Graphs and Combinatorics (2020) 36:1273–1295

https://doi.org/10.1007/s00373-020-02207-3(0123456789().,-volV)(0123456789().,-volV)

ORIGINAL PAPER

On q-Series and Split Lattice Paths

Megha Goyal1

Received: 24 April 2016 / Revised: 10 May 2020 / Published online: 13 August 2020
Ó Springer Japan KK, part of Springer Nature 2020

Abstract
In this paper a natural question which arise to study the graphical aspect of split
ðn þ tÞ-color partitions, is answered by introducing a new class of lattice paths,
called split lattice paths. A direct bijection between split ðn þ tÞ-color partitions and
split lattice paths is proved. This new combinatorial object is applied to give new
combinatorial interpretations of two basic functions of Gordon-McIntosh. Some
generalized q-series are also discussed. We further explore these paths by providing
combinatorial interpretations of some Rogers-Ramanujan type identities which
reveal their rich structure and great potential for further research.

Keywords q-series  Split ðn þ tÞ-color partitions  Split lattice paths 

Combinatorial identities

Mathematics Subject Classification 05A15  05A17  05A19  11P81

1 Introduction

The graphical prospect of partitions and compositions has always drawn the
attention of mathematicians. Graphical representation is very useful when
applications of partitions and compositions are considered [9, 19, 20]. In 1987,
Agarwal and Andrews [1] defined ðn þ tÞ-color partitions. To study the graphical
aspect of this new set of partitions, Agarwal and Bressoud [2] introduced weighted
lattice paths. Several basic series identities had been interpreted combinatorially
using ordinary partitions, colored partitions, Frobenius partitions, lattice paths,
associated lattice paths etc. in [1–5, 8, 11, 12, 14, 15, 18]. In 2014, Agarwal and
Sood [6] defined a new class of partitions, called split ðn þ tÞ-color partitions. Using

& Megha Goyal

[email protected]
1
Department of Mathematical Sciences, IK Gujral Punjab Technical University Jalandhar, Main
Campus, Kapurthala 144603, India

123
1274 Graphs and Combinatorics (2020) 36:1273–1295

this new combinatorial object, Agarwal and Sood gave combinatorial meaning to
two basic functions of Gordon–McIntosh [10]. This new set of partitions has the
potential to provide combinatorial interpretations to q-series which cannot be
interpreted combinatorially using ordinary partitions or ðn þ tÞ-color partitions, for
instance see [13, 17]. But the graphical aspect of split ðn þ tÞ-color partitions is still
remained an unexplored concept.
The present article is a contribution to the quest for settling the graphical
representation of split ðn þ tÞ-color partitions. We have originally started this
problem by defining a new combinatorial object which we call split lattice paths.
These objects are natural combinatorial structures associated with basic series. It
turned out, however, that the split lattice paths provide not only the graphical
representation of split ðn þ tÞ-color partitions but split lattice paths will also provide
graphical representations of all ordered split ðn þ tÞ-color partitions which may be
studied in future. The next section is dedicated to the introduction of split lattice
paths and furthermore, we will establish bijectively the equivalence of the two
sets—split ðn þ tÞ-color partitions of a nonnegative integer l and split lattice paths
of weight l. Another useful aspect of split lattice paths is explored in Sect. 3.
Agarwal and Sood [6] proved that the summations series in the following basic
functions of Gordon and McIntosh [10] are the generating functions of certain
restricted classes of split ðn þ tÞ-color partitions.
X
1 k2
q ðq; q2 Þ k
V0 ðqÞ ¼1 þ 2 ; ð1Þ
k¼1
ðq; q2 Þk

X
1 k2
q ðq; q2 Þ k1
V1 ðqÞ ¼ ; ð2Þ
k¼1
ðq; q2 Þk

where
Y1
ð1  aqi Þ
ða; qÞk ¼ :
i¼0
ð1  aqkþi Þ

They used recurrence equations to prove their results, and their idea of the proof is
algebraic. Here we will establish that certain restricted classes of split lattice paths
are also generated by the summation series on R.H.S. of (1) and (2). But our idea of
the proof is purely combinatorial and constructive. Here we develop a new com-
binatorial method to study these identities and exploit a new possibility of the
combinatorial arguments on the split lattice paths. This new method yields more
combinatorial and insightful proofs of these identities. Furthermore, we will prove a
1–1 correspondence between the combinatorial objects generated by the summa-
tions in (1) and (2). In Sect. 4 we will further explore the concept of split lattice
paths by interpreting three generalized basic series combinatorially. Our study
moreover produces some particular cases of these three combinatorial identities
which ultimately yield Rogers-Ramanujan type identities. These results are dis-
cussed in Sect. 5. The article closes in Sect. 6 with remarks on the combinatorial

123
 Graphs and Combinatorics (2020) 36:1273–1295 1275

connections between split lattice paths, q-series and split ðn þ tÞ-color partitions.
We conclude this section by posing an open problem.
Before we state our main results we first recall some definitions from [1].
Definition 1 A partition with ‘‘ðn þ tÞ copies of n’’, t  0, is a partition in which a
part of size n, n  0, can come in ðn þ tÞ different colors denoted by subscripts:
n1 ; n2 ; . . .; nnþt . Note that zeros are permitted if and only if t is greater than or equal
to one. Also, zeros are not permitted to repeat in any partition.
Remark 1 We note that if we take t ¼ 0, then these are nothing but the n-color
partitions. If order of the parts is considered then these are nothing but the n-color
compositions.
Definition 2 The weighted difference of two parts gk , hl (g  h) is defined by
g  h  k  l and is denoted by ððgk  hl ÞÞ.
In [6] the split (n þ t)-color partitions are defined as:
Definition 3 Let ap be a part in an (n þ t)-color partition of a non-negative integer
l. We split the color ‘p’ into two parts-‘the green part’ and ‘the red part’ and denote
them by ‘g’ and ‘r’ respectively, such that 1  g  p; 0  r  p  1 and p ¼ g þ r.
An (n þ t)-color partition in which each part is split in this manner is called split
(n þ t)-color partition.
Example 1 In 52þ1 , the green part is 2 and the red part is 1.
Remark 2 We note that if r ¼ 0, then we will not write it. Thus for instance, we
will write 53 for 53þ0 .
Anand and Agarwal [7] gave the following description of associated lattice
paths.
Definition 4 All paths will be of finite length lying in the first quadrant. They start
on the Y-axis (origin included), end on the X-axis and use three kinds of unitary
steps:
northeast: from (x, y) to ðx þ 1; y þ 1Þ.
southeast: from (x, y) to ðx þ 1; y  1Þ, only allowed if y [ 0.
horizontal: from (x, y) to ðx þ 1; yÞ, only allowed when the first step of a sequence
of consecutive horizontal steps is preceded by a northeast step and the
last is followed by a southeast step.

The following terminology is used in describing associated lattice paths:

Truncated Isosceles Trapezoidal Section (TITS): A section of the path which
starts on the X-axis with northeast steps followed by horizontal steps and then
followed by southeast steps ending on the X-axis forms a Truncated Isosceles
Trapezoidal Section. Since the lower base lies on X-axis and is not a part of the
path, hence the term truncates.
Slant Section (SS): A section of the path consisting of only southeast steps which
starts on the Y-axis (origin not included) and ends on the X-axis.

123
1276 Graphs and Combinatorics (2020) 36:1273–1295

3
b

a
1

0
0 1 2 3 4 5 6 7 8 9

Fig. 1 One SS of height 1 and one TITS with ordered pair f2; 3g

Height of a Slant Section: It is ‘t’ if it starts from (0, t). Clearly, a path can have an
SS only in the beginning of the path. An associated lattice path can have at most
one SS.
Weight of a TITS: Every TITS is represented by an ordered pair {a, b}, where
a denotes its altitude and b the length of the upper base. The weight of a TITS
with ordered pair {a, b} is a.
Weight of an Associated Lattice Path: It is the sum of weights of its TITSs.
Note that the weight of the Slant Section is zero.

Example 2 In Fig. 1, the associated lattice path has one SS of height 1 and one
TITS with ordered pair f2; 3g and its weight is 2.

2 Split Lattice Paths and Split ðn + tÞ-Color Partitions

Now we introduce a new class of lattice paths which we call split lattice paths and
we describe the terminology used in split lattice paths.
Definition 5 In split lattice paths, we split the length b of the upper base of each
TITS in an associated lattice path into two parts—the left part will be called ‘ray’
and the right part a ‘segment’ and we denote their lengths by ‘r’ and ‘s’ respectively,
such that 1  r  b, 0  s  b  1 and b ¼ r þ s. An associated lattice path in which
the lengths of the upper bases of all the TITSs are split into rays and segments is
called a split lattice path. In a split lattice path, the ray is represented by a ‘solid
line’ and the segment is represented by a ‘dotted line’.
Remark 3 For all type of calculations and logics, we will consider the length of the
upper base b as a whole and not its parts r and s, separately.
Definition 6 The following order is defined on the set of all TITSs of a split lattice
path which firstly depends upon their weights and then on the length of their upper
bases:
If a\c then TITS with ordered pair fa; bg will appear before the TITS with
ordered pair fc; dg and if a ¼ c then the TITS with ordered pair fa; bg will appear
before the TITS with ordered pair fc; dg, where b\d. Further if a ¼ c and b ¼ d,

123
Graphs and Combinatorics (2020) 36:1273–1295 1277

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Fig. 2 One SS of height 1 and two TITSs with same ordered pair f2; 3g

then the order of these TITSs depends upon the length of the ray, that is, say
b ¼ r1 þ s1 and d ¼ r2 þ s2 , then TITS with ordered pair fa; bg will appear
before the TITS with ordered pair fc; dg, where r1 \r2 . Thus, the TITSs satisfy the
order: f1; 1g\f1; 1 þ 1g\f1; 2g\f2; 1g\f2; 1 þ 1g\f2; 2g\f3; 1g\f3; 1 þ
1g \f3; 2g\f3; 1 þ 2g\f3; 2 þ 1g\f3; 3g\   .
Example 3 In Fig. 2, the split lattice path has one SS of height 1 and two TITS with
ordered pair f2; 3g. Here the weight of the first TITS is 2 and the upper base is split
into a ray of length 1 and a segment of length 2. The weight of the second TITS is
also 2 but here the upper base is split into a ray of length 2 and a segment of length
1. Thus in view of Definition 6, TITS with a ray of length 1 will appear before the
TITS with a ray of length 2 in the corresponding split lattice path. Here the weight
of the split lattice path is 4.
Remark 4 It is worthy to note here that if we do not consider the condition of
arranging the TITSs in a non decreasing order of their altitudes and the lengths of
their upper bases, we get a split lattice path representation for split ðn þ tÞ-color
compositions.
Definition 7 Mt ðlÞ denote the set of all split lattice paths of weight l such that for
any TITS with ordered pair {a, b}, b does not exceed a þ t. When t [ 0 and a ¼ 0,
then there will be an SS of height t and the TITSs are arranged in the order as
defined in Definition 6.
Definition 8 Nt ðlÞ denote the set of all split ðn þ tÞ-color partitions of a
nonnegative integer l.
Now we prove the existence of 1-1 correspondence between the sets Mt ðlÞ and
Nt ðlÞ.
Theorem 1 For t  0, there exists a 1-1 correspondence between the split lattice
paths of weight l enumerated by Mt ðlÞ and the split ðn þ tÞ-color partitions of l
enumerated by Nt ðlÞ. That is,
jMt ðlÞj ¼ jNt ðlÞj for all l  0;
where |S| denotes the cardinality of a finite set S.
Proof We will prove this theorem by mapping each part ap , a  1 of a split colored
partition enumerated by Nt ðlÞ to a TITS with ordered pair {a, p}. If in the split

123
1278 Graphs and Combinatorics (2020) 36:1273–1295

ðn þ tÞ-color partitions enumerated by Nt ðlÞ, t [ 0, then the part 0t in the partition

is mapped to an SS of height t. Also, in the split colored partitions enumerated by
Nt ðlÞ, p can not exceed a þ t. So in this way, we get a split lattice path enumerated
by Mt ðlÞ corresponding to each split ðn þ tÞ-color partition enumerated by Nt ðlÞ.
Clearly, this process is reversible and thus corresponding to each split lattice path
enumerated by Mt ðlÞ, we get a split ðn þ tÞ-color partition enumerated by Nt ðlÞ.
This completes the proof of Theorem 1. h
To illustrate the constructed bijections we consider split n-color partitions of
l ¼ 3 and the corresponding split lattice paths in Table 1.

3 Two Basic Series Functions of Gordon-McIntosh

Agarwal and Sood [6] interpreted the two basic functions (1) and (2) of Gordon and
McIntosh [10] as the generating functions of some restricted split ðn þ tÞ-color
partitions enumerated by A1 ðlÞ and A2 ðlÞ which are defined in Theorems 2 and 3
respectively. Our objective in this section is to extend further these results using
split lattice paths. We will show constructively that certain restricted split lattice
paths are also generated by the summation series on R.H.S of (1) and (2).

Table 1 Split n-color partitions and split lattice paths for l ¼ 3

123
Graphs and Combinatorics (2020) 36:1273–1295 1279

Theorem 2 Let A1 ðk; lÞ denote the set of split n-color partitions of l with exactly
k parts which satisfy the following conditions:
1. the parts and their subscripts have the same parity,
2. the red part of the subscripts can not exceed 1,
3. the least part is either aa ða  1Þ or aða1Þþ1 ða  2Þ,
4. the weighted difference between any two consecutive parts is 0.

Let B1 ðk; lÞ denote the set of split lattice paths of weight l with exactly k TITSs and
no SS which satisfy the following conditions:
1. for any TITS with ordered pair {a, b}, b does not exceed a and a  bðmod 2Þ,
2. there is no segment with length [ 1,
3. if fP; ug is the ordered pair of the first TITS in the split lattice path then P ¼ u,
4. for any two TITSs with respective ordered pairs {a1 ; b1 } and {a2 ; b2 }, (a1  a2 ),
a2  b2 ¼ a1 þ b1 holds.

[
1 [
1
Let A1 ðlÞ ¼ A1 ðk; lÞ and B1 ðlÞ ¼ B1 ðk; lÞ, then
k¼1 k¼1

X
1 X
1 X
1 k2
q ðq; q2 Þ
l l k
A1 ðlÞq ¼ B1 ðlÞq ¼ ;
l¼1 l¼1 k¼1
ðq; q2 Þk

where A1 ðlÞ ¼ jA1 ðlÞj and B1 ðlÞ ¼ jB1 ðlÞj.

Theorem 3 Let A2 ðk; lÞ denote the set of split n-color partitions of l with exactly
k parts which satisfy the following conditions:
1. the parts and their subscripts have the same parity,
2. the red part of the subscripts can not exceed 1,
3. the least part is aa ða  1Þ,
4. the weighted difference between any two consecutive parts is 0.

Let B2 ðk; lÞ denote the set of split lattice paths of weight l with exactly k TITSs and
no SS which satisfy the following conditions:
1. for any TITS with ordered pair {a, b}, b does not exceed a and a  bðmod 2Þ,
2. there is no segment with length [1,
3. if fP; ug is the ordered pair of the first TITS in the split lattice path then P ¼ u
and it has no segment,
4. for any two TITSs with respective ordered pairs {a1 ; b1 } and {a2 ; b2 }, (a1  a2 ),
a2  b2 ¼ a1 þ b1 holds.

[
1 [
1
Let A2 ðlÞ ¼ A2 ðk; lÞ and B2 ðlÞ ¼ B2 ðk; lÞ, then
k¼1 k¼1

123
1280 Graphs and Combinatorics (2020) 36:1273–1295

X
1 X
1 X
1 k2
q ðq; q2 Þ
l l k1
A2 ðlÞq ¼ B2 ðlÞq ¼ ;
l¼1 l¼1 k¼1
ðq; q2 Þk

where A2 ðlÞ ¼ jA2 ðlÞj and B2 ðlÞ ¼ jB2 ðlÞj.

As both theorems have similar proofs, so we will discuss the detailed proof of
Theorem 2 and provide an outline of the proof of Theorem 3.
Remark 5 By defining a similar mapping as defined in Theorem 1, one can easily
establish bijections between the set of split n-color partitions Ai ðlÞ and the set of
split lattice paths Bi ðlÞ for each 1  i  2.

3.1 Proof of Theorem 2

We will prove this theorem in three steps. First we will show that the summation
series on R.H.S. of (1) also generates the split lattice paths enumerated by B1 ðlÞ.
Further we define a set of pairs ða; bÞ denoted by C1 ðk; lÞ where a ¼ ða1 ; a2 ; . . .; ak Þ
and b ¼ ðb1 ; b2 ; . . .; bk Þ are k-tuples of integers and satisfy:
(i) ai 2 Z  0 =f0; 1; 2; 3; . . .g for 1  i  k,
(ii) bi 2 f0; 1g for 1  i  k,
(iii) if the weight of the pair ða; bÞ is defined by
X
k X
k
wða; bÞ ¼ k2 þ ð2i  1Þai þ ð2i  1Þbi ;
i¼1 i¼1

then ða; bÞ should satisfy wða; bÞ ¼ l.

One can easily establish that the generating function for C1 ðk; lÞ is
X 2
qk ðq; q2 Þk
l
C1 ðk; lÞq ¼ ;
l
ðq; q2 Þk

where C1 ðk; lÞ ¼ jC1 ðk; lÞj. In the second and third step we will prove that there is
a bijection between the set of split lattice paths B1 ðk; lÞ and the set of pairs
C1 ðk; lÞ.
Proof Step I. We shall prove that
X
1 X
1 k2
q ðq; q2 Þ
B1 ðlÞql ¼ k
: ð3Þ
l¼1 k¼1
ðq; q2 Þk
2
qk ðq;q2 Þk 2
In ðq;q2 Þk the factor qk generates a split lattice path having k TITSs such that the
altitude of ith TITS is 2i  1 and the length of its ray is 1. Thus the associated
ordered pair is f2i  1; 1g.
For k ¼ 3, the path begins as shown in Fig. 3.

123
Graphs and Combinatorics (2020) 36:1273–1295 1281

6
5
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

Fig. 3 Split lattice path for k ¼ 3

In Fig. 3, we consider two successive TITSs, say, ith and ði þ 1Þth . Their corre-
sponding ordered pairs are f2i  1; 1g and f2i þ 1; 1g respectively as shown in
Fig. 4.
The factor ðq;q12 Þ generates k nonnegative multiples of ð2i  1Þ, 1  i  k; say,
k

a1  1; a2  3;    ; ak  ð2k  1Þ. This is encoded by increasing the altitude of ith

TITS by 2ðak þ ak1 þ    þ akiþ2 Þ þ akiþ1 and the length of the ray by akiþ1 .
So the corresponding ordered pair becomes f2i  1 þ 2ðak þ ak1
þ    þ akiþ2 Þ þ akiþ1 ; 1 þ akiþ1 g. Fig. 4 now changes to Fig. 5.
The factor ðq; q2 Þk generates k distinct nonnegative multiples of ð2i  1Þ,
1  i  k; say, b1  1; b2  3; . . .; bk  ð2k  1Þ; where each bi ð1  i  kÞ is 0 or 1.
This is encoded by increasing the altitude of ith TITS by 2ðbk þ bk1 þ    þ
bkiþ2 Þ þ bkiþ1 and by putting a segment of length bkiþ1 adjacent to the ray. So
the ordered pair becomes f2i  1 þ 2ðak þ ak1 þ    þ akiþ2 Þ þ akiþ1 þ 2ðbk þ
bk1 þ    þbkiþ2 Þ þ bkiþ1 ; 1 þ akiþ1 þ bkiþ1 g. So Fig. 5 now changes to
Fig. 6.

(i+1)st

th
i

Fig. 4 i th and ði þ 1Þth TITSs

st
Fig. 5 i th and ði þ 1Þth TITSs (i+1)

th
i

123
1282 Graphs and Combinatorics (2020) 36:1273–1295

Fig. 6 i th and ði þ 1Þth TITSs st

(i+1)

th
i

Every split lattice path enumerated by B1 ðlÞ is uniquely generated in this

manner. This proves (3).
Furthermore, we see that constructively
 there exists a well-defined map between
the pairs ða; bÞ ¼ ðai Þki¼1 ; ðbi Þki¼1 of the set C1 ðk; lÞ and the k TITSs represented
by the ordered pairs fai ; bi gki¼1 ¼ fai ; ri þ si gki¼1 of the set B1 ðk; lÞ, say,
/ : C1 ðk; lÞ ! B1 ðk; lÞ
defined by
 
/ ðai Þki¼1 ; ðbi Þki¼1 ¼ fakiþ1 ; rkiþ1 þ skiþ1 gki¼1 ;

such that, for all 1  i  k


ai ¼ rkiþ1  1;
/: ð4Þ
bi ¼ skiþ1 :
and akiþ1 ¼ 2ðk  i þ 1Þ  1 þ 2ðak þ ak1 þ    þaiþ1 Þ þ ai þ 2ðbk þbk1 þ
   þ biþ1 Þ þ bi .
Step II. We now show that /ða; bÞ is actually an element of B1 ðk; lÞ. Let us
denote the ith and ði þ 1Þth TITS in Fig. 6 by fai ; bi g and faiþ1 ; biþ1 g respectively,
then
ai ¼ 2i  1 þ 2ðak þ ak1 þ    þ akiþ2 Þ þ akiþ1
þ 2ðbk þ bk1 þ    þ bkiþ2 Þ þ bkiþ1 ;
bi ¼ 1 þ akiþ1 þ bkiþ1 ;
ð5Þ
aiþ1 ¼ 2i þ 1 þ 2ðak þ ak1 þ    þ akiþ1 Þ þ aki
þ 2ðbk þ bk1 þ    þ bkiþ1 Þ þ bki ;
biþ1 ¼ 1 þ aki þ bki :
Clearly, ai  bi and the parity of ai and bi depends upon akiþ1 þ bkiþ1 . If akiþ1 þ
bkiþ1 is odd then both ai and bi are even and when akiþ1 þ bkiþ1 is even then
both ai and bi are odd. This confirms that for any TITS with ordered pair {a, b}, b
does not exceed a and a  bðmod 2Þ.
Also, we know from condition (ii) on the pairs of the set C1 ðk; lÞ, bi 2 f0; 1g for
1  i  k. Thus using (4), there cannot be any TITS with segment of length greater
than 1 in the split lattice paths of the set B1 ðk; lÞ. If fa1 ; b1 g is the ordered pair of

123
Graphs and Combinatorics (2020) 36:1273–1295 1283

first TITS and using (5), it is obvious that a1 ¼ b1 . Now again using (5), we consider
aiþ1  ai ¼ 2 þ aki þ bki þ akiþ1 þ bkiþ1 ¼ biþ1 þ bi .
Step III. Finally we show that / is injective as well as surjective.
1. / is 1-1:    
Suppose ðai Þki¼1 ; ðbi Þki¼1 and ðci Þki¼1 ; ðdi Þki¼1 be any two pairs in the set
C1 ðk; lÞ such that:
   
/ ðai Þki¼1 ; ðbi Þki¼1 ¼ / ðci Þki¼1 ; ðdi Þki¼1
) fakiþ1 ; rkiþ1 þ skiþ1 g ¼ fa0kiþ1 ; rkiþ1
0
þ s0kiþ1 g; for 1  i  k
ð6Þ
We know by Definition 6 two TITSs are equal if their weights as well as the
lengths of their rays and segments are equal. Thus from (6) we have
0
rkiþ1 ¼ rkiþ1 and skiþ1 ¼ s0kiþ1
) 1 þ ai ¼ 1 þ c i and bi ¼ di ; using (4)
) ai ¼ c i and bi ¼ di ; for 1  i  k
   
) ðai Þi¼1 ; ðbi Þki¼1
k
¼ ðci Þki¼1 ; ðdi Þki¼1 :

Hence / is injective.
2. / is surjective:
Consider TITSs with ordered pairs fai ; bi g for 1  i  k of split lattice paths
enumerated by B1 ðk; lÞ, where bi ¼ ri þ si for 1  i  k. Now by the conditions
of Theorem 2 on the split lattice paths in the set B1 ðk; lÞ, we know every ri  1,
si ¼ 0 or 1 for 1  i  k. So, by definition of the map /, we have ai ¼
rkiþ1  1  0 and bi ¼ skiþ1 ¼ 0 or 1 for 1  i  k. Thus ðai Þki¼1 ; ðbi Þki¼1 is
a pair in the set C1 ðk; lÞ. Hence / is surjective.

And the inverse mapping is:


1 ri ¼ akiþ1 þ 1;
/ : for 1  i  k
si ¼ bkiþ1 :
and ai ¼ 2i  1 þ 2ðak þ ak1 þ    þ akiþ2 Þ þ akiþ1 þ 2ðbk þ bk1 þ    þ
bkiþ2 Þ þ bkiþ1 . h

3.2 Outline of the Proof of Theorem 3

The proof of Theorem 3 is treated in the same manner as Theorem 2. Here the
factor ðq; q2 Þk1 generates k  1 distinct nonnegative multiples of 2i  1,
1  i  k  1. So, we will not put any segment adjacent to the ray in the first
TITS. Thus the length of the segment in the first TITS of corresponding split lattice
path is zero. Clearly, this makes the smallest part of the form aa in the associated
split n-color partition. Here we define the set of pairs ða; bÞ denoted by C2 ðk; lÞ

123
1284 Graphs and Combinatorics (2020) 36:1273–1295

where a ¼ ða1 ; a2 ; . . .; ak Þ is a k-tuple and b ¼ ðb1 ; b2 ; . . .; bk1 Þ is a ðk  1Þ-tuple

of integers and satisfy:
(i) ai 2 Z  0 =f0; 1; 2; 3; . . .g for 1  i  k,
(ii) bi 2 f0; 1g for 1  i  k  1,
(iii) if the weight of the pair ða; bÞ is defined by

X
k X
k1
wða; bÞ ¼ k2 þ ð2i  1Þai þ ð2i  1Þbi ;
i¼1 i¼1

then ða; bÞ should satisfy wða; bÞ ¼ l. Clearly, the generating function for C2 ðk; lÞ
will be
X 2
qk ðq; q2 Þk1
l
C2 ðk; lÞq ¼ :
l
ðq; q2 Þk

To establish a bijection between the set of pairs C2 ðk; lÞ and the set of split lattice
paths B2 ðk; lÞ, we define the map

w : C2 ðk; lÞ ! B2 ðk; lÞ
such that

ai ¼ rkiþ1  1; for 1  i  k
w:
bi ¼ skiþ1 ; for 1  i  k  1
and akiþ1 ¼ 2ðk  i þ 1Þ  1 þ 2ðak þ ak1 þ    þ aiþ1 Þ þ ai þ 2ðbk þ bk1 þ
   þ biþ1 Þ þbi .
By using the similar arguments as used in the Proof of Theorem 2, one can easily
verify that the map w is bijective and the inverse map is given by:

1 ri ¼ akiþ1 þ 1; for 1  i  k;
w :
si ¼ bkiþ1 ; for 1  i  k  1

4 Three Generalized Basic Series

Definition 9 Let T ¼ f1; 1; 3; 5; 7; . . .g. For jqj\1, t 2 T and 1  k  3, we

define fkt ðqÞ by
X
1 k½1þðk1Þðtþ3Þ=2
q ðq; q2 Þ k
f1t ðqÞ ¼ ; ð7Þ
k¼0
ðq2 ; q2 Þk ðq; q2 Þk

X
1 kðkþ1Þðtþ3Þ=2
q ðq; q2 Þ k
f2t ðqÞ ¼ ; ð8Þ
k¼0
ðq2 ; q2 Þk ðq; q2 Þkþ1

123
Graphs and Combinatorics (2020) 36:1273–1295 1285

X
1 k½1þðkþ1Þðtþ3Þ=2
q ðq; q2 Þ k
f3t ðqÞ ¼ : ð9Þ
k¼0
ðq2 ; q2 Þk ðq; q2 Þkþ1

Sood and Agarwal [17] proved that the q-series (7)–(9) have their combinatorial
counterparts in the form of split ðn þ tÞ-color partitions enumerated by Atk ðlÞ, for
1  k  3, which are defined in Theorems 4, 5 and 6 respectively. Our objective in
this section is to extend these results in terms of split lattice paths.

4.1 Main Results

Theorem 4 Let At1 ðk; lÞ denote the set of split n-color partitions of l with exactly
k parts which satisfy the following conditions:
1. the parts and their subscripts have the same parity,
2. the red part of the subscripts can not exceed 1,
3. the weighted difference between any two consecutive parts is [ t.

Let Bt1 ðk; lÞ denote the set of split lattice paths of weight l with exactly k TITSs and
no SS which satisfy the following conditions:
1. for any TITS with ordered pair {a, b}, b does not exceed a and a  bðmod 2Þ,
2. there is no segment with length [1,
3. for any two TITSs with respective ordered pairs {a1 ; b1 } and {a2 ; b2 }, (a1  a2 ),
a2  b2 [ a1 þ b1 þ t holds.

[
1 [
1
Let At1 ðlÞ ¼ At1 ðk; lÞ and Bt1 ðlÞ ¼ Bt1 ðk; lÞ, then
k¼1 k¼1

X
1 X
1
At1 ðlÞql ¼ Bt1 ðlÞql ¼ f1t ðqÞ;
l¼0 l¼0

where At1 ðlÞ ¼ jAt1 ðlÞj and Bt1 ðlÞ ¼ jBt1 ðlÞj.

Theorem 5 Let At2 ðk; lÞ denote the set of split ðn þ 1Þ-color partitions of l with
exactly k parts which satisfy the following conditions:
1. the parts and their subscripts have opposite parity,
2. the red part of the subscripts can not exceed 1,
3. the smallest part is of the form aaþ1 , for some a,
4. the red part of the subscript of the smallest part is 0,
5. the weighted difference between any two consecutive parts is [ t.

Let Bt2 ðk; lÞ denote the set of split lattice paths of weight l with exactly k TITSs
which satisfy the following conditions:

123
1286 Graphs and Combinatorics (2020) 36:1273–1295

1. for any TITS with ordered pair {a, b}, b does not exceed a þ 1 and
a 6 bðmod 2Þ,
2. there is no segment with length [1,
3. there is an SS of height 1 or a TITS with ordered pair fa; a þ 1g,
4. the length of the segment in first TITS is 0,
5. for any two TITSs with respective ordered pairs {a1 ; b1 } and {a2 ; b2 }, (a1  a2 ),
a2  b2 ¼ a1 þ b1 þ t holds.

[
1 [
1
Let At2 ðlÞ ¼ At2 ðk; lÞ and Bt2 ðlÞ ¼ Bt2 ðk; lÞ, then
k¼1 k¼1

X
1 X
1
At2 ðlÞql ¼ Bt2 ðlÞql ¼ f2t ðqÞ;
l¼0 l¼0

where At2 ðlÞ ¼ jAt2 ðlÞj and Bt2 ðlÞ ¼ jBt2 ðlÞj.

Theorem 6 Let At3 ðk; lÞ denote the set of split ðn þ 2Þ-color partitions of l with
exactly k parts which satisfy the following conditions:
1. the parts and their subscripts have the same parity,
2. the red part of the subscripts can not exceed 1,
3. the least part is of the form aaþ2 for some a,
4. the red part of the subscript of the smallest part is 0,
5. the weighted difference between any two consecutive parts is [ t.

Let Bt3 ðk; lÞ denote the set of split lattice paths of weight l with exactly k TITSs
which satisfy the following conditions:
1. for any TITS with ordered pair {a, b}, b does not exceed a þ 2 and
a  bðmod 2Þ,
2. there is no segment with length [1,
3. there is an SS of height 2 or a TITS with ordered pair fa; a þ 2g,
4. the length of the segment in first TITS is 0,
5. for any two TITSs with respective ordered pairs {a1 ; b1 } and {a2 ; b2 }, (a1  a2 ),
a2  b2 ¼ a1 þ b1 þ t holds.

[
1 [
1
Let At3 ðlÞ ¼ At3 ðk; lÞ and Bt3 ðlÞ ¼ Bt3 ðk; lÞ, then
k¼1 k¼1

X
1 X
1
At3 ðlÞql ¼ Bt3 ðlÞql ¼ f3t ðqÞ;
l¼0 l¼0

where At3 ðlÞ ¼ jAt3 ðlÞj and Bt3 ðlÞ ¼ jBt3 ðlÞj.
Again as these theorems have similar proofs, so we will discuss the detailed proof
of Theorem 4 and provide an outline of the proofs of Theorems 5–6.

123
Graphs and Combinatorics (2020) 36:1273–1295 1287

Remark 6 Again by defining a similar mapping as defined in Theorem 1 one can

easily establish bijections between the set of split ðn þ tÞ-color partitions Atk ðlÞ and
the set of split lattice paths Btk ðlÞ for each 1  k  3 and t 2 T.

4.2 Proof of Theorem 4

This theorem is also proved in three steps. Here, we define a set of triplets ða; b; cÞ
denoted by Ct1 ðk; lÞ where a ¼ ða1 ; a2 ;    ; ak Þ, b ¼ ðb1 ; b2 ;    ; bk Þ and c ¼
ðc1 ; c2 ;    ; ck Þ are k-tuples of integers and satisfy:
(i) ai 2 Z  0 =f0; 1; 2; 3; . . .g for 1  i  k,
(ii) bi 2 Z  0 =f0; 1; 2; 3; . . .g for 1  i  k,
(iii) ci 2 f0; 1g for 1  i  k,
(iv) if the weight of the triplet ða; b; cÞ is defined by
 ðk  1Þðt þ 3Þ  X k Xk Xk
wða; b; cÞ ¼ k 1 þ þ 2iai þ ð2i  1Þbi þ ð2i  1Þci ;
2 i¼1 i¼1 i¼1

then ða; b; cÞ should satisfy wða; b; cÞ ¼ l.

One can easily establish that the generating function for Ct1 ðk; lÞ is
X qk½1þðk1Þðtþ3Þ=2 ðq; q2 Þk
C1t ðk; lÞql ¼ ;
l
ðq2 ; q2 Þk ðq; q2 Þk

where C1t ðk; lÞ ¼ jCt1 ðk; lÞj. In the second and third step we will prove that there is
a bijection between the split lattice paths enumerated by Bt1 ðk; lÞ and the triplets
enumerated by C1t ðk; lÞ.
Proof Step I. We shall prove that
X
1 X
1 k½1þðk1Þðtþ3Þ=2
q ðq; q2 Þ
Bt1 ðlÞql ¼ k
: ð10Þ
l¼0 k¼0
ðq2 ; q2 Þk ðq; q2 Þk

qk½1þðk1Þðtþ3Þ=2 ðq;q2 Þk
In ðq2 ;q2 Þk ðq;q2 Þk the factor qk½1þðk1Þðtþ3Þ=2 generates a split lattice path having k
th
TITSs such that i TITS has the ordered pair f2i  1 þ ði  1Þðt þ 1Þ; 1g.
For k ¼ 3 and t ¼ 1, the path begins as shown in Fig. 7.

Fig. 7 Split lattice path for k ¼ 6

3 and t ¼ 1 5
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

123
1288 Graphs and Combinatorics (2020) 36:1273–1295

In Fig. 7, we consider two successive TITSs, say, i th and ði þ 1Þth. Their cor-
responding ordered pairs are f2i  1 þ ði  1Þðt þ 1Þ; 1g and f2i þ 1 þ iðt þ 1Þ; 1g
respectively.
The factor ðq2 ;q1 2 Þ generates k nonnegative multiples of 2, say a1  a2     ak  0,
k

which are encoded by increasing the altitude of the ith TITS by akiþ1 ; 1  i  k.
Thus the ordered pair associated with the ith TITS becomes
fð2i  1Þ þ ði  1Þðt þ 1Þ þ akiþ1 ; 1g. The ith and ði þ 1Þth TITSs will look like as
in Fig. 8:
The factor ðq;q12 Þ generates k nonnegative multiples of ð2i  1Þ, 1  i  k; say,
k

b1  1; b2  3;    ; bk  ð2k  1Þ: This is encoded by increasing the altitude of ith

TITS by 2ðbk þ bk1 þ    þ bkiþ2 Þ þ bkiþ1 and the length of the ray by bkiþ1 .
So the associated ordered pair becomes f2i  1 þ akiþ1 þ 2ðbk þ bk1
þ    þ bkiþ2 Þ þ bkiþ1 ; 1 þ bkiþ1 g. Now, Fig. 8 changes to Fig. 9.
The factor ðq; q2 Þk generates k distinct nonnegative multiples of ð2i  1Þ,
1  i  k; say, c1  1; c2  3; . . .; ck  ð2k  1Þ; where each ci ð1  i  kÞ is 0 or 1.
This is encoded by increasing the altitude of ith TITS by 2ðck þ ck1 þ    þ
ckiþ2 Þ þ ckiþ1 and by putting a segment of length ckiþ1 adjacent to the ray. So
the ordered pair becomes f2i  1 þ ði  1Þðt þ 1Þ þ akiþ1 þ 2ðbk þ bk1 þ    þ
bkiþ2 Þ þ bkiþ1 þ 2ðck þ ck1 þ    þ ckiþ2 Þ þ ckiþ1 ; 1 þ bkiþ1 þ ckiþ1 g.
Figure 9 now changes to Fig. 10
Every split lattice path enumerated by Bt1 ðlÞ is uniquely generated in this
manner. This proves (10).

Fig. 8 i th and ði þ 1Þth TITSs (i+1)

st

th
i

Fig. 9 i th and ði þ 1Þth TITSs

st
(i+1)

th
i

Fig. 10 i th and ði þ 1Þth TITSs

(i+1)st

th
i

123
Graphs and Combinatorics (2020) 36:1273–1295 1289

Furthermore, we see that constructively there

 exists a well-defined map between
the triplets ða; b; cÞ ¼ ðai Þki¼1 ; ðbi Þki¼1 ; ðci Þki¼1 of the set Ct1 ðk; lÞ and the k TITSs
represented by ordered pairs fai ; bi gki¼1 ¼ fai ; ri þ si gki¼1 of the set Bt1 ðk; lÞ, say,
/t1 : Ct1 ðk; lÞ ! Bt1 ðk; lÞ;
defined by
 
/t1 ðai Þki¼1 ; ðbi Þki¼1 ; ðci Þki¼1 ¼ fakiþ1 ; rkiþ1 þ skiþ1 gki¼1 ;

such that, for all 1  i  k


bi ¼ rkiþ1  1;
/t1 : ð11Þ
ci ¼ skiþ1 :
and akiþ1 ¼ 2ðk  i þ 1Þ  1 þ ðk  iÞðt þ 1Þ þ ai þ 2ðbk þ bk1 þ    þ biþ1 Þ þ
bi þ 2ðck þ ck1 þ    þ ciþ1 Þ þci .
Step II. We now show that /t1 ða; b; cÞ is actually an element of Bt1 ðk; lÞ. Let us
denote the ith and ði þ 1Þth TITS in Fig. 10 by fai ; bi g and faiþ1 ; biþ1 g respectively,
then
ai ¼ 2i  1 þ ði  1Þðt þ 1Þ þ akiþ1 þ 2ðbk þ bk1 þ    þ bkiþ2 Þ
þ bkiþ1 þ 2ðck þ ck1 þ    þ ckiþ2 Þ þ ckiþ1 ;
bi ¼ 1 þ bkiþ1 þ ckiþ1 ;
ð12Þ
aiþ1 ¼ ð2i þ 1Þ þ iðt þ 1Þ þ aki þ 2ðbk þ bk1 þ    þ bkiþ1 Þ
þ bki þ 2ðck þ ck1 þ    þ ckiþ1 Þ þ cki ;
biþ1 ¼ 1 þ bki þ cki :
Clearly, ai  bi and the parity of ai and bi depends upon bkiþ1 þ ckiþ1 . If bkiþ1 þ
ckiþ1 is odd then both ai and bi are even and when bkiþ1 þ ckiþ1 is even then
both ai and bi are odd. This confirms that for any TITS with ordered pair {a, b}, b
does not exceed a and a  bðmod 2Þ.
Also, we know from condition (iii) on the triplets of the set Ct1 ðk; lÞ, ci 2 f0; 1g
for 1  i  k. Thus using (11), there cannot be any segment with length greater than
1 in the split lattice paths of the set Bt1 ðk; lÞ. Now using (12), we consider
aiþ1  biþ1 ¼ ai þ bi þ t þ 1 þ aki þ akiþ1 [ t.
Step III. Finally we show that /t1 is injective as well as surjective.
1. /t1 is 1-1:
   
Suppose ðai Þki¼1 ; ðbi Þki¼1 ; ðci Þki¼1 and ðdi Þki¼1 ; ðfi Þki¼1 ; ðgi Þki¼1 be any two
triplets in the set Ct1 ðk; lÞ such that:

123
1290 Graphs and Combinatorics (2020) 36:1273–1295

   
/t1 ðai Þki¼1 ; ðbi Þki¼1 ; ðci Þki¼1 ¼ /t1 ðdi Þki¼1 ; ðfi Þki¼1 ; ðgi Þki¼1
) fakiþ1 ; rkiþ1 þ skiþ1 g ¼ fa0kiþ1 ; rkiþ1
0
þ s0kiþ1 g; for 1  i  k
ð13Þ
We know by Definition 6 two TITSs are equal if their weights as well as the
lengths of their rays and segments are equal. Thus from (13) we have
0
rkiþ1 ¼ rkiþ1 and skiþ1 ¼ s0kiþ1
) 1 þ bi ¼ 1 þ fi and ci ¼ gi ; using (11)
) bi ¼ fi and ci ¼ gi ; for 1  i  k
) ai ¼ di ; for 1  i  k
   
) ðai Þki¼1 ; ðbi Þki¼1 ; ðci Þki¼1 ¼ ðdi Þki¼1 ; ðfi Þki¼1 ; ðgi Þki¼1 :

Hence /t1 is injective.

2. /t1 is surjective:
Consider TITSs with ordered pairs fai ; bi g for 1  i  k of split lattice paths
enumerated by Bt1 ðk; lÞ, where bi ¼ ri þ si for 1  i  k. Now by the conditions
of Theorem 4 on the split lattice paths in the set Bt1 ðk; lÞ, we know every ri  1,
si ¼ 0 or 1 for 1  i  k. So, by definition of the map /t1 , we have bi ¼

rkiþ1  1  0 and ci ¼ skiþ1 ¼ 0 or 1 for 1  i  k. Thus ðai Þki¼1 ; ðbi Þki¼1 ;

ðci Þki¼1 is a triplet in the set Ct1 ðk; lÞ. Hence /t1 is surjective.

And the inverse mapping is:

8
< ri ¼ bkiþ1 þ 1;
>
t 1
ð/1 Þ : for 1  i  k
>
:
si ¼ ckiþ1 :
and ai ¼ 2i  1 þ ði  1Þðt þ 1Þ þ akiþ1 þ 2ðbk þ bk1 þ    þ bkiþ2 Þþ bkiþ1 þ
2ðck þ ck1 þ    þ ckiþ2 Þ þckiþ1 . h

4.3 Outline of the Proofs of Theorems 5–6

Here, the changes required to prove the remaining theorems are discussed briefly.
Theorem 5: The proof of Theorem 5 is treated in the same manner as Theorem 4.
The only difference is that now the path begins with an SS of height 1 or a TITS
with ordered pair fa; a þ 1g. Here we define the set of triplets ða; b; cÞ denoted by
C2 ðk; lÞ where a ¼ ða1 ; a2 ; . . .; ak Þ, c ¼ ðc1 ; c2 ; . . .; ck Þ are k-tuples of integers and
b ¼ ðb1 ; b2 ; . . .; bkþ1 Þ is ðk þ 1Þ-tuple of integers and satisfy:
(i) ai 2 Z  0 =f0; 1; 2; 3; . . .g for 1  i  k,
(ii) bi 2 Z  0 =f0; 1; 2; 3; . . .g for 1  i  k þ 1,
(iii) ci 2 f0; 1g for 1  i  k,
(iv) if the weight of the triplet ða; b; cÞ is defined by

123
Graphs and Combinatorics (2020) 36:1273–1295 1291

kðk þ 1Þðt þ 3Þ X k X
kþ1 Xk
wða; b; cÞ ¼ þ 2iai þ ð2i  1Þbi þ ð2i  1Þci ;
2 i¼1 i¼1 i¼1

then ða; b; cÞ should satisfy wða; b; cÞ ¼ l.

Clearly, the generating function for Ct2 ðk; lÞ will be

X
Ct2 ðk; lÞql ¼ f2t ðqÞ:
l

We will consider two cases to establish a bijection between the set of triplets
Ct2 ðk; lÞ and the set of split lattice paths Bt2 ðk; lÞ.
kðtþ2Þ
q
1. Case 1.: When bkþ1 ¼ 0, then the factor ð1q 2kþ1 Þ makes the split lattice path to

begin with an SS of height 1 which clearly corresponds to the part 01 in the

corresponding split ðn þ 1Þ-color partition.
kðtþ2Þ
q
2. Case 2.: When bkþ1 6¼ 0, then the factor ð1q 2kþ1 Þ makes the split lattice path to

begin with a TITS with ordered pair {bkþ1 ; 1 þ bkþ1 } which clearly corresponds
to the part aaþ1 in the corresponding split ðn þ 1Þ-color partition.

In both the cases, we define the map

/t2 : Ct2 ðk; lÞ ! Bt2 ðk; lÞ;

such that, for all 1  i  k


bi ¼ rkiþ2  1;
/t2 : ð14Þ
ci ¼ skiþ2 :
and akiþ2 ¼ 2ðk  i þ 1Þ  1 þ ðk  iÞðt þ 1Þ þ ai þ 2ðbk þ bk1 þ    þ biþ1 Þ þ
bi þ 2ðck þ ck1 þ    þ ciþ1 Þ þci .
By using the similar arguments as used in the Proof of Theorem 4, one can easily
verify that the map /t2 is bijective and the inverse map is given by:
8
< riþ1 ¼ bkiþ1 þ 1;
>
ð/t2 Þ1 : for 1  i  k; ð15Þ
>
:
siþ1 ¼ ckiþ1 :
Theorem 6: The proof of Theorem 6 is also treated in the same manner as Theo-
rem 4. Here, the only difference is that now the path begins with an SS of height 2
or a TITS with ordered pair fa; a þ 2g. We define the set of triplets ða; b; cÞ denoted
by Ct3 ðk; lÞ where a ¼ ða1 ; a2 ; . . .; ak Þ, c ¼ ðc1 ; c2 . . .; ck Þ are k-tuples of integers
and b ¼ ðb1 ; b2 ; . . .; bkþ1 Þ is ðk þ 1Þ-tuple of integers and satisfy:
(i) ai 2 Z  0 =f0; 1; 2; 3; . . .g for 1  i  k,
(ii) bi 2 Z  0 =f0; 1; 2; 3; . . .g for 1  i  k þ 1,
(iii) ci 2 f0; 1g for 1  i  k,

123
1292 Graphs and Combinatorics (2020) 36:1273–1295

(iv) if the weight of the triplet ða; b; cÞ is defined by

 ðk þ 1Þðt þ 3Þ  X k X
kþ1 Xk
wða; b; cÞ ¼ k 1 þ þ 2iai þ ð2i  1Þbi þ ð2i  1Þci ;
2 i¼1 i¼1 i¼1

then ða; b; cÞ should satisfy wða; b; cÞ ¼ l.

Clearly, the generating function for Ct3 ðk; lÞ will be

X
Ct3 ðk; lÞql ¼ f3t ðqÞ:
l

Again we will consider two case to establish a bijection between the set of triplets
Ct3 ðk; lÞ and the set of split lattice paths Bt3 ðk; lÞ.
kðtþ3Þ
q
1. Case 1.: When bkþ1 ¼ 0, then the factor ð1q 2kþ1 Þ makes the split lattice path to

begin with an SS of height 2 which clearly corresponds to the part 02 in the

corresponding split ðn þ 2Þ-color partition.
kðtþ3Þ
q
2. Case 2.: When bkþ1 6¼ 0, then the factor ð1q 2kþ1 Þ makes the split lattice path to

begin with a TITS with ordered pair fbkþ1 ; 2 þ bkþ1 g which clearly corresponds
to the part aaþ2 in the corresponding split ðn þ 2Þ-color partition.

In both the cases, we define the map

/t3 : Ct3 ðk; lÞ ! Bt3 ðk; lÞ
as defined in (14) and the inverse mapping as defined in (15).

5 Rogers–Ramanujan Type Identities

For some particular values of t, Theorems 4–6 enable us to provide new

combinatorial meanings to the following Rogers–Ramanujan type identities.
X
1 n2
q ðq; q2 Þ Y
1
ð1 þ q2n1 Þð1 þ q6n2 Þð1 þ q6n4 Þð1  q6n Þ
n
¼ ; ð16Þ
n¼0
ðq; qÞ2n n¼1
ð1  q2n Þ

X
1 nðnþ1Þ
q ðq; q2 Þ Y
1
ð1  q12n4 Þð1  q12n8 Þð1  q12n Þ
n
¼ ; ð17Þ
n¼0
ðq; qÞ2nþ1 n¼1
ð1  qn Þ

X
1 nðnþ2Þ
q ðq; q2 Þ Y
1
ð1  q12n2 Þð1  q12n10 Þð1  q12n Þ
n
¼ : ð18Þ
n¼0
ðq; qÞ2nþ1 n¼1
ð1  qn Þ

Identities (16)–(18) are appearing in the Slater’s long list of Rogers–Ramanujan type
identities [16, p.154, I(29), I(51), I(50)]. Now an appeal to Theorems 4–6 give the
following 3-way combinatorial interpretations of Identities (16)–(18), respectively:

123
Graphs and Combinatorics (2020) 36:1273–1295 1293

Theorem 7 Let D1 ðlÞ denote the number of partitions of l such that the odd parts
are distinct, even parts are  2; 4ðmod 12Þ and two colors of the parts which
are  2ðmod 12Þ are allowed. Then
X1 X
1 X
1
D1 ðlÞql ¼ A1
1 ðlÞq l
¼ B1 l
1 ðlÞq ;
l¼0 l¼0 l¼0

where A1 1
1 ðlÞ and B1 ðlÞ are as defined in Theorem 4 for t ¼ 1. The following
Table 2 illustrates this theorem more precisely.

Theorem 8 Let D2 ðlÞ denote the number of partitions of l such that the parts are
6 0; 4ðmod 12Þ. Then
X
1 X
1 X
1
D2 ðlÞql ¼ A1 l
2 ðlÞq ¼ B1 l
2 ðlÞq ;
l¼0 l¼0 l¼0

where A1 1
2 ðlÞ and B2 ðlÞ are as defined in Theorem 5 for t ¼ 1.

Theorem 9 Let D3 ðlÞ denote the number of partitions of l such that the parts are
6 0; 2ðmod 12Þ. Then
X
1 X
1 X
1
D3 ðlÞql ¼ A1 l
3 ðlÞq ¼ B1 l
3 ðlÞq ;
l¼0 l¼0 l¼0

where A1 1
3 ðlÞ and B3 ðlÞ are as defined in Theorem 6 for t ¼ 1.

Table 2 Split n-color partitions and split lattice paths for Theorem 7

123
1294 Graphs and Combinatorics (2020) 36:1273–1295

6 Conclusion

Elegant connections between different combinatorial parameters are observed in

this paper. Theorems 2–3 enable us to look at the Gordon–McIntosh eight order
mock theta functions in a new combinatorial way. The Theorems 4–6 yield three
infinite classes of combinatorial identities. The new combinatorial interpretations of
Rogers–Ramanujan type identities in terms of split lattice paths reveal the fact that
this new combinatorial object is quite helpful to understand the graphical aspect of
‘‘Sum-Product’’ identities. It would be of interest if more q-series identities can be
interpreted combinatorially using split lattice paths.

References
1. Agarwal, A.K., Andrews, G.E.: Rogers–Ramanujan identities for partitions with ‘‘N copies of N’’.
J. Combin. Theory Ser. A 45(1), 40–49 (1987)
2. Agarwal, A.K., Bressoud, D.M.: Lattice paths and multiple basic hypergeometric series. Pac. J. Math.
136(2), 209–228 (1989)
3. Agarwal, A.K., Goyal, M.: Lattice paths and Rogers identities. Open J. Discr. Math. 1, 89–95 (2011)
4. Agarwal, A.K., Goyal, M.: New partition theoretic interpretations of Rogers–Ramanujan identities.
Int. J. Combin. 2012, 6 (2012)
5. Agarwal, A.K. Goyal, M.: On 3-way combinatorial identities. Proc. Indian Acad. Sci. Math. Sci.
(2018). https://doi.org/10.1007/s12044-018-0378-3
6. Agarwal, A.K., Sood, G.: Split ðn þ tÞ-color partitions and Gordon-M-cIntosh eight order mock theta
functions. Electron. J. Combin. 21(2), 2.46 (2014)
7. Anand, S., Agarwal, A.K.: A new class of lattice paths and partitions with n copies of n. Proc. Indian
Acad. Sci. Math. Sci. 122(1), 23–39 (2012)
8. Connor, W.G.: Partition theorems related to some identities of Rogers and Watson. Trans. Am. Math.
Soc. 214, 95–111 (1975)
9. Ferrari, L.: Some combinatorics related to central binomial coefficients: Grand-Dyck paths, coloured
noncrossing Partitions and Signed Pattern Avoiding Permutations. Graphs Combin. 26(1), 51–70
(2010)
10. Gordon, B., McIntosh, R.J.: Some eight order mock theta functions. J. Lond. Math. Soc. 62(2),
321–335 (2000)
11. Goyal, M.: New combinatorial interpretations of some Rogers–Ramanujan type identities. Contrib.
Discr. Math. 11(2), 43–57 (2017)
12. Goyal, M.: On combinatorial extensions of Rogers–Ramanujan type identities. Contrib. Discr. Math.
12, 33–51 (2017)
13. Goyal, M.: Rogers–Ramanujan type identities for split ðn þ tÞ-color partitions. Math. Rep. 21(71), 3,
339–349 (2019)
14. Goyal, M., Agarwal, A.K.: Further Rogers–Ramanujan identities for n-color partitions. Utilitas Math.
95, 141–148 (2014)
15. Goyal, M., Agarwal, A.K.: On a new class of combinatorial identities. ARS Combin. 127, 65–77
(2016)
16. Slater, L.J.: Further identities of the Rogers–Ramanujan type. Proc. Lond. Math. Soc. 54, 147–167
(1952)
17. Sood, G., Agarwal, A.K.: Rogers–Ramanujan identities for split ðn þ tÞ-color partitions. J. Combin.
Number Theory. 7(2), 141–151 (2015)
18. Subbarao, M.V.: Some Rogers–Ramanujan type partition theorems. Pac. J. Math. 120, 431–435
(1985)
19. Wenchang, C.: A partition identity on binomial coefficients and its applications. Graphs Combin.
5(1), 197–200 (1989)

123
Graphs and Combinatorics (2020) 36:1273–1295 1295

20. Yan, S.H.F., Zhang, Y.: On lattice paths with four types of steps. Graphs Combin. 31(4), 1077–1084
(2014)

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

123