Parallel covering a rhombus with equilateral triangles
Jingjing Wang
Yanxun Chang
Note: Corresponding Author.
Affiliation: 21118010@bjtu.edu.cn; yxchang@bjtu.edu.cnAffiliation: School of Mathematics and Statistics,
Affiliation: Beijing Jiaotong University, Beijing 100044, P.R. China.
Abstract
Suppose that is a rhombus with side length and with an interior angle , where
.
Let be an equilateral triangle with a side parallel to a side of and let be a collection of homothetic copies of .
In this paper, we show the following two results:
if and the sum of the areas of equilateral triangles from is at least , then these equilateral triangles can parallel cover the rhombus ;
if and the sum of the areas of equilateral triangles from is at least , then they can parallel cover the rhombus .
Furthermore, these bounds are optimal on their respective intervals.
Let , , and be convex bodies in the plane.
Let denote the area of .
We recall the following definitions from [3].
We say that the collection permits a covering of if there exist rigid motions such that .
If each is a translation, then permits a translative covering of .
When and all are polygons, a covering is called parallel if there exists a side of such that each has a side parallel to it.
Such a side of is called a base of .
Let be the smallest positive number such that any collection of positive homothetic copies of whose total area is not less than permits a translative covering of .
It is known that for any square [5].
It is shown in [3] that for an equilateral triangle and a square .
For parallel coverings by squares, results for isosceles triangles are given in [4], while those for a rhombus, an obtuse triangle and a parallelogram are given in [7, 9, 8], respectively.
Results on covering a triangle with triangles can be found in [1, 2].
A result on covering a square with isosceles right triangles can be found in [6].
In this paper, we consider the problem of parallel covering a rhombus with equilateral triangles.
Throughout the paper, the admissible rigid motions are translations and rotations (denoted by ).
We still refer to such coverings as parallel coverings.
Let be a rhombus with side length and with an interior angle , where .
Let be an equilateral triangle with a side parallel to a side of and let be a collection of homothetic copies of .
Let be the smallest positive number such that any collection of homothetic copies of with total area not less than permits a parallel covering of .
The present problem differs from the related square-covering problems in that the interaction between the geometry of equilateral triangles and the angle of the rhombus leads to different
residual configurations.
The aim of this paper is to determine the value of .
More precisely, we prove that if , then
(see Theorem 3.1);
if , then
(see Theorems 3.2–3.3).
2 Preliminaries
The line segment connecting points and is denoted by , and its length is denoted by .
The line through points and is denoted by .
To prove our main result, we need the following result from [1].
Lemma 2.1.
Let be a triangle and let be any collection of positive homothetic copies of .
If , then permits a parallel covering of .
Lemma 2.2.
Let be a trapezoid with , , height , base angles and (see Fig. 1).
Then .
Let be an equilateral triangle with a side parallel to and let be a collection of homothetic copies of .
Let denote the side length of for .
Without loss of generality we may assume that .
If
then permits a parallel covering of .
Fig. 1: A covering method of
Proof.
We use the following procedure to construct a parallel covering of with equilateral triangles, following the method of [3, 5] (see Fig. 1).
We place the equilateral triangles layer by layer from left to right, with one side of each triangle parallel to .
In particular, the base of the first layer contains the segment .
In each layer, the first triangle is positioned so that one of its sides lies along and another side lies along the base of the layer.
Each subsequent triangle is placed to the right of the preceding one so
that one of its sides is contained in a side of the preceding triangle,
with the two sides having a common endpoint on the base of the layer,
and the interiors of the triangles are pairwise disjoint.
A layer is called full if there is at least one unused triangle
in the collection and the first unused triangle in the given order
cannot be placed, according to the above placement rule, to the right
of the last triangle already placed, in such a way that its interior
has a nonempty intersection with .
If the collection is finite and all its triangles have been used before
is completely covered, the construction stops. We then classify
the last layer by considering a hypothetical equilateral triangle whose
side length does not exceed that of the last triangle placed.
If no such triangle can be placed in the same way, the last layer is regarded as
full; otherwise, it is the final non-full layer.
For each , let be a rigid motion used to place , and define .
For simplicity, each placed triangle is labeled in all figures.
For the -th full layer, let be the unique intersection point of the base of this layer with the side of , and let denote the first triangle in the left-to-right placement order whose boundary contains .
To determine the last triangle of the -th full layer and the corresponding layer height , we first check whether is the first triangle in that layer.
If is not the first triangle in that layer, we consider the following three cases according to the position of the top vertex of relative to the line .
1.
The top vertex of lies strictly to the right of the line .
In this case, is the last triangle of the layer.
We set and (the height of ).
For example, in Fig. 1, has its top vertex strictly to the right of the line .
Hence, and .
2.
The top vertex of lies strictly to the left of the line .
If , then is the last triangle of the layer.
We set and .
Otherwise, the layer ends with . We set and
.
For example, in the second layer of Fig. 1, the top vertex of lies strictly to the left of the line and .
Therefore, and .
In the third layer, the opposite inequality holds for , so and .
3.
The top vertex of lies on the line .
As in item 1, is the last triangle of the layer. We set and .
For example, this occurs for in the fifth layer of Fig. 1, and hence and .
It remains to consider the case in which is the first triangle in that layer.
If the top vertex of lies on or strictly to the right of the line , then the covering is complete, since already covers the remaining uncovered part of .
If the top vertex of lies strictly to the left of the line , then the covering is complete if covers the remaining uncovered part of ;
otherwise, and are determined by the rule in item 2.
Once the last triangle of the -th full layer
and the corresponding height have been determined, if the
construction proceeds to a new layer, its base is placed at a vertical
distance above the base of the -th full layer. The first
triangle of the new layer is , with one side lying
along and another along the base of the new layer.
This process is repeated until is completely covered or all triangles in the collection have been used.
Contrary to the statement, we suppose that it is impossible to cover with the equilateral triangles by this method.
We first consider the case where finitely many full layers are created.
Let denote the number of full layers.
If , then all triangles are placed in the first layer.
Hence,
.
This leads to a contradiction.
Therefore, we may assume that .
By construction, the first triangle in the -th layer is .
By the definition of , we have whenever the -th layer exists.
Clearly, .
The total area of the equilateral triangles lying in the first layer is less than , where is the area of a parallelogram with base length and height , and is the area of an equilateral triangle with height .
Similarly, using , we know that the total area of the equilateral triangles lying in the -th layer, for , is less than
Moreover, the total area of the equilateral triangles lying in the -th layer, if it has been created, does not exceed
Observe that the sum
does not exceed the area of a trapezoid with base lengths and and with height , that is, does not exceed .
Hence,
From and , we obtain
.
Moreover, since , we have
Thus,
which is a contradiction.
Now consider the case where infinitely many full layers are created.
Using the strict estimates obtained above together with
we obtain
Since and , using again the sign of , we obtain the same contradiction.
∎
Corollaries 2.3 and 2.5 below are direct applications
of Lemma 2.2, whereas Corollaries 2.4
and 2.6 follow from the same layer-by-layer construction in
the degenerate case .
In Corollaries 2.3–2.6, let be an equilateral triangle with one side parallel to , and let be a collection of homothetic copies of with side lengths .
All coverings considered below are parallel.
Without loss of generality we may assume that .
Corollary 2.3.
Let be the trapezoid described in Lemma 2.2(see Fig. 1).
If
then permits a parallel covering of .
Moreover, if is the smallest integer such that , , , permit a covering of (see Fig. 1), then
Corollary 2.4.
Let be a triangle with base , height (from to ), and base angles and (see Fig. 2).
If
then permits a parallel covering of .
Moreover, if is the smallest integer such that , , , permit a covering of , then
Fig. 2:
Corollary 2.5.
Let be a trapezoid with , , height , base angles and (see Fig. 3).
Then .
If , then permits a parallel covering of .
Moreover, if is the smallest integer such that permit a covering of , then
.
Fig. 3:
Corollary 2.6.
Let be a triangle with base , height (from to ), and base angles and (see Fig. 4).
If , then permits a parallel covering of .
Fig. 4:
3 Main Result
Suppose that is a rhombus with side length and with an interior angle , where
.
Let be an equilateral triangle with a side parallel to the base of .
Let be a collection of homothetic copies of .
Let denote the side length of for .
Without loss of generality we may assume that .
Theorem 3.1.
If and
, then permits a parallel covering of and thus
.
Proof.
We claim that permits a parallel covering of .
We may assume that , otherwise can be parallel covered by .
Fig. 5:
We place as in Fig. 5.
The remaining equilateral triangles are used for the covering of the triangle with base length and height
.
By Corollary 2.4, we deduce that if cannot be covered, then
This inequality leads to a contradiction.
By the discussions above we know that .
However, any equilateral triangle with side length less than cannot parallel cover , hence
.
As a consequence, .
∎
Theorem 3.2.
If and
, then permits a parallel covering of and thus .
Proof.
We claim that permits a parallel covering of .
We may assume that , otherwise can be parallel covered by .
Throughout the upper-bound estimates below, we fix an arbitrary in the stated range.
For this fixed , each right-hand side is a quadratic polynomial in the relevant variables , considered on the corresponding admissible domain.
Its maximum over the closure of that domain is determined by checking the stationary points in the interior and considering all boundary cases.
In each case, we give a point at which the maximum is attained.
The same procedure is used in the proof of Theorem 3.3.
Case 1: .
Fig. 6:
We assume that , otherwise can be parallel covered by and as shown in Fig. 6.
Subcase 1.1: .
Fig. 7:
We place and as in Fig. 7(a).
The remaining equilateral triangles are used for the covering of the equilateral triangle
with side length
and height .
By Lemma 2.1, we deduce that if cannot be covered, then
The maximum of the right-hand side over the closure of the feasible region is (which is attained at and ) provided that the feasible region is given by ,
and .
This leads to a contradiction.
Subcase 1.2: .
We place as in Fig. 7(b).
The remaining equilateral triangles are used for the covering of the triangle with base length and height
.
By Corollary 2.4, we deduce that if cannot be covered, then
The maximum of the right-hand side over the closure of the feasible region is
(which is attained at )
provided that the feasible region is given by .
This leads to a contradiction.
Case 2: .
Fig. 8:
We assume that , otherwise can be parallel covered by and , as shown in Fig. 8(a).
Subcase 2.1: .
We assume that
, otherwise can be parallel covered by , and , as shown in Fig. 8(b).
(i) .
Fig. 9:
We place , and as in Fig. 9(a).
The remaining equilateral triangles are used for the covering of the right triangle with leg lengths
and .
By Corollary 2.6 and , we deduce that if cannot be covered, then
The maximum of the right-hand side is
provided that the feasible region is given by , , and .
This maximum, which is attained at and
, is less than .
This leads to a contradiction.
(ii) .
We place , and as in Fig. 9(b).
The remaining equilateral triangles are used for the covering of the trapezoid
with base lengths
and
and height .
By Corollary 2.5 and , we deduce that if cannot be covered, then
The maximum of the right-hand side over the closure of the feasible region is
provided that the feasible region is given by
, ,
and .
This maximum, which is attained at
and ,
is less than .
This leads to a contradiction.
(iii) .
We place and as in Fig. 9(c).
The remaining equilateral triangles are used for the covering of the trapezoid
with base lengths
and and height .
By Corollary 2.5, we deduce that if cannot be covered, then
The feasible region is defined by , ,
and .
The maximum of the right-hand side over the closure of the feasible region is attained at
and .
Moreover, this maximum is less than .
This leads to a contradiction.
Subcase 2.2: .
Fig. 10:
We place as in Fig. 10.
The remaining equilateral triangles are used for the covering of the trapezoid with base lengths and and height .
By Lemma 2.2, we deduce that if cannot be covered, then
The maximum is
provided that the feasible region is given by
,
and .
This maximum, which is attained at and , is less than
.
This leads to a contradiction.
Case 3: .
Subcase 3.1: .
Subcase 3.1.1: .
Fig. 11:
We assume that , otherwise can be parallel covered by , , and , as shown in Fig. 11.
(i) .
Fig. 12:
We place , , and as in Fig. 12(a).
The remaining equilateral triangles are used for the covering of the trapezoid
with base lengths
and
and height
.
By Lemma 2.2 and , we deduce that if cannot be covered, then
The maximum of the right-hand side over the closure of the feasible region is
provided that the feasible region is given by
,
,
, and .
This maximum, which is attained at , , is less than
.
This leads to a contradiction.
(ii) .
We place , and as in Fig. 12(b).
The remaining equilateral triangles are used for the covering of the equilateral triangle with side length
and height .
By Corollary 2.4, we deduce that if cannot be covered, then
The feasible region is defined by ,
,
,
and .
The
maximum of the right-hand side is attained at
and
.
Moreover, this maximum is less than .
This leads to a contradiction.
Subcase 3.1.2: .
Subcase 3.1.2.1: .
Fig. 13:
Throughout this subcase, let denote the
trapezoid with base lengths
and
and height , as shown in Fig. 13.
In this subcase, we distinguish two possibilities according to .
Subcase 3.1.2.1.1: .
In this subcase, we distinguish three possibilities according to .
(i) .
We place , , , and as in Fig. 13(a).
The remaining equilateral triangles are used for the covering of the trapezoid .
By Lemma 2.2 and , we deduce that if cannot be covered, then
The maximum over the closure of the feasible region is
provided that the feasible region is given by
,
,
,
,
and .
This maximum, which is attained at , is less than
.
This leads to a contradiction.
(ii) .
We place , , , and as in Fig. 13(b).
The remaining equilateral triangles are used for the covering of and , where
has base length and height .
By Corollaries 2.3 and 2.4, together with for , we deduce that if cannot be covered, then
The feasible region is defined by
,
,
, ,
,
and .
The
maximum of the right-hand side over the closure of the feasible region is attained at
and
.
Moreover, this maximum is less than .
This leads to a contradiction.
(iii) .
We place , , and as in Fig. 13(c).
The remaining equilateral triangles are used for the covering of and , where
has base lengths
and
and height .
By Corollaries 2.3 and 2.5, together with for , we deduce that if cannot be covered, then
The feasible region is defined by
,
,
, ,
,
and .
The
maximum of the right-hand side over the closure of the feasible region is attained at
,
and
.
Moreover, this maximum is less than .
This leads to a contradiction.
Subcase 3.1.2.1.2: .
We place , and as in Fig. 13(d).
The remaining equilateral triangles are used for the covering of and , where
has base lengths and and height .
By Corollary 2.3 and for , we deduce that if cannot be covered, then
The feasible region is defined by
,
,
, , and .
The
maximum of the right-hand side over the closure of the feasible region is attained at
, and .
Moreover, this maximum is less than
.
This leads to a contradiction.
Subcase 3.1.2.2: .
Fig. 14:
We place and as in Fig. 14(a).
The remaining equilateral triangles are used for the covering of the trapezoid
with base lengths and and height .
By Corollary 2.5, we deduce that if cannot be covered, then
The feasible region is defined by
,
,
, and .
The
maximum of the right-hand side over the closure of the feasible region is attained at
and .
Moreover, this maximum is less than .
This leads to a contradiction.
Subcase 3.2: .
We place as in Fig. 14(b).
The remaining equilateral triangles are used for the covering of the trapezoid with base lengths and and height .
By Lemma 2.2, we deduce that if cannot be covered, then
The feasible region is defined by
,
and
.
The
maximum of the right-hand side is attained at
.
Moreover, this maximum is less than .
This leads to a contradiction.
By the discussions above we know that
.
However, any equilateral triangle with side length less than
cannot parallel cover , hence
.
As a consequence, .
∎
The angle comes from Subcase 2.1(ii) of Theorem 3.2.
For the corresponding upper-bound expression and feasible region,
the maximum is attained at
, ,
when ,
whereas it is attained at
,
,
when
.
At , we have
.
Hence this is the transition value at which
the relevant active boundary constraint changes.
Theorem 3.3.
If and
,
then permits a parallel covering of and thus
.
Proof.
We claim that permits a parallel covering of .
We may assume that , otherwise can be parallel covered by .
Case 1: .
The argument in this case is the same as that in Case 1 of Theorem 3.2.
Case 2: .
Fig. 15:
We assume that , otherwise can be parallel covered by and , as shown in Fig. 15(a).
Subcase 2.1: .
We assume that , otherwise can be parallel covered by , and , as shown in Fig. 15(b).
Subcase 2.1.1: .
Fig. 16:
We place , and as in Fig. 16(a).
The remaining equilateral triangles are used for the covering of the right triangle with leg lengths
and .
By Corollary 2.6 and , we deduce that if cannot be covered, then
The maximum of the right-hand side over the closure of the feasible region is
provided that the feasible region is given by , , and .
This maximum, which is attained at
and , is less than .
This leads to a contradiction.
Subcase 2.1.2: .
We place and as in Fig. 16(b).
The remaining equilateral triangles are used for the covering of the right triangle with leg lengths
and
.
By Corollary 2.6, we deduce that if cannot be covered, then
The maximum of the right-hand side over the closure of the feasible region is provided that the feasible region is given by , , and
.
This maximum, which is attained at
and , is less than .
This leads to a contradiction.
Subcase 2.2: .
Subcase 2.2.1: .
Fig. 17:
We place , and as in Fig. 17(a).
The remaining equilateral triangles are used for the covering of the trapezoid with base lengths and
and height .
By Corollary 2.5 and , we deduce that if cannot be covered, then
The maximum of the right-hand side over the closure of the feasible region is
provided that the feasible region is given by , , and .
This maximum, which is attained at
and , is less than .
This leads to a contradiction.
Subcase 2.2.2: .
We place and as in Fig. 17(b).
The remaining equilateral triangles are used for the covering of the trapezoid with base lengths and and height .
By Corollary 2.5, we deduce that if cannot be covered, then
The maximum of the right-hand side over the closure of the feasible region is
provided that the feasible region is given by , , and .
This maximum, which is attained at
, and , is less than
.
This leads to a contradiction.
Subcase 2.3: .
We place as in Fig. 17(c).
The remaining equilateral triangles are used for the covering of the trapezoid
with base lengths
and and height .
By Lemma 2.2, we deduce that if cannot be covered, then
The maximum of the right-hand side is
provided that the feasible region is given by , and .
This maximum, which is attained at and , is less than .
This leads to a contradiction.
Case 3: .
The covering construction is the same as that in Case 3 of Theorem 3.2, and the same right-hand side expressions are obtained in the corresponding subcases.
For each subcase, we recheck the maximum of the corresponding expression over the closure of its admissible domain for the present range of and verify that it is still bounded above by .
Thus the same contradictions follow in Case 3.
By the discussions above we know that .
However, any equilateral triangle with side length less than cannot parallel cover , hence .
As a consequence, .
∎
Funding. This research was supported by the National Natural Science
Foundation of China (Grant Number 12371326).
References
[1] Z. Füredi, Covering a triangle with homothetic copies, in: Discrete Geometry, Monogr. Textbooks Pure Appl. Math., 253, Dekker (New York, 2003), pp. 435–445.
[2] J. Januszewski, Covering a triangle with sequences of its homothetic copies, Period. Math. Hungar., 36 (1998), 183–189.
[3] J. Januszewski, Parallel packing and covering of an equilateral triangle with sequences of
squares, Acta Math. Hungar., 125 (2009), 249–260.
[4] M. Lu and Z. Su, Parallel covering of isosceles triangles with squares, Acta Math. Hungar., 155 (2018), 266–297.
[5] J. W. Moon and L. Moser, Some packing and covering theorems, Colloq. Math., 17
(1967), 103–110.
[6] F. Song and Z. Su, Parallel covering a square with isosceles right triangles,
Results Math., 80 (2025), Paper No. 124, 10 pp.
[7] C. Su and X. Li, Parallel covering a rhombus with squares, Period. Math. Hungar., 88 (2024), 190–203.
[8] Z. Su and Y. Wu, Parallel covering an obtuse triangle with squares, Studia Sci. Math. Hungar., 61 (2024), 161–184.
[9] Z. Su and J. Zhang, Parallel covering a parallelogram with squares, Acta Math. Hungar., 172 (2024), 264–286.