeta function, Contrib.
Discrete. Math. 15 (2020), 108–116.
38 ZHI-WEI SUN
[10] H. Cohen and J. Guillera, Rational hypergeometric identities for 1/π c : survey and
generalizations, arXiv:2101.12592, 2021.
[11] S. Cooper, Ramanujan’s Theta Functions, Springer, Cham, 2017.
[12] J.W.L. Glaisher, On the residues of the sums of products of the first p − 1 numbers,
and their powers, to modulus p2 or p3 , Quart. J. Math. 31 (1900), 321–353.
[13] J. Guillera, About a new kind of Ramanujan-type series, Experiment. Math. 12 (2003),
507-510.
[14] J. Guillera, Hypergeometric identities for 10 extended Ramanujan-type series, Ra-
manujan J. 15 (2008), 219–234.
[15] J. Guillera, A new Ramanujan-like series for 1/π 2 , Ramanujan J. 26 (2011), 369–374.
[16] J. Guillera, WZ-proofs of “divergent” Ramanujan-type series, in: Advances in Com-
binatorics (eds., I. Kotsireas and E. Zima), Springer, 2013, 187–195.
[17] J. Guillera and M. Rogers, Ramanujan series upside-down, J. Austral. Math. Soc. 97
(2014), 78–106.
[18] V.J.W. Guo and X. Lian, Some q-congruenceson double basic hypergeometric sums,
J. Difference Equ. Appl. 27 (2021), 453–461.
[19] 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.
[20] Q.-H. Hou, C. Krattenthaler and Z.-W. Sun, On q-analogues of some series for π and
π 2 , Proc. Amer. Math. Soc. 147 (2019), 1953–1961.
[21] T. Huber, D. Schultz and D. Ye, Series for 1/π of level 20, J. Number Theory 188
(2018), 121–136.
[22] L. Long, Hypergeometric evaluation identities and supercongruences, Pacific J. Math.
249 (2011), 405–418.
[23] G.-S. Mao and H. Pan, Congruences corresponding to hypergeometric identities I. 2 F1
transformations, J. Math. Anal. Appl. 505 (2022), Article ID: 125527.
[24] S. Ramanujan, Modular equations and approximations to π, Quart. J. Math. (Oxford)
(2) 45 (1914), 350–372.
[25] Z.-H. Sun, Congruences concerning Legendre polynomials, Proc. Amer. Math. Soc.
139 (2011), 1915–1929.
[26] Z.-W. Sun, On congruences related to central binomial coefficients, J. Number Theory
131 (2011), 2219–2238.
[27] Z.-W. Sun, Super congruences and Euler numbers, Sci. China Math. 54 (2011), 2509–
2535.
[28] Z.-W. Sun, Arithmetic theory of harmonic numbers, Proc. Amer. Math. Soc. 140
(2012), 415–428.
[29] Z.-W. Sun, Products and sums divisible by central binomial coefficients, Electron. J.
Combin. 20 (2013), no. 1, #P9, 1-14.
[30] Z.-W. Sun, Supercongruences involving products of two binomial coefficients, Finite
Fields Appl. 22 (2013), 24–44.
[31] Z.-W. Sun, Conjectures and results on x2 mod p2 with 4p = x2 + dy 2 , in: Number
Theory and Related Area (eds., Y. Ouyang, C. Xing, F. Xu and P. Zhang), Adv. Lect.
Math. 27, Higher Education Press and International Press, Beijing-Boston, 2013, pp.
149–197.
[32] Z.-W. Sun, p-adic congruences motivated by series, J. Number Theory 134 (2014),
181–196.
[33] Z.-W. Sun, New series for some special values of L-functions, Nanjing Univ. J. Math.
Biquarterly 32 (2015), no. 2, 189–218.
[34] Z.-W. Sun, A new series for π 3 and related congruences, Internat. J. Math. 26 (2015),
no. 8, Article ID: 1550055 (23 pages).
[35] Z.-W. Sun, Supercongruences motivated by e, J. Number Theory 147 (2015), 326–341.
[36] Z.-W. Sun, Open conjectures on congruences, Nanjing Univ. J. Math. Biquarterly 36
(2019), no. 1, 1–99.
[37] Z.-W. Sun, New series for powers of π and related congruences, Electron. Res. Arch.
28 (2020), no. 3, 1273–1342.
[38] Z.-W. Sun, New type series for powers of π, J. Comb. Number Theory 12 (2020),
157–208.
[39] Z.-W. Sun, New Conjectures in Number Theory and Combinatorics (in Chinese),
Harbin Institute of Technology Press, 2021.
[40] Z.-W. Sun and L.-L. Zhao, Arithmetic theory of harmonic numbers (II), Colloq. Math.
130 (2013), 67–78.
[41] H. Swisher, On the supercongruence conjecture of van Hamme, Res. Math. Sci. 2
(2015), Article ID: 18.
[42] L. van Hamme, Some conjectures concerning partial sums of generalized hypergeo-
metric series, in: p-adic Functional Analysis, Lecture Notes in Pure and Appl. Math.,
192, Dekker, New York, 1997, 223–236.
[43] C. Wang and Z.-W. Sun, Proof of some conjectural hypergeometric supercongruences
via curious identities, 505 (2022), Article ID: 125575.
[44] C. Wei, On two double series for π and their q-analogues, Ramanujan J. 60 (2023),
615–625.
[45] C. Wei and G. Ruan, Double series for π and their q-ananlogues, arXiv:2210.01331,
2022.
[46] J. Wolstenholme, On certain properties of prime numbers, Quart. J. Appl. Math. 5
(1862), 35–39.
[47] 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

Zhi Wei SUN CONJECTURES

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

Zhi Wei SUN CONJECTURES

Uploaded by

Copyright:

Available Formats

To appear in: M. B. Nathanson (ed.), Combinatorial and Additive Number Theory, Springer.

SERIES WITH SUMMANDS

We discover exact values of some series involving harmonic numbers of

For each m ∈ Z+ = {1, 2, 3, . . .}, the harmonic numbers of order m are

J. Wolstenholme [46] established two fundamental congruences for harmonic

In 2012, K. N. Boyadzhiev [7] proved that

modulo p2 for any prime p > 3. By [6], we have

Proof. By taking derivatives of both sides of (1.3), we get

The Bernoulli numbers B0 , B1 , B2 , . . . are defined by

In 1900 J.W.L. Glaiser [12] proved that

The Euler numbers E0 , E1 , E2 , . . . are defined by

Clearly E2n+1 = 0 for all n ∈ N. It is also known that

The Euler polynomials are given by

The Dirichlet beta function is given by

· ) the Kronecker symbol. For a prime p and an integer a 6≡ 0 (mod p),

2. Series with summands containing

(ii) For any prime p > 5, we have

Remark 2.1. In 1979 R. Apéry [2] proved the irrationality of ζ(3) =

In contrast with (2.5), we have

by applying (1.5) with x = 1.

Remark 2.4. The author’s conjectural identities

(ii) For any odd prime p, we have the congruence

Remark 2.5. In 1974 R. W. Gosper announced the identity

an elegant proof of which can be found in [1].

(ii) For any prime p > 3, we have

Remark 2.6. For any positive integer n, we clearly have

Using this trick we can deduce that

Conjecture 2.7 (2023-05-28). We have

Remark 2.8. For any x ∈ (−1, 1), we have

which can be proved directly or via Mathematica. In particular,

For x ∈ (−1, 1), we obviously have

by using (1.1) and (1.2). In particular,

With the aid of Mathematica, we obtain that

Conjecture 2.9 (2022-12-30). We have

where Γ(x) is the well-known Gamma function.

For any prime p ≡ 1 (mod 4) with p = x2 +y 2 (x, y ∈ Z and x ≡ 1 (mod 4)),

Conjecture 2.10 (2022-11-14). We have the identity

Remark 2.10. In contrast with (2.19), Mathematica yields that

Conjecture 2.11 (2022-12-30). We have

Remark 2.11. Mathematica yields the identity

Conjecture 2.12 (2022-12-30). We have

Remark 2.12. For any prime p > 3, we have

by [30, Corollary 1.4], and

as conjectured by the author [27, Conjecture 5.13] and proved by C. Wang

Remark 2.13. Mathematica yields the identity

By [30, Corollary 1.3], for any prime p ≡ 5, 7 (mod 8) we have

Conjecture 2.14 (2022-12-30). We have

Remark 2.14. Let p > 3 be a prime. By [30, Corollary 1.4],

as conjectured by the author [27, Conjecture 5.14(iii)] and proved by Wang

Remark 2.15. Let p > 3 be a prime. By [30, Corollary 1.4], we have

If p = x2 + 3y 2 with x, y ∈ Z and x ≡ 1 (mod 3), then

as conjectured by the author [27, Conjecture 5.14] and confirmed by G.-S.

as conjectured by the author [27, Conjecture 5.14] and confirmed by Wang

Remark 2.16. Let p > 3 be a prime with p 6= 7. By [30, Corollary 1.4],

The author [27, Conjecture 5.14(ii)] conjectured that

Conjecture 2.17 (2022-12-30). We have

Remark 2.17. Mathematica yields the identity

3. Series and congruences with summands

The author [34] proved that

Remark 3.2. The identity (3.5) looks quite challenging.

Conjecture 3.3 (2022-10-13). (i) We have

(iii) Let p be an odd prime. Then