Does there exist a quadrilateral with the side lengths and diagonal lengths as indicated above?
Note: We are working in R 2 with the usual Euclidean geometry.
This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try
refreshing the page, (b) enabling javascript if it is disabled on your browser and,
finally, (c)
loading the
non-javascript version of this page
. We're sorry about the hassle.
If the quadrilateral is cyclic (that is, if it can be circumscribed by a circle), this is Ptolemy's equality (which has the same terms, but an equals sign instead of an inequality).
Shouldn't it be product of opposite sides; rather than parallel?
Log in to reply
ow, i thought it is parallel. the right word should be opposite then. thanks sir. :) ~
Log in to reply
Who sir? Where sir? You can just call me by name or bro... :) :)
Indeed, that's how I chose the sides.
We do not need to know if the quadrilateral is cyclic or not, Ptolemy's inequality applies nonetheless.
Log in to reply
Could the quadrilateral exist in a non-Euclidean geometry? Perhaps where spacetime is curved near a massive object?
Log in to reply
That's a great extension! Let me build on it.
Can this system exist as 4 points in 3-D space? Why, or why not?
Is it possible that this system could exist on the surface of a sphere? Why, or why not?
Considering the curvature of the space would help us understand what kind of geometry is desired. How much bending do we want, and in what direction?
Log in to reply
@Calvin Lin – Looks like it can't exist as a tetrahedron in 3-D Euclidean space. Each point has 3 angles between edges and in some cases one angle is greater than the sum of the other two which prevents the tetrahedron from coming together in the right way. It seems to work as geodesics on the surface of a sphere though. The radius of the sphere is approximately 3.38 units and the Gaussian curvature about 0.09.
Log in to reply
@Justin Travers – Great! Considering the vertex where 4, 5 and 9 meet, then the angle formed by 4 − 5 − 8 is larger than the sum of the angles formed by 4 − 9 − 7 and 9 − 5 − 6 , hence we cannot have a 3-D figure.
Note: The "standard" test for the 3-D case uses the Cayley-Menger determinant (after checking that the faces satisfy the triangle inequality, and there is a vertex where all the 3 angles are acute).
I've not thought about the sphere case. That's interesting that you could find such a setup. Does it seem to be the case that any 6 lengths can be represented as geodesics on a sphere?
Unless I am mistaken, Ptolemy's theorem is applicable on quadrilaterals that are inscribable in a circle. And in this case their is no info about it
Log in to reply
That is a true statement about Ptolemy's theorem.
On the other hand, Ptolemy's inequality applies to all quadrilaterals. I have updated the wiki link so that it lands on the relevant section (instead of just the main page)
If we relax the constraint on one of the lengths, that is, if we want to create a quadrilateral that satisfies any five of the six given dimensions, then would we be able to create a quadrilateral?
Suppose we replace the two diagonals 8 and 9 by two others, 7 and 8, without changing the four sides. Ptolemy's inequality now holds, but does this new quad exist? If Ptolemy's inequality is violated, the quad does not exist. If it isn't violated, the quad may or may not exist. In such a case we need to resort to another test. Baruch L.
Log in to reply
Yes. Ptolemy's inequality is a necessary but not sufficient condition.
When 5 lengths are given, the 6th will be determined uniquely. We can do so using sine/cosine rules to calculate what it should be.
Thanks. I don't remember ever encountering Ptolemy's inequality.
This is incorrect. The figure is a tetrahedron NOT a quadrilateral. Therefore, the quadrilateral does not exist. Am I the only one who finds this kind of sloppy wording maddening. I was pretty sure what you meant, but as framed, it is wrong.
Log in to reply
"We are working in R 2 with the usual Euclidean geometry", meaning that all elements lie on the same plane.
Law of Cosines , we have
By9 2 = 4 2 + 7 2 − 2 ( 4 ) ( 7 ) cos A ⟹ A = 1 0 6 . 6 0 1 5 4 9 6 ∘
9 2 = 5 2 + 6 2 − 2 ( 5 ) ( 6 ) cos C ⟹ C = 1 0 9 . 4 7 1 2 2 0 6 ∘
8 2 = 4 2 + 5 2 − 2 ( 4 ) ( 5 ) cos B ⟹ B = 1 2 5 . 0 9 9 6 3 2 2 ∘
8 2 = 6 2 + 7 2 − 2 ( 6 ) ( 7 ) cos D ⟹ D = 7 5 . 5 2 2 4 8 7 8 1 ∘
Adding the angles, we have
A + B + C + D = 1 0 6 . 6 0 1 5 4 9 6 ∘ + 1 0 9 . 4 7 1 2 2 0 6 ∘ + 1 2 5 . 0 9 9 6 3 2 2 ∘ + 7 5 . 5 2 2 4 8 7 8 1 ∘ = 4 1 6 . 6 9 4 8 9 0 2 ∘ ⟹ It is not equal to 360.
∴ It is not a quadrilateral.
Careful. Since all your calculations are estimates (and not exact), you should prove that the exact value of the sum of angles is not 3 6 0 ∘ .
To tighten up your proof, consider rewriting the cosines of A,B,C,D as the (exact) simplified fractions. Then show that cos − 1 A + cos − 1 B + cos − 1 C + cos − 1 D = 3 6 0 ∘ . Do you know how to proceed?
Log in to reply
His proof is perfectly fine. He estimated each angle to nearest 0.0000001, which means his margin of error on the sum is +/- 0.0000002, and the difference between the sum and 360 was far out of that range.
Log in to reply
Ahh fair enough. For some reason, I thought there's a huge margin of error. Thanks for clearing things up
If this quadrilateral exists; then the sum of the areas of the 8-6-7 triangle and the 8-5-4 triangle must be equal to the sum of the areas of the 9-5-6 triangle and 9-7-4 triangle.
By heron's formula; the above situation means;
0
.
5
×
8
.
5
×
3
.
5
×
4
.
5
+
1
0
.
5
×
2
.
5
×
3
.
5
×
4
.
5
=
1
0
×
1
×
5
×
4
+
1
0
×
1
×
6
×
3
⟹
6
6
.
9
3
7
5
+
4
1
3
.
4
3
7
5
=
2
0
0
+
1
8
0
A quick look at the calculator or a quick approximation in mind and we can conclude that the LHS is roughly 1 more than the RHS.
Hence; this quadrilateral doesnt exist.
Nice alternative way of disproving the existance :)
As it turns out, giving 5 side lengths of the quadrilateral uniquely determines it (and whether or not it could exist). Using Heron's formula, we can obtain an equation for the 6th side (and if there are no positive solutions, then it does not exist). So, this is a neceesary condition. Is it a sufficient condition? IE If the 2 "vertical" triangles and the 2 "horizontal" triangles have the same summed area, does that imply that the quadrilaterial exists?
Log in to reply
I would guess, YES; till all four are possible triangles. Will try to prove or disprove this...
'.' quadrilateral can't have 3 obtuse angles
.'. the shown sidescan't make a quadrilateral
Notice that the left (B) and right (D) points lie on the ellipse with the other two points (A and C) as the two foci, c=4.5 and a=5.5. Midpoi t between AC=O. Then 2(DO^2+4.5^2)=5^2+6^2. Easily DO<4. Similarly BO<4. So it is impossible for BD=8.
Let ∠ 4 5 be the angle between sides of lengths 4 and 5 (for arbitrary values of 4 and 5).
Naturally we need that ∠ 4 9 + ∠ 5 9 = ∠ 4 5 .
By the cosine rule: 7 2 = 4 2 + 9 2 − 2 ⋅ 4 ⋅ 9 cos ∠ 4 9 ⇒ cos ∠ 4 9 = 2 / 3 ⇒ sin ∠ 4 9 = 3 5 6 2 = 5 2 + 9 2 − 2 ⋅ 5 ⋅ 9 cos ∠ 5 9 ⇒ cos ∠ 5 9 = 7 / 9 ⇒ sin ∠ 5 9 = 9 4 2 8 2 = 4 2 + 5 2 − 2 ⋅ 4 ⋅ 5 cos ∠ 4 5 ⇒ cos ∠ 4 5 = 2 3 / 4 0 sin ∠ 4 9 sin ∠ 5 9 = 2 7 4 1 0
But cos ( A + B ) = cos A cos B − sin A sin B . We already have a problem: Since cos ∠ 4 9 cos ∠ 5 9 and cos ∠ 4 5 are both rational, sin ∠ 4 9 sin ∠ 5 9 would also need to be rational. But sin ∠ 4 9 sin ∠ 5 9 is irrational, so we have a problem.
I realise we have here a corollary of the cosine rule: If all side lengths of a triangle are rational, then the angles have rational cosines. Therefore it wasn't necessary to calculate cos ∠ 4 5 , since we already know that it's rational. Essentially, I was showing that by reaching a contradiction (rational - irrational = rational), we can save ourselves a bit of arithmetic.
If OTOH we had found sin ∠ 4 9 sin ∠ 5 9 to be rational, we /would/ need to do the arithmetic to find out whether they're equal.
Log in to reply
Nice! So many ways to get at a contradiction. :)
Verifying that all the angles add up is a good way to ensure we have a geometrically correct figure :)
[This is not a complete solution.It is just for discussion doubts and anomalies.]
Assume that the angle formed by length 4 and length side is A.The diagonal of length 8 divides this angle in two parts namely P and Q.
First find c o s ∠ A by triangle cosine rule whose sides are 4 , 7 , 9 .
Now calculate c o s ∠ P whose sides are 4 , 8 , 5 and c o s ∠ Q whose corresponding sides are 8 , 7 , 6 .
Apply the formula c o s ∠ A = c o s ∠ ( P + Q ) .
While calculating these two values will not be equal
It makes sure the construction of this quadrilateral is not possible.
Yup, that's another way of approaching it, by showing that the angles do not add up. For completeness, can you add in the calculations for the angles?
This is why diagrams can be so misleading.
Log in to reply
The angle between side size 9 and 7 is 2 5 . 2 1 ∘ , between 9 and 6 it is 3 1 . 5 9 ∘ . From that I got what should have been an 8 to be actually 6 . 2 4 . (I didn't do the third angle but rather a side, same idea.)
Log in to reply
Yup, lots of contradictions abound, once you start looking closely :)
The side whose length is 4 is longer than the one with 6 this isn't possible and hence such a figure can't exist.
Using the 9 diagonal as common base, drop a perpendicular (h1) from the apex of the 9/6/5 triangle. Since the base on 6 side must be <6, the base on the 5 side must be >3, and h is therefore <4. Using the same method on the other triangle (9/7/4) it is clear that the h2 <3. The shortest distance between the two apexes will be greater than the sum of the two heights (which will be less than 8, as demonstrated); but the hypotenuses of the small right triangles between the heights and the "8" diagonal are nevertheless <5 and <3 respectively, so that total diagonal must be <8. (If, like me, you didn't know Ptolemy's inequality).
I'm trying to follow what you're saying, but I got lost.
I agree that the heights of the 9-6-5 is less than 4, and the 9-7-4 triangles is less than 3, and so the "total height" would be less than 7.
However, how does that contradict that the diagonal has a length of 8? To me, line connecting these two points would have a greater length than the sum of the heights, which seems to agree with your sentence of "The shortest distance between the two apexes will be greater than the sum of the two heights". Can you elaborate on the contradiction
The almost right triangle with sides 5 and 6 would never Have a 9 as the hypotenuse
It is not a right triangle, so 5^2+6^2=9^2 is not satisfied. So, what's your point?
I think you're assuming that the image is accurate. If so, the images on our site are not accurate unless otherwise indicated. They are meant to give you a starting point to think about how the lines / angles relate to each other.
9 2 = 4 2 + 7 2 − 5 6 cos A → cos A = 5 6 − 3 7 9 2 = 5 2 + 6 2 − 6 0 cos C → cos C = 3 − 1 8 2 = 4 2 + 5 2 − 4 0 cos B → cos B = 4 0 − 2 3 8 2 = 7 2 + 6 2 − 8 4 cos D → cos D = 4 1
There would be three obtuse angles. Impossible!
arccos ( − 6 0 3 7 ) + ( arccos ( − 3 1 ) + ( arccos ( − 4 0 2 3 ) + ( arccos ( 4 1 ) = 7 . 6 . . . r a d
100, 100 , 100, 60 are 3 obtuse angles that sum to 360, what's wrong with that?
Check using pythagorean theroem if the square of the diagonal is equal to the square of the sum of the other sides.
Pythagorean theorem only works for right angled triangles. None of the triangles inside this quadrilateral are right angled.
You don't have to calculate anything. 6-7-8 is nearly equilateral, so the 9 will extend way beyond the 8 line. Which it can't do because the 8-4-5 triangle is nearly a straight line. QED.
What do you mean by "nearly equilateral"? What is the threshold of calling a triangle to be equilateral?
You seem to be providing an intuitive reason for why the diagonal length cannot be 9. In such cases, we have to be careful about the extent of approximation. As it turns out, that length would be 7-ish, which is pretty close to 9.
One comment says 7 is close to 9; the other implies I'm too loose saying 6-7-8 is "nearly equilateral." I think I picked a nice mean (trusting my intuition!)
By inspection, at least one inscribed triangle violates the Triangle Inequality, hence this quadrilateral is impossible.
Which triangle violates the triangle inequality?
If you're looking at the "small" triangles (using part of the diagonal, instead of the full diagonal), then it's still possible to assign side lengths so that the triangle inequality isn't violated. E.g. splitting them evenly (though of course, this would cause the 4 sides to have the same length).
Simple: because the triangle formed by the sides of 4, 5, and 9 would not exist, since 4+5=9 and for a triangle to exist, the sum of two sides must be greater than the third.
the full Pythagorean theorum:
if (C x C) = (B x B) + (A x A), the triangle is right. if (C x C) is greater than (B + B) + (A x A), the triangle is obtuse. if (C x C) is less than (B x B) + (A x A), the triangle is acute.
Also, the biggest angle in a triangle is opposite the biggest side.
Plugging in the top left triangle, 4 x 4 is less than 8x8 + 9x9, so the top left angle in the center would be acute (if it were real).
In order for the top right side to be obtuse to match, the 5 from the perimeter would need to be greater than the 9 from the center to the top.
Because 2 acute angles can not add up to 180 degrees, the figure has artificial measurements.
Can this be invalidated by proving the triangles represented can't exist as shown?
Each of the 4 individual triangles could exist as triangles, since they satisfy the triangle inequality. For this problem, we need to show that they cannot be put together to form an equilateral triangle.
How about this: the cyan diagonal is split by the blue diagonal into two segments having length (8-x) and x. We then have two triangles, (4; 9; 8-x) and (9; 5; x).
For the first triangle, 8 - x > (9-4) gives x < 3; for the second triangle, x > (9-5), so x > 4. Since x cannot be greater than 4 and lower than 3 at the same time, the figure cannot exist. Would this be enough to prove it?
Log in to reply
Unfortunately no. Notice that the 9 diagonal is also split up into 2 parts. We do not have a 4:9:8-x triangle. Instead, we have a 4:9:7 triangle.
Problem Loading...
Note Loading...
Set Loading...
Use Ptolemy's inequality . That is, for all quadrilaterials,
p r o d u c t o f 2 o p p o s i t e s i d e s ∑ ≥ p r o d u c t o f 2 d i a g o n a l s ∑
But...
( 4 ⋅ 6 ) + ( 7 ⋅ 5 ) ≥ ( 8 ⋅ 9 )
Hence, the quadrilateral doesn't exist.