Journal of Computational and Applied Mathematics 473 (2026) 116861

Contents lists available at ScienceDirect

Journal of Computational and Applied Mathematics

journal homepage: [Link]/locate/cam

The exact region determined by Blomqvist’s beta, Spearman’s

footrule and Gini’s gamma
∗
Damjana Kokol Bukovšek , Blaž Mojškerc
University of Ljubljana, School of Economics and Business, and Institute of Mathematics, Physics and Mechanics, Ljubljana, Slovenia

ARTICLE INFO ABSTRACT

MSC: To quantify the degree of association between random variables, concordance measures are
62H05 employed. To express such a degree, a single measure might give too much space, so several
62H20 are used for comparison. In this paper we study the ternary relation between three well-known
60E05
(weak) concordance measures, namely Blomqvist’s beta, Spearman’s footrule and Gini’s gamma.
Keywords: In other words, given the values of Blomqvist’s beta and Spearman’s footrule, we determine the
Copula degree of freedom a copula has at taking the value of Gini’s gamma. We explicitly determine
Dependence concepts
the 3-dimensional region representing the relation. We also provide copulas where bounds of
Concordance measure
the region are attained.
Gini’s gamma
Blomqvist’s beta
Spearman’s footrule

1. Introduction

Two continuous random variables are concordant when large values of the first are associated with large values of the second. It
is often more appropriate to study the association of random variables instead of their linear correlation due to the invariance
of measures of concordance to monotonously increasing transformations. This directly implies that measures of concordance
are independent of marginal distributions of continuous random variables. Rather, they rely solely on the copula modeling the
dependence of the two random variables.
The most commonly used concordance measures are Spearman’s rho, Kendall’s tau, Gini’s gamma and Blomqvist’s beta, and
a weak concordance measure Spearman’s footrule. These measures have been studied extensively since their introduction. Recent
references for bivariate concordance measures include [1–7] and their multivariate generalizations were studied in [8–11], to name
just a few.
Recently, studying binary relations between five most common (weak) measures of concordance on copulas has become
increasingly popular. Since Spearman’s rho and Kendall’s tau are the most common and well known concordance measures, it is
only natural to try and determine their relation. The problem of the exact region determined by them was open from 1960’s and was
solved only recently by Schreyer, Paulin and Trutschnig in [12]. The relation between Blomqvist’s beta and other (possibly weak)
measures of concordance is relatively easy to tackle due to the nature of beta, while relations between others might and do prove to
be more challenging. The region determined by Gini’s gamma and Spearman’s footrule is characterized in [13]. The problem of the
region determined by Spearman’s rho and Spearman’s footrule is partially solved in [14], see also [15]. Characterizing the upper
bound of the region appears to be a very hard problem that has so far evaded the solution. The region determined by Spearman’s
rho and Gini’s gamma is still open.

∗ Corresponding author.
E-mail addresses: [Link]@[Link] (D. Kokol Bukovšek), [Link]@[Link] (B. Mojškerc).

[Link]
Received 27 September 2024; Received in revised form 9 May 2025
Available online 24 June 2025
0377-0427/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC license
([Link]
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

In this paper we take a step further and characterize the exact region representing the ternary relation between Blomqvist’s beta,
Spearman’s footrule and Gini’s gamma. Given the values of Blomqvist’s beta and Spearman’s footrule for some copula, we give the
lower and the upper bound that Gini’s gamma can take. We also provide copulas where these bounds are attained.

2. Preliminaries

Let I be the unit interval [0, 1] ⊆ R and 𝑥1 , 𝑥2 , 𝑦1 , 𝑦2 ∈ I be such that 𝑥1 ≤ 𝑥2 and 𝑦1 ≤ 𝑦2 . The Cartesian product of
intervals 𝐵 = [𝑥1 , 𝑥2 ] × [𝑦1 , 𝑦2 ] is called a rectangle in I2 . Let 𝐻 ∶ I2 → R be a real function. We define the 𝐻-volume of 𝐵 as
Vol𝐻 (𝐵) = 𝐻(𝑥2 , 𝑦2 ) − 𝐻(𝑥2 , 𝑦1 ) − 𝐻(𝑥1 , 𝑦2 ) + 𝐻(𝑥1 , 𝑦1 ). A bivariate copula is a function 𝐶 ∶ I2 → I with the following properties:

• 𝐶(0, 𝑦) = 𝐶(𝑥, 0) = 0 for all 𝑥, 𝑦 ∈ I (𝐶 is grounded),

• 𝐶(𝑥, 1) = 𝑥 and 𝐶(1, 𝑦) = 𝑦 for all 𝑥, 𝑦 ∈ I (𝐶 has uniform marginals), and
• Vol𝐶 (𝐵) ≥ 0 for every rectangle 𝐵 ⊆ I2 (𝐶 is 2−increasing ).

Bivariate copulas are therefore functions of two variables which couple bivariate distribution functions with their one-dimensional
marginal distribution functions, a famous Theorem by Sklar [16].
Let  be the set of all bivariate copulas. We introduce some transformations that are naturally defined on : We denote by
𝐶 𝑡 the transpose of the copula 𝐶, i.e., 𝐶 𝑡 (𝑥, 𝑦) = 𝐶(𝑦, 𝑥). By 𝐶 𝜎1 and 𝐶 𝜎2 we denote the two reflections of a copula 𝐶 defined by
𝐶 𝜎1 (𝑥, 𝑦) = 𝑦 − 𝐶(1 − 𝑥, 𝑦) and 𝐶 𝜎2 (𝑥, 𝑦) = 𝑥 − 𝐶(𝑥, 1 − 𝑦) (see [17, §1.7.3]), and by 𝐶̂ = (𝐶 𝜎1 )𝜎2 the survival copula of 𝐶. We write
𝐶 ≤ 𝐷 if 𝐶(𝑥, 𝑦) ⩽ 𝐷(𝑥, 𝑦) for all (𝑥, 𝑦) ∈ I2 . This is the so-called pointwise order of copulas. It is well known that  is a partially
ordered set with respect to the order, but not a lattice [18, Theorem 2.1], and that 𝑊 (𝑥, 𝑦) = max{0, 𝑥+𝑦−1} and 𝑀(𝑥, 𝑦) = min{𝑥, 𝑦}
are the lower and upper bounds of all copulas, respectively. Copulas 𝑊 and 𝑀 are called Fréchet-Hoeffding lower and upper bounds.
We denote the main diagonal of copula 𝐶 by 𝛿𝐶 and its opposite diagonal by 𝜔𝐶 , i.e.,

𝛿𝐶 (𝑥) = 𝐶(𝑥, 𝑥), and 𝜔𝐶 (𝑥) = 𝐶(𝑥, 1 − 𝑥).

The function 𝛿𝐶 satisfies 0 ≤ 𝛿𝐶 (𝑥) ≤ 𝑥, 𝛿𝐶 (1) = 1 and it is increasing and 2-Lipschitz. The function 𝜔𝐶 satisfies 𝜔𝐶 (0) = 𝜔𝐶 (1) = 0
and it is nonnegative and 1-Lipschitz (see [19]).
A mapping 𝜅 ∶  → [−1, 1] is called a concordance measure if it satisfies the following properties (see [17, Definition 2.4.7]):

(C1) 𝜅(𝐶) = 𝜅(𝐶 𝑡 ) for every 𝐶 ∈ .

(C2) 𝜅(𝐶) ⩽ 𝜅(𝐷) when 𝐶 ⩽ 𝐷.
(C3) 𝜅(𝑀) = 1.
(C4) 𝜅(𝐶 𝜎1 ) = 𝜅(𝐶 𝜎2 ) = −𝜅(𝐶).
(C5) If a sequence of copulas 𝐶𝑛 , 𝑛 ∈ N, converges pointwise to 𝐶 ∈ , then lim𝑛→∞ 𝜅(𝐶𝑛 ) = 𝜅(𝐶).

Certain properties that are sometimes stated in definitions of a concordance measure follow from the properties listed above.
Indeed, a concordance measure also satisfies the following properties (see [4, §3] for more details):

(C6) 𝜅(𝛱) = 0, where 𝛱 is the independence copula 𝛱(𝑢, 𝑣) = 𝑢𝑣.

(C7) 𝜅(𝑊 ) = −1.
̂ for every 𝐶 ∈ .
(C8) 𝜅(𝐶) = 𝜅(𝐶)

The four most commonly used concordance measures of a copula 𝐶 are Spearman’s rho, Kendall’s tau, Gini’s gamma, and
Blomqvist’s beta. If we replace property (C4) with property (C6) in the definition of a concordance measure, we get a weak
concordance measure (see [5]). Spearman’s footrule is an example of a weak concordance
[ ] measure. The range of a concordance
measure is the interval [−1, 1], while the range of Spearman’s footrule is equal to − 12 , 1 (see [11, §4]). In this paper we are going
to consider the relation between Blomqvist’s beta (𝛽), Spearman’s footrule (𝜙) and Gini’s gamma (𝛾). They are defined as follows:
( )
𝛽(𝐶) = 4 𝐶 12 , 12 − 1, (1)

1
𝜙(𝐶) = 6 𝛿𝐶 (𝑥)𝑑𝑥 − 2, (2)
∫0
1 1
𝛾(𝐶) = 4 𝛿 (𝑥)𝑑𝑥 + 4 𝜔 (𝑥)𝑑𝑥 − 2. (3)
∫0 𝐶 ∫0 𝐶
In recent years the relations between pairs of (weak) concordance measures got lots of attention. The exact region determined
by Kendall’s tau and Spearman’s rho was determined in [12]. The regions determined by Blomqvist’s beta and Spearman’s rho,
Kendall’s tau, and Gini’s gamma are given in [20] as an exercise for the reader. The region determined by Blomqvist’s beta and
Spearman’s footrule was given in [21]. The region determined by Spearman’s footrule and Gini’s gamma was given in [13], the
regions determined by Spearman’s footrule and Kendall’s tau and determined by Gini’s gamma and Kendall’s tau in [22]. The region
between Spearman’s footrule and Spearman’s rho was partially determined in [14] (see also [15]). The region determined by Gini’s
gamma and Spearman’s rho is still open. Here we give the results we are going to need in the sequel.

2
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Proposition 2.1 ([13]). For any copula 𝐶 ∈  we have

4
3
𝜙(𝐶) − 13 ≤ 𝛾(𝐶) ≤ min{ 43 𝜙(𝐶) + 16 , 23 𝜙(𝐶) + 13 }.

The bounds are attained by shuffles of 𝑀.

The exact region determined by Spearman’s footrule and Gini’s gamma

𝛺𝜙,𝛾 = {(𝜙(𝐶), 𝛾(𝐶)) ∈ [− 12 , 1] × [−1, 1] ∶ 𝐶 ∈ }

is quadrilateral 𝐴1 𝐵1 𝐶1 𝐷1 with vertices 𝐴1 (− 12 , −1), 𝐵1 (1, 1), 𝐶1 ( 14 , 12 ), 𝐷1 (− 12 , − 12 ) depicted in Fig. 8.

Proposition 2.2. For any copula 𝐶 ∈  we have

3
16
(1 + 𝛽(𝐶))2 − 12 ≤ 𝜙(𝐶) ≤ 1 − 38 (1 − 𝛽(𝐶))2 , (4)
3 3
8
(1 + 𝛽(𝐶))2 − 1 ≤ 𝛾(𝐶) ≤ 1 − 8
(1 − 𝛽(𝐶))2 . (5)

The bounds are attained by shuffles of 𝑀.

3. Main result

In this section we locate the exact three-dimensional region determined by Blomqvist’s beta, Spearman’s footrule and Gini’s
gamma. In other words, we determine all possible triples (𝛽(𝐶), 𝜙(𝐶), 𝛾(𝐶)), where 𝐶 runs over the set of all copulas . We will
find the upper and lower bound for 𝛾(𝐶), if 𝛽(𝐶) and 𝜙(𝐶) are given and satisfy Proposition 2.2, and show that the bounds can be
attained. Proposition 2.1 already gives two bounds which will be part of our final result, namely,
4
3
𝜙(𝐶) − 13 ≤ 𝛾(𝐶) ≤ 43 𝜙(𝐶) + 16 . (6)

In next two propositions we prove two more bounds.

Proposition 3.1. For any copula 𝐶 ∈  we have

𝛾(𝐶) ≤ 23 𝜙(𝐶) + 14 𝛽(𝐶) − 18 𝛽(𝐶)2 + 5

24
.

Proof. Let 𝐶 be any copula and let 𝑏 = 𝜔𝐶 ( 12 ) = 𝐶( 12 , 12 ) = 𝛽(𝐶)+1 4

. Since 𝜔𝐶 is 1-Lipschitz, we have for any 𝑥 ∈ [0, 12 ] that
𝜔𝐶 ( 2 ) − 𝜔𝐶 (𝑥) ≥ −( 2 − 𝑥), so that 𝜔𝐶 (𝑥) ≤ −𝑥 + 𝑏 + 2 . Furthermore, for any 𝑥 ∈ [ 12 , 1] 1-Lipschitz property implies that
1 1 1

𝜔𝐶 (𝑥) − 𝜔𝐶 ( 12 ) ≤ 𝑥 − 12 , and thus 𝜔𝐶 (𝑥) ≤ 𝑥 + 𝑏 − 12 . Since 𝜔𝐶 (0) = 𝜔𝐶 (1) = 0, we have also 𝜔(𝑥) ≤ min{𝑥, 1 − 𝑥} for any 𝑥 ∈ I.
It follows that
⎧𝑥; if 0 ≤ 𝑥 ≤ 1
+ 2𝑏 ,
⎪ 1 1
4
⎪−𝑥 + 𝑏 + 2 ; if 4
+ 𝑏
2
< 𝑥 ≤ 12 ,
𝜔𝐶 (𝑥) ≤ 𝜔0 (𝑥) = ⎨ 1 1 3
⎪𝑥 + 𝑏 − 2 ; if 2
<𝑥≤ 4
− 2𝑏 ,
⎪1 − 𝑥; if 3
− 𝑏
< 𝑥 ≤ 1.
⎩ 4 2
1 𝜙(𝐶)+2
From equality (2) it follows that ∫0 𝛿𝐶 (𝑥)𝑑𝑥 = 6
. Now,
1 1
𝛾(𝐶) = 4 𝛿𝐶 (𝑥)𝑑𝑥 + 4 𝜔𝐶 (𝑥)𝑑𝑥 − 2
∫0 ∫0
1 1
≤4 𝛿 (𝑥)𝑑𝑥 + 4 𝜔 (𝑥)𝑑𝑥 − 2
∫0 𝐶 ∫0 0
𝜙(𝐶) + 2 ( )
=4⋅ + 4 ⋅ 12 𝑏 − 12 𝑏2 + 18 − 2
6
= 23 𝜙(𝐶) + 2𝑏 − 2𝑏2 − 16
= 23 𝜙(𝐶) + 14 𝛽(𝐶) − 18 𝛽(𝐶)2 + 5
24
,

which finishes the proof. □

Proposition 3.2. For any copula 𝐶 ∈  we have

𝛾(𝐶) ≥ 23 𝜙(𝐶) + 12 𝛽(𝐶) + 14 𝛽(𝐶)2 − 5

12
.

3
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Proof. As above, for any copula 𝐶 let 𝑏 = 𝜔𝐶 ( 12 ) = 𝐶( 12 , 12 ) = 𝛽(𝐶)+1 4

. Since 𝜔𝐶 is 1-Lipschitz, we have for any 𝑥 ∈ [0, 12 ]
that 𝜔𝐶 ( 2 ) − 𝜔𝐶 (𝑥) ≤ 2 − 𝑥, so that 𝜔𝐶 (𝑥) ≥ 𝑥 + 𝑏 − 2 . Furthermore, for any 𝑥 ∈ [ 12 , 1] 1-Lipschitz property implies that
1 1 1

𝜔𝐶 (𝑥) − 𝜔𝐶 ( 12 ) ≥ −(𝑥 − 12 ), and thus 𝜔𝐶 (𝑥) ≤ −𝑥 + 𝑏 + 12 . Since 𝜔𝐶 is nonnegative, it follows that

⎧0; if 0 ≤ 𝑥 ≤ 1
− 𝑏,
⎪ 1 1
2
⎪𝑥 + 𝑏 − 2 ; if 2
− 𝑏 < 𝑥 ≤ 12 ,
𝜔𝐶 (𝑥) ≥ 𝜔1 (𝑥) = ⎨ 1 1 1
⎪−𝑥 + 𝑏 + 2 ; if 2
<𝑥≤ 2
+ 𝑏,
⎪0; if 1
+ 𝑏 < 𝑥 ≤ 1.
⎩ 2

Now,
1 1
𝛾(𝐶) ≥ 4 𝛿 (𝑥)𝑑𝑥 + 4 𝜔 (𝑥)𝑑𝑥 − 2
∫0 𝐶 ∫0 1
𝜙(𝐶) + 2
=4⋅ + 4 ⋅ 𝑏2 − 2
6
= 23 𝜙(𝐶) + 4𝑏2 − 23
= 23 𝜙(𝐶) + 12 𝛽(𝐶) + 14 𝛽(𝐶)2 − 5
12
,

which finishes the proof. □

In next examples we give six families of shuffles of 𝑀, which attain the proved bounds. A shuffle of 𝑀

𝑀(𝑛, 𝐽 , 𝜋, 𝜀)

is determined by a positive integer 𝑛, a partition of interval I into 𝑛 pieces

𝐽 = {[0, 𝑥1 ], [𝑥1 , 𝑥2 ], … , [𝑥𝑛−2 , 𝑥𝑛−1 ], [𝑥𝑛−1 , 1]},

shortly written as an (𝑛 − 1)-tuple 𝐽 = (𝑥1 , 𝑥2 , … , 𝑥𝑛−1 ) where 0 ≤ 𝑥1 ≤ 𝑥2 ≤ ... ≤ 𝑥𝑛−1 ≤ 1, a permutation 𝜋 ∈ 𝑆𝑛 , written as an 𝑛-tuple
of images 𝜋 = (𝜋(1), 𝜋(2), … , 𝜋(𝑛)), and a mapping 𝜀 ∶ {1, 2, … , 𝑛} → {−1, 1}, written as an 𝑛-tuple of images 𝜀 = (𝜀(1), 𝜀(2), … , 𝜀(𝑛)).
For more details see [20, §3.2.3]. Notice that we allow some of the intervals in the partition 𝐽 to be singletons.

Example 3.3. Let 𝑏 ∈ [0, 12 ] and let 𝐶𝑏 be a shuffle of 𝑀

𝐶𝑏 = 𝑀(4, ( 12 − 𝑏, 12 , 1
2
+ 𝑏), (4, 2, 3, 1), (−1, −1, −1, −1)).

Notice that 𝐶0 = 𝑊 and 𝐶 1 = 𝑀(2, 12 , (1, 2), (−1, −1)). We have 𝐶𝑏 ( 12 , 12 ) = 𝑏,

2

⎧0; if 0 ≤ 𝑥 ≤ 1
− 2𝑏 ,
⎪ 1
2
⎪2𝑥 + 𝑏 − 1; if 2
− 𝑏
2
< 𝑥 ≤ 12 ,
𝛿𝐶𝑏 (𝑥) = ⎨ 1 1
⎪𝑏; if 2
<𝑥≤ 2
+ 2𝑏 ,
⎪2𝑥 − 1; if 1
+ 𝑏
< 𝑥 ≤ 1,
⎩ 2 2

and
⎧0; if 0 ≤ 𝑥 ≤ 1
− 𝑏,
⎪ 1 1
2
⎪𝑥 + 𝑏 − 2 ; if 2
− 𝑏 < 𝑥 ≤ 12 ,
𝜔𝐶𝑏 (𝑥) = ⎨ 1 1 1
⎪−𝑥 + 𝑏 + 2 ; if 2
<𝑥≤ 2
+ 𝑏,
⎪0; if 1
+ 𝑏 < 𝑥 ≤ 1.
⎩ 2

The mass distribution of copula 𝐶𝑏 , the graphs of functions 𝛿𝐶𝑏 and 𝜔𝐶𝑏 , and 3D plot of copula 𝐶𝑏 are depicted in Fig. 1. It follows
that
1
𝛽(𝐶𝑏 ) = 4𝑏 − 1, 𝜙(𝐶𝑏 ) = 3𝑏2 − 2
and 𝛾(𝐶𝑏 ) = 6𝑏2 − 1,

so that
3 1
𝜙(𝐶𝑏 ) = 16
(1 + 𝛽(𝐶𝑏 ))2 − 2
and 𝛾(𝐶𝑏 ) = 23 𝜙(𝐶𝑏 ) + 12 𝛽(𝐶𝑏 ) + 14 𝛽(𝐶𝑏 )2 − 5
12
,

and both the lower bound from inequality (4) and Proposition 3.2 are attained.

Example 3.4. Let 𝑏 ∈ [0, 12 ] and let 𝐶𝑏 be a shuffle of 𝑀

𝐷𝑏 = 𝑀(2, ( 12 − 𝑏, 1
2
+ 𝑏), (3, 2, 1), (−1, 1, −1)).

4
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Fig. 1. The mass distribution of copula 𝐶𝑏 (top left), the graphs of functions 𝛿𝐶𝑏 and 𝜔𝐶𝑏 (top right), and a 3D plot of copula 𝐶𝑏 (bottom) from Example 3.3.

As in the previous example, 𝐷0 = 𝑊 , but 𝐷 1 = 𝑀. We have 𝐷𝑏 ( 12 , 12 ) = 𝑏,

2

⎧0; if 0 ≤ 𝑥 ≤ 1
− 𝑏,
⎪ 2
𝛿𝐷𝑏 (𝑥) = ⎨𝑥 + 𝑏 − 1 ; if 1
−𝑏<𝑥≤ 1
+ 𝑏,
2 2 2
⎪ 1
⎩2𝑥 − 1; if 2
+ 𝑏 < 𝑥 ≤ 1,
and
⎧0; if 0 ≤ 𝑥 ≤ 1
− 𝑏,
⎪ 1 1
2
⎪𝑥 + 𝑏 − 2 ; if 2
− 𝑏 < 𝑥 ≤ 12 ,
𝜔𝐷𝑏 (𝑥) = ⎨ 1 1 1
⎪−𝑥 + 𝑏 + 2 ; if 2
<𝑥≤ 2
+ 𝑏,
⎪0; if 1
+ 𝑏 < 𝑥 ≤ 1.
⎩ 2
Fig. 2 depicts the mass distribution of copula 𝐷𝑏 , the graphs of functions 𝛿𝐷𝑏 and 𝜔𝐷𝑏 , and 3D plot of copula 𝐷𝑏 . It follows that
1
𝛽(𝐷𝑏 ) = 4𝑏 − 1, 𝜙(𝐷𝑏 ) = 6𝑏2 − 2
and 𝛾(𝐷𝑏 ) = 8𝑏2 − 1,
so that
𝛾(𝐷𝑏 ) = 43 𝜙(𝐷𝑏 ) − 1
3
= 23 𝜙(𝐷𝑏 ) + 12 𝛽(𝐷𝑏 ) + 14 𝛽(𝐷𝑏 )2 − 5
12
,
and both the lower bound from inequality (6) and Proposition 3.2 are attained.

5
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Fig. 2. The mass distribution of copula 𝐷𝑏 (top left), the graphs of functions 𝛿𝐷𝑏 and 𝜔𝐷𝑏 (top right), and a 3D plot of copula 𝐷𝑏 (bottom) from Example 3.4.

Example 3.5. Let 𝑏 ∈ [0, 12 ] and let 𝐸𝑏 be a shuffle of 𝑀

𝐸𝑏 = 𝑀(3, (𝑏, 1 − 𝑏), (1, 2, 3), (1, −1, 1)).

Again 𝐸0 = 𝑊 and 𝐸 1 = 𝑀. We have 𝐸𝑏 ( 12 , 12 ) = 𝑏,

2

⎧𝑥; if 0 ≤ 𝑥 ≤ 𝑏,
⎪
⎪𝑏; if 𝑏 < 𝑥 ≤ 12 ,
𝛿𝐸𝑏 (𝑥) = ⎨ 1
⎪2𝑥 + 𝑏 − 1; if 2
< 𝑥 ≤ 1 − 𝑏,
⎪𝑥; if 1 − 𝑏 < 𝑥 ≤ 1,
⎩
and
⎧𝑥; if 0 ≤ 𝑥 ≤ 𝑏,
⎪
𝜔𝐸𝑏 (𝑥) = ⎨𝑏; if 𝑏 < 𝑥 ≤ 1 − 𝑏,
⎪
⎩1 − 𝑥; if 1 − 𝑏 < 𝑥 ≤ 1.

The mass distribution of copula 𝐸𝑏 , the graphs of functions 𝛿𝐸𝑏 and 𝜔𝐸𝑏 , and 3D plot of copula 𝐸𝑏 can be found in Fig. 3. It follows
that
1
𝛽(𝐸𝑏 ) = 4𝑏 − 1, 𝜙(𝐸𝑏 ) = −6𝑏2 + 6𝑏 − 2
and 𝛾(𝐸𝑏 ) = −8𝑏2 + 8𝑏 − 1,

6
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Fig. 3. The mass distribution of copula 𝐸𝑏 (top left), the graphs of functions 𝛿𝐸𝑏 and 𝜔𝐸𝑏 (top right), and a 3D plot of copula 𝐸𝑏 (bottom) from Example 3.5.

so that

𝜙(𝐸𝑏 ) = 1 − 38 (1 − 𝛽(𝐸𝑏 ))2 and 𝛾(𝐸𝑏 ) = 43 𝜙(𝐸𝑏 ) − 13 ,

and the upper bound from inequality (4) and the lower bound from inequality (6) are attained.

Example 3.6. Let 𝑏 ∈ [0, 12 ] and let 𝐹𝑏 be a shuffle of 𝑀

𝐹𝑏 = 𝑀(4, (𝑏, 12 , 1 − 𝑏), (1, 3, 2, 4), (1, 1, 1, 1)).

Note that 𝐹0 = 𝑀(2, 12 , (2, 1), (1, 1)) and 𝐹 1 = 𝑀. We have 𝐹𝑏 ( 12 , 12 ) = 𝑏,

2

⎧𝑥; if 0 ≤ 𝑥 ≤ 𝑏,
⎪
⎪𝑏; if 𝑏 < 𝑥 ≤ 12 ,
𝛿𝐹𝑏 (𝑥) = ⎨ 1
⎪2𝑥 + 𝑏 − 1; if 2
< 𝑥 ≤ 1 − 𝑏,
⎪𝑥; if 1 − 𝑏 < 𝑥 ≤ 1,
⎩

7
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Fig. 4. The mass distribution of copula 𝐹𝑏 (top left), the graphs of functions 𝛿𝐹𝑏 and 𝜔𝐹𝑏 (top right), and a 3D plot of copula 𝐹𝑏 (bottom) from Example 3.6.

and

⎧𝑥; if 0 ≤ 𝑥 ≤ 1
+ 2𝑏 ,
⎪ 1 1
4
⎪−𝑥 + 𝑏 + 2 ; if 4
+ 𝑏
2
< 𝑥 ≤ 12 ,
𝜔𝐹𝑏 (𝑥) = ⎨ 1 1 3
⎪𝑥 + 𝑏 − 2 ; if 2
<𝑥≤ 4
− 2𝑏 ,
⎪1 − 𝑥; if 3
− 𝑏
< 𝑥 ≤ 1.
⎩ 4 2

Fig. 4 shows the mass distribution of copula 𝐹𝑏 , the graphs of functions 𝛿𝐹𝑏 and 𝜔𝐹𝑏 , and 3D plot of copula 𝐹𝑏 . It follows that

1
𝛽(𝐹𝑏 ) = 4𝑏 − 1, 𝜙(𝐹𝑏 ) = −6𝑏2 + 6𝑏 − 2
and 𝛾(𝐹𝑏 ) = −6𝑏2 + 6𝑏 − 12 ,

so that

𝜙(𝐹𝑏 ) = 1 − 38 (1 − 𝛽(𝐹𝑏 ))2 and 𝛾(𝐹𝑏 ) = 23 𝜙(𝐹𝑏 ) + 14 𝛽(𝐹𝑏 ) − 18 𝛽(𝐹𝑏 )2 + 5

24
,

and both the upper bound from inequality (4) and Proposition 3.1 are attained.

Example 3.7. Let 𝑏 ∈ [0, 12 ] and let 𝐺𝑏 be a shuffle of 𝑀

𝐺𝑏 = 𝑀(6, ( 14 − 2𝑏 , 1
4
+ 2𝑏 , 12 , 3
4
− 2𝑏 , 3
4
+ 2𝑏 ), (4, 2, 6, 1, 5, 3), (1, −1, 1, 1, −1, 1)).

8
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Notice that 𝐺0 = 𝑀(2, 12 , (2, 1), (1, 1)) and 𝐺 1 = 𝑀(2, 12 , (1, 2), (−1, −1)). We have 𝐺𝑏 ( 12 , 12 ) = 𝑏,
2

⎧0; if 0 ≤ 𝑥 ≤ 1
,
⎪ 4
⎪2𝑥 − 1 ; if 1
<𝑥≤ 1
+ 2𝑏 ,
⎪ 2 4
1
4
⎪𝑏; if 4
+ 𝑏
2
< 𝑥 ≤ 12 ,
𝛿𝐺𝑏 (𝑥) = ⎨ 1 3
⎪2𝑥 + 𝑏 − 1; if 2
<𝑥≤ 4
− 2𝑏 ,
⎪1; if 3
− 𝑏
< 𝑥 ≤ 34 ,
⎪2 4 2
⎪2𝑥 − 1; if 3
< 𝑥 ≤ 1,
⎩ 4

and
⎧𝑥; if 0 ≤ 𝑥 ≤ 1
+ 2𝑏 ,
⎪ 1 1
4
⎪𝑏 − 𝑥 + 2 ; if 4
+ 𝑏
2
< 𝑥 ≤ 12 ,
𝜔𝐺𝑏 (𝑥) = ⎨ 1 1 3
⎪𝑥 + 𝑏 − 2 ; if 2
<𝑥≤ 4
− 2𝑏 ,
⎪1 − 𝑥; if 3
− 𝑏
< 𝑥 ≤ 1.
⎩ 4 2

The mass distribution of copula 𝐺𝑏 , the graphs of functions 𝛿𝐺𝑏 and 𝜔𝐺𝑏 , and 3D plot of copula 𝐺𝑏 are drawn in Fig. 5. It follows
that
1
𝛽(𝐺𝑏 ) = 4𝑏 − 1, 𝜙(𝐺𝑏 ) = −3𝑏2 + 3𝑏 − 2
and 𝛾(𝐺𝑏 ) = −4𝑏2 + 4𝑏 − 12 ,

so that

𝛾(𝐺𝑏 ) = 43 𝜙(𝐺𝑏 ) + 1
6
= 23 𝜙(𝐺𝑏 ) + 14 𝛽(𝐺𝑏 ) − 18 𝛽(𝐺𝑏 )2 + 5
24
,

and both the upper bound from inequality (6) and Proposition 3.1 are attained.

Example 3.8. Let 𝑏 ∈ [0, 12 ] and let 𝐻𝑏 be a shuffle of 𝑀

1
𝐻𝑏 = 𝑀(6, ( 2𝑏 , 2
− 2𝑏 , 12 , 1
2
+ 2𝑏 , 1 − 2𝑏 ), (3, 5, 1, 6, 2, 4), (1, 1, 1, 1, 1, 1)).

Observe that 𝐻0 = 𝑀(2, 12 , (2, 1), (1, 1)) and 𝐻 1 = 𝑀(4, ( 14 , 12 , 34 ), (2, 1, 4, 3), (1, 1, 1, 1)). We have 𝐻𝑏 ( 12 , 12 ) = 𝑏,
2

⎧0; if 0 ≤ 𝑥 ≤ 1
− 2𝑏 ,
⎪ 1
2
⎪2𝑥 + 𝑏 − 1; if 2
− 𝑏
2
< 𝑥 ≤ 12 ,
𝛿𝐻𝑏 (𝑥) = ⎨ 1 1
⎪𝑏; if 2
<𝑥≤ 2
+ 2𝑏 ,
⎪2𝑥 − 1; if 1
+ 𝑏
< 𝑥 ≤ 1,
⎩ 2 2

and
⎧𝑥; if 0 ≤ 𝑥 ≤ 14 ,
⎪
⎪−𝑥 + 1 ; if 1
<𝑥≤ 1
− 2𝑏 ,
⎪ 2
1
4
1
2
⎪𝑥 + 𝑏 − 2 ; if 2
− 𝑏
2
< 𝑥 ≤ 12 ,
𝜔𝐻𝑏 (𝑥) = ⎨ 1 1 1
⎪−𝑥 + 𝑏 + 2 ; if 2
<𝑥≤ 2
+ 2𝑏 ,
⎪𝑥 − 1 ; if 1
+ 𝑏
< 𝑥 ≤ 34 ,
⎪ 2 2 2
⎪1 − 𝑥; if 3
< 𝑥 ≤ 1.
⎩ 4

The mass distribution of copula 𝐻𝑏 , the graphs of functions 𝛿𝐻𝑏 and 𝜔𝐻𝑏 , and 3D plot of copula 𝐻𝑏 are shown in Fig. 6. It follows
that
1
𝛽(𝐻𝑏 ) = 4𝑏 − 1, 𝜙(𝐻𝑏 ) = 3𝑏2 − 2
and 𝛾(𝐻𝑏 ) = 4𝑏2 − 12 ,

so that
3 1
𝜙(𝐻𝑏 ) = 16
(1 + 𝛽(𝐻𝑏 ))2 − 2
and 𝛾(𝐻𝑏 ) = 43 𝜙(𝐻𝑏 ) + 16 ,

and the lower bound from inequality (4) and the upper bound from inequality (6) are attained.

We are now able to prove our main theorem.

Theorem 3.9. Let

𝛺𝛽,𝜙,𝛾 = {(𝛽(𝐶), 𝜙(𝐶), 𝛾(𝐶)) ∈ [−1, 1] × [− 12 , 1] × [−1, 1] ∶ 𝐶 ∈ }

9
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Fig. 5. The mass distribution of copula 𝐺𝑏 (top left), the graphs of functions 𝛿𝐺𝑏 and 𝜔𝐺𝑏 (top right), and a 3D plot of copula 𝐺𝑏 (bottom) from Example 3.7.

be the exact region determined by Blomqvist’s beta, Spearman’s footrule and Gini’s gamma. Then 𝛺𝛽,𝜙,𝛾 equals the region
{
𝛺 = (𝛽, 𝜙, 𝛾) ∈ [−1, 1] × [− 12 , 1] × [−1, 1] ∶ 16
3
(1 + 𝛽)2 − 12 ≤ 𝜙 ≤ 1 − 38 (1 − 𝛽)2 ,
5 }
max{ 3 𝜙 − 3 , 3 𝜙 + 2 𝛽 + 4 𝛽 − 12 } ≤ 𝛾 ≤ min{ 43 𝜙 + 16 , 23 𝜙 + 14 𝛽 − 18 𝛽 2 + 24
4 1 2 1 1 2 5
} .

Proof. The inclusion 𝛺𝛽,𝜙,𝛾 ⊆ 𝛺 follows from Propositions 2.1, 2.2, 3.1, and 3.2. The region 𝛺 is bounded by six surfaces
3
𝑆1 ∶ 𝜙= 16
(1 + 𝛽)2 − 12 ,
2
𝑆2 ∶ 𝛾= 3
𝜙 + 12 𝛽 + 14 𝛽 2 − 5
12
,
4
𝑆3 ∶ 𝛾= 3
𝜙 − 13 ,
𝑆4 ∶ 𝜙= 1 − 38 (1 − 𝛽)2 ,
2
𝑆5 ∶ 𝛾= 3
𝜙 + 14 𝛽 − 18 𝛽 2 + 5
24
,
4
𝑆6 ∶ 𝛾= 3
𝜙 + 16 .
𝛽0 +1
For any fixed 𝛽 the intersection of any of these surfaces with the plane 𝛽 = 𝛽0 is a line. Let 𝑏 = 4
and denote by 𝛴 the plane
𝛽 = 4𝑏 − 1. We have the following intersections:
𝑆1 ∩ 𝑆2 ∩ 𝛴 ∶ 𝐶(4𝑏 − 1, 3𝑏2 − 12 , 6𝑏2 − 1),
𝑆2 ∩ 𝑆3 ∩ 𝛴 ∶ 𝐷(4𝑏 − 1, 6𝑏2 − 12 , 8𝑏2 − 1),

10
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Fig. 6. The mass distribution of copula 𝐻𝑏 (top left), the graphs of functions 𝛿𝐻𝑏 and 𝜔𝐻𝑏 (top right), and a 3D plot of copula 𝐻𝑏 (bottom) from Example 3.8.

𝑆3 ∩ 𝑆4 ∩ 𝛴 ∶ 𝐸(4𝑏 − 1, −6𝑏2 + 6𝑏 − 12 , −8𝑏2 + 8𝑏 − 1),

𝑆4 ∩ 𝑆5 ∩ 𝛴 ∶ 𝐹 (4𝑏 − 1, −6𝑏2 + 6𝑏 − 12 , −6𝑏2 + 6𝑏 − 12 ),
𝑆5 ∩ 𝑆6 ∩ 𝛴 ∶ 𝐺(4𝑏 − 1, −3𝑏2 + 3𝑏 − 12 , −4𝑏2 + 4𝑏 − 12 ),
𝑆6 ∩ 𝑆1 ∩ 𝛴 ∶ 𝐻(4𝑏 − 1, 3𝑏2 − 12 , 4𝑏2 − 12 ).

So the intersection of the region 𝛺 with the plane 𝛴 is the hexagon 𝐶𝐷𝐸𝐹 𝐺𝐻. This hexagon has three pairs of parallel sides: the
projections of the sides 𝐸𝐹 and 𝐻𝐶 to the 𝜙-𝛾 plane are vertical, the projections of the sides 𝐶𝐷 and 𝐹 𝐺 have slope 23 , and the
projections of the sides 𝐷𝐸 and 𝐺𝐻 have slope 43 . Each of the vertices 𝐶, 𝐷, 𝐸, 𝐹 , 𝐺, and 𝐻 is attained by copula 𝐶𝑏 , 𝐷𝑏 , 𝐸𝑏 , 𝐹𝑏 ,
𝐺𝑏 , and 𝐻𝑏 from Examples 3.3–3.8, respectively. In the case 𝑏 = 0 the hexagon is reduced to the line segment from 𝐶(−1, − 12 , −1)
to 𝐹 (−1, − 12 , − 12 ). In the case 𝑏 = 0 the hexagon is reduced to the line segment from 𝐶(1, 14 , 12 ) to 𝐷(1, 1, 1). Since Blomqvist’s beta,
Spearman’s footrule and Gini’s gamma are all linear functions of copulas, a convex combination of two copulas 𝐶1 and 𝐶2 , which
is a copula, is mapped by 𝛽, 𝜙, and 𝛾 to the convex combination of their images. This means that any point (𝛽0 , 𝜙, 𝛾) in the hexagon
𝐶𝐷𝐸𝐹 𝐺𝐻 is attained by some copula with 𝛽(𝐶) = 𝛽0 , and thus any point (𝛽, 𝜙, 𝛾) ∈ 𝛺 is attained by some copula 𝐶 ∈ . □

In Fig. 7 the region 𝛺 is shown. The curves 𝑆1 ∩ 𝑆2 , 𝑆2 ∩ 𝑆3 , 𝑆3 ∩ 𝑆4 , 𝑆4 ∩ 𝑆5 , 𝑆5 ∩ 𝑆6 , and 𝑆6 ∩ 𝑆1 , are drawn blue, black,
orange, red, magenta, and green, respectively. In Fig. 8 the projection of the hexagon 𝐶𝐷𝐸𝐹 𝐺𝐻 to the 𝜙-𝛾 plane is drawn in red
for the case 𝛽0 = 13 . The whole region 𝛺𝜙,𝛾 = {(𝜙(𝐶), 𝛾(𝐶)) ∈ [− 12 , 1] × [−1, 1] ∶ 𝐶 ∈ } is drawn in black. In the case 𝛽0 = 13 we have
𝐶(− 16 , − 13 ), 𝐷( 16 , − 19 ), 𝐸( 56 , 79 ), 𝐹 ( 56 , 56 ), 𝐺( 16 , 18
7
), 𝐻(− 16 , − 18
1
).

11
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

1
Fig. 7. The region 𝛺 from Theorem 3.9. The hexagon 𝐶𝐷𝐸𝐹 𝐺𝐻 for the case 𝛽0 = 3
has edges depicted in yellow.

1
Fig. 8. The hexagon 𝐶𝐷𝐸𝐹 𝐺𝐻 for the case 𝛽0 = 3
(red) and the region 𝛺𝜙,𝛾 (black).

It is obvious that the projection of the region 𝛺 to the 𝛽-𝜙 plane is the region determined by Blomqvist’s beta and Spearman’s
footrule

𝛺𝛽,𝜙 = {(𝛽(𝐶), 𝜙(𝐶)) ∈ [−1, 1] × [− 12 , 1] ∶ 𝐶 ∈ }

= {(𝛽, 𝜙) ∈ [−1, 1] × [− 12 , 1] ∶ 3
16
(1 + 𝛽)2 − 1
2
≤ 𝜙 ≤ 1 − 38 (1 − 𝛽)2 }.

It is also clear that the projection of the region 𝛺 to the 𝜙-𝛾 plane is the region 𝛺𝜙,𝛾 , i.e. the quadrilateral 𝐴1 𝐵1 𝐶1 𝐷1 in Fig. 8. A
short calculation shows that the projection of the region 𝛺 to the 𝛽-𝛾 plane is the region determined by Blomqvist’s beta and Gini’s
gamma

𝛺𝛽,𝛾 = {(𝛽(𝐶), 𝛾(𝐶)) ∈ [−1, 1]2 ∶ 𝐶 ∈ } = {(𝛽, 𝛾) ∈ [−1, 1]2 ∶ 38 (1 + 𝛽)2 − 1 ≤ 𝛾 ≤ 1 − 38 (1 − 𝛽)2 }.

12
D. Kokol Bukovšek and B. Mojškerc Journal of Computational and Applied Mathematics 473 (2026) 116861

Fig. 8 shows that for any copula 𝐶 with given value of 𝜙(𝐶), the spread of 𝛾(𝐶), i.e., the difference between the maximal possible
value and the minimal possible value that 𝛾(𝐶) can attain, is at most 12 . If 𝜙(𝐶) is close to 1, this spread is even smaller. On average
the spread of 𝛾(𝐶) given 𝜙(𝐶) is
area(𝛺𝜙,𝛾 ) 3
= = 0.375.
length([− 12 , 1]) 8
1
If we know both, 𝛽(𝐶) and 𝜙(𝐶), the spread of 𝛾(𝐶) can still be equal to 2
, for example in the case 𝛽(𝐶) = − 12 , 𝜙(𝐶) = − 14 , when
𝛾(𝐶) ∈ [− 23 , − 16 ]. But the average spread of 𝛾(𝐶) given 𝛽(𝐶) and 𝜙(𝐶) is
volume(𝛺𝛽,𝜙,𝛾 ) 19
= ≈ 0.3167.
area(𝛺𝛽,𝜙 ) 60
So, if we know 𝛽(𝐶) and 𝜙(𝐶) on average the spread of 𝛾(𝐶) is reduced by 15.6% with respect to the average spread of 𝛾(𝐶) given
𝜙(𝐶) only.

4. Concluding remarks

In this paper we characterize the exact region representing the ternary relation between Blomqvist’s beta, Spearman’s footrule
and Gini’s gamma. Given the values of Blomqvist’s beta and Spearman’s footrule for some copula, we give the lower and the upper
bound that Gini’s gamma can take. We also provide copulas where these bounds are attained.
Future work might include studying other combinations of triplets of classical (weak) concordance measures, and relations
between other (weak) concordance measures like convex (weak) concordance measures introduced in [23,24].

Acknowledgments

The authors acknowledge financial support from the ARIS (Slovenian Research and Innovation Agency, research core funding
No. P1-0222).

Data availability

No data was used for the research described in the article.

References

[1] Juan Fernández Sánchez, Wolfgang Trutschnig, A link between Kendall’s 𝜏, the length measure and the surface of bivariate copulas, and a consequence
to copulas with self-similar support, Depend. Model. 11 (1) (2023) 14, Paper No. 20230105.
[2] Sebastian Fuchs, Klaus D. Schmidt, Bivariate copulas: transformations, asymmetry and measures of concordance, Kybern. (Prague) 50 (1) (2014) 109–125.
[3] Takaaki Koike, Marius Hofert, Comparison of correlation-based measures of concordance in terms of asymptotic variance, J. Multivariate Anal. 201 (2024)
20, Paper No. 105265.
[4] Damjana Kokol Bukovšek, Tomaž Košir, Blaž Mojškerc, Matjaž Omladič, Relation between non-exchangeability and measures of concordance of copulas,
J. Math. Anal. Appl. 487 (1) (2020) 123951, 26.
[5] Eckhard Liebscher, Copula-based dependence measures, Depend. Model. 2 (1) (2014) 49–64.
[6] Martynas Manstavičius, A few generalizations of Kendall’s tau. Part I: construction, Nonlinear Anal. Model. Control 29 (4) (2024) 783–801.
[7] Qingyang Zhang, Asymptotic expected sensitivity function and its applications to measures of monotone association, Ann. Inst. Statist. Math. 76 (5) (2024)
877–896.
[8] Javad Behboodian, Ali Dolati, Manuel Úbeda-Flores, A multivariate version of Gini’s rank association coefficient, Statist. Papers 48 (2) (2007) 295–304.
[9] Fabrizio Durante, Sebastian Fuchs, Reflection invariant copulas, Fuzzy Sets and Systems 354 (2019) 63–73.
[10] M.D. Taylor, Multivariate measures of concordance for copulas and their marginals, Depend. Model. 4 (1) (2016) 224–236.
[11] Manuel Úbeda-Flores, Multivariate versions of Blomqvist’s beta and Spearman’s footrule, Ann. Inst. Statist. Math. 57 (4) (2005) 781–788.
[12] Manuela Schreyer, Roland Paulin, Wolfgang Trutschnig, On the exact region determined by Kendall’s 𝜏 and Spearman’s 𝜌, J. R. Stat. Soc. Ser. B. Stat.
Methodol. 79 (2) (2017) 613–633.
[13] Damjana Kokol Bukovšek, Blaž Mojškerc, On the exact region determined by Spearman’s footrule and Gini’s gamma, J. Comput. Appl. Math. 410 (2022)
13, Paper No. 114212.
[14] Damjana Kokol Bukovšek, Nik Stopar, On the exact region determined by Spearman’s rho and Spearman’s footrule, J. Comput. Appl. Math. 437 (2024)
23, Paper No. 115463.
[15] Marco Tschimpke, Manuela Schreyer, Wolfgang Trutschnig, Revisiting the region determined by Spearman’s 𝜌 and Spearman’s footrule 𝜙, J. Comput. Appl.
Math. 457 (2025) 18, Paper No. 116259.
[16] M. Sklar, Fonctions de répartition à 𝑛 dimensions et leurs marges, Publ. Inst. Stat. Univ. Paris 8 (1959) 229–231.
[17] Fabrizio Durante, Carlo Sempi, Principles of Copula Theory, CRC Press, Boca Raton, FL, 2016, p. xvi+315.
[18] Roger B. Nelsen, Manuel Úbeda Flores, The lattice-theoretic structure of sets of bivariate copulas and quasi-copulas, C. R. Math. Acad. Sci. Paris 341 (9)
(2005) 583–586.
[19] Gregory A. Fredricks, Roger B. Nelsen, The Bertino family of copulas, in: Distributions with Given Marginals and Statistical Modelling, Kluwer Acad. Publ.,
Dordrecht, 2002, pp. 81–91.
[20] Roger B. Nelsen, An Introduction to Copulas, second ed., in: Springer Series in Statistics, Springer, 2006, p. xiv+269.
[21] Damjana Kokol Bukovšek, Tomaž Košir, Blaž Mojškerc, Matjaž Omladič, Spearman’s footrule and Gini’s gamma: local bounds for bivariate copulas and
the exact region with respect to Blomqvist’s beta, J. Comput. Appl. Math. 390 (2021) 23, Paper No. 113385.
[22] Damjana Kokol Bukovšek, Nik Stopar, On the exact regions determined by Kendall’s tau and other concordance measures, Mediterr. J. Math. 20 (3) (2023)
16, Paper No. 147.
[23] Radko Mesiar, Anna Kolesárová, Ayyub Sheikhi, Convex concordance measures, Fuzzy Sets and Systems 441 (2022) 366–377.
[24] Radko Mesiar, Anna Kolesárová, Ayyub Sheikhi, Svitlana Shvydka, Convex weak concordance measures and their constructions, Fuzzy Sets and Systems
478 (2024) 13, Paper No. 108841.

13