The above curve-stitching is formed by joining ( i , 0 ) to ( 0 , 1 0 − i ) for i = 1 , 2 , … , 9 .
If i is all the numbers between 1 and 10, not just integers, the outer curve will be smooth rather than kinked.
Now, after a rotation and a translation shown below, this smooth curve can be expressed as y = b a x 2 , where a and b are positive integers with a square-free.
What is the value of a b 2 ?
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.
In general, this kind of problems can be solved by envelopes . EDIT: I just realised Mark Henning has detailed this method below.
Log in to reply
Thanks for the link! I should have mentioned it. I just thought it would be nice to have a method which relies on more elementary things (properties of parabola, rotations and translations).
Log in to reply
Ohh, your solution's really well written by the way. Regarding the second part on "How do we know it's a parabola", you can complete it by deriving the definition of an envelope since you are already halfway there.
Log in to reply
@Julian Poon – Thanks for the compliment! I added the "enveloppe" part, hope some people will find it useful.
BTW, I just read in your status that you dislike Singapore's math education system. Over here in Europe people are praising Singapore's math education like crazy. I would love to hear your opinion/experiences on the topic!
Woah, I used the exact method as this but backwards. I rotated the parabola y=ax^2 45 degrees and fit it through the translated point (-2.5,7.5).
The problem statement is actually incorrect as it states: "If is all the numbers between 1 and 9" instead of "If is all the numbers between 1 and 10", which leads to the curve going through the point (9, 0) instead of (10, 0) and a solution that does not fit the requirements for "a" and "b"...
Log in to reply
Same as I thought and that's why I was getting the wrong answer
The problem is ill-posed, it should actually read "if all the numbers between 0 and 10" and restrict the solution to the first quadrant. Or look at all real numbers, as I said in my solution.
But if you go with "If is all the numbers between 1 and 9" and try to extrapolate the last line so that it goes through the point (9,0), then you should have gotten to the conclusion that you do not get a parabola (you get part of a parabola which is the extended by two lines), so it does not make sense to speak of "a" and "b"...
Anyways, it's always best to report such things in the "report section"
We want to identify the envelope of the family of straight lines ( 1 0 − u ) x + u y = u ( 1 0 − u ) 0 < u < 1 0 To do this we differentiate partially with respect to u y − x = 1 0 − 2 u and eliminate u from the two equations, obtaining 2 1 0 − x + y x + 2 1 0 + x − y y 5 ( x + y ) − 2 1 ( x − y ) 2 5 ( x + y ) = 4 ( 1 0 + x − y ) ( 1 0 − x + y ) = 2 5 − 4 1 ( x − y ) 2 = 2 5 + 4 1 ( x − y ) 2 which becomes Y = 2 5 + 2 0 2 X 2 where X = 2 x − y Y = 2 x + y In other words, we obtain the parabola y = 2 0 2 x 2 after rotation about the origin through 4 5 ∘ and a further translation. Thus a = 2 , b = 2 0 , so that a b 2 = 2 0 0 .
Buuut in the problem lines it is said that 1≤u≤9 not 0<u<10. I think that this problem should be rewritten so it is correct
Log in to reply
I agree that the question could be better phrased - the only way to end up with a unique smooth curve is to take the infinite collection of lines, not just a finite set.
Great idea! I'm not fully sure that I understand how it works though.. Differentiating the family of curves and eliminating the parameter..
Log in to reply
I wrote an introduction to enveloppes in my answer, in case you're interested. Otherwise, check the wikipedia page!
It is enough to know the slope and point of tangency to the smooth curve of one line. You can use that along with the derivation of sqrt(a)/b x^2 to get the solution. I used the line connecting 0 and 10, giving me that the derivation of sqrt(a)/b x^2 at x=10/sqrt(2) has to be 1.
Smart solution!!
(1) Take line x+y=5 and it's perpendicular distance from point (0,10)=5/√2 as "y" value in parabola y=(√a/b)x^2
(2) Find distance of point (5/2,5/2) along the line x+y=5, where, the perpendicular from point (0,10) cuts x+y=5 as equal to 5√2. Take this as the "x" value in the parabola y=(√a/b)x^2.
(3) Using (5/√2)=(√a/b){(5√2)^2}, we get the answer for b^2/a=200.
Answer=200
Problem Loading...
Note Loading...
Set Loading...
Part of the solution is given in the problem since we are told that the shape that emerges from all those tangent is a parabola. The problem is that the different lines we are given are tangents to the parabola, but we don't know where is the point these line are tangent to.
The first step is to consider the full family of lines which go through the points ( t , 0 ) and ( 0 , 1 0 − t ) (where t is any real number). These lines are in symmetry with the 45° in this system. The parabola also has one line of symmetry. Thanks to this we conclude ( 2 5 , 2 5 ) is the vertex of the parabola.
Next because it's a parabola, the tangent line only touch the graph (and do not cross it). This will give us a second point on the parabola. Indeed for t close to 0 but positive, the lines have negative slope while for t close to 0 but positive they have positive slope. Furthermore, these line all intersect very close to the point ( 0 , 1 0 ) . This means that ( 0 , 1 0 ) is also on the parabola (see below for details). By symmetry ( 1 0 , 0 ) is also on the parabola.
So we have the vertex at ( 2 5 , 2 5 ) and another point at ( 1 0 , 0 ) . We just need to rotate (so that the symmetry axis is vertical) and translate (so that the vertex is at ( 0 , 0 ) ) . ( 2 5 , 2 5 ) ( 1 0 , 0 ) → rotate by 45° ( 0 , 2 5 2 ) ( 5 2 , 5 2 ) → translate down by 2 5 2 ( 0 , 0 ) ( 5 2 , 2 5 2 ) So we have a point on the end-parabola: x = 5 2 , y = 2 5 2 . Pluging in will show that the equation for the parabola is y = 2 0 2 x 2 so that a = 2 and b = 2 0 . The answer is thus 2 0 0 .
How do we know it's a parabola?
The idea is already given above: we have a family of lines between the points ( t , 0 ) and ( 0 , 1 0 − t ) . Say L t is this line, then the equation for L t is y = − t 1 0 − t x + ( 1 0 − t ) . Note that any two such lines have exactly one intersection point. Want to know which point on L t is on the parabola, then just look at the point P t , ϵ which is at the intersection of L t and L t + ϵ . Then as ϵ → 0 , P t , ϵ will converge to a point P ˉ t on L t and this point will be on the curve. You then get a parametrisation t ↦ P ˉ t and you can check that this parametrises a parabola. Caution: the parametrisation can be singular; here for example the equation for L t is already singular when t = 0 .
As an example, for t = 1 0 , L 1 0 is the line y = 0 and L 1 0 + ϵ is y = 1 0 + ϵ ϵ x − ϵ . The intersection of L 1 0 and L 1 0 + ϵ has a y -coordinate which is obviously 0 and the x -coordinate is obtained by solving 0 = 1 0 + ϵ ϵ x − ϵ . This gives x = 1 0 + ϵ . Hence P 1 0 , ϵ = ( 1 0 + ϵ , 0 ) . Letting ϵ → 0 shows that the point ( 1 0 , 0 ) was on the curve.
From there to enveloppes... [an edit motivated by Julian Poon's comment]
In general, you have a familiy of curve C t given as the solution of an equation f ( t , x , y ) = 0 (in this problem C t are the lines L t and f ( t , x , y ) = − y − t 1 0 − t x + ( 1 0 − t ) ). The intersection point of C t and C t + ϵ will satisfy the equations f ( t , x , y ) = 0 and f ( t + ϵ , x , y ) = 0 or equivalently f ( t , x , y ) = 0 and f ( t + ϵ , x , y ) − f ( t , x , y ) = 0 Now the equation on the right will lose meaning if ϵ → 0 , which is a bugger because [since we have three variables] we need two independent equations to get a curve. So the trick is to rewrite it as f ( t , x , y ) = 0 and ϵ f ( t + ϵ , x , y ) − f ( t , x , y ) = 0 If f ( t , x , y ) is differentiable in t , then as ϵ → 0 you get two independent equations (hurray!): f ( t , x , y ) = 0 and d t d f ( t , x , y ) = 0 See Mark Henning's answer for the implementation of this solution method.