Avoiding traffic lights

An infinite road has traffic lights at intervals of 1500 m. The lights are all synchronised and are alternately green for \dfrac{3}{2} minutes and red for 1 minute. For which v can a car travel at a constant speed of v m/s without ever going through a red light?

Answer.

Let x(t) be the position of the car at time t s and let x(0)=x_0 for some 0 \le x_0 <1500. With speed v, the car will be at position x(t) = x_0 + vt at time t s. Without loss of generality, we assume that 0 \le x_0 < 1500 and assume that the light is green at time t s where \displaystyle t \in A= \bigcup_{k=0}^\infty [150k, 150k+90)

We want such that whenever x(t)=1500n if and only if t \in A.

The equation is equivalent to 1500n = x_0 + vt meaning that \dfrac{1500n-x_0}{v}=t. Hence, we want for each n, there exists an integer k such that

150k \le \dfrac{1500n-x_0}{v} < 150k+90

for some integer k. This means that for every integer n,

k \le \dfrac{10n - y_0}{v} < k + \dfrac{3}{5} for some integer k

i.e. 0 \le \left \{ \dfrac{10n-y_0}{v} \right \} <\dfrac{3}{5}

where y_0 = \dfrac{x_0}{150}. Here, we have that 0 \le y_0 < 10.

Let us discuss the case for y_0=0. If we let \alpha = \dfrac{10}{v}, then the inequality becomes 0 \le \left \{n \alpha \right \} < \dfrac{3}{5} for every positive integer n.

Lemma. If \alpha is a positive real number such that \{\alpha\},\{2\alpha\},\{3\alpha\},\{4\alpha\},\dots are all in the interval \left[0,\dfrac{3}{5}\right), then \left \{ \alpha \right \} = \dfrac{1}{2}.

Sketch of proof of lemma. If \{\alpha\} = \dfrac{1}{2}, then we have that \{n \alpha\} = 0 when n is even and \dfrac{1}{2} when n is odd. This clearly satisfies the condition.

If \alpha is an irrational number, then the sequence \{\alpha\},\{2\alpha\},\{3\alpha\},\dots is equidistributed on the interval [0,1). This is Weyl’s Equidistribution Theorem, but we can easily visualize what happens and prove a weaker statement regarding it being bouded by \dfrac{3}{5}.

Suppose now that \{\alpha\}=\dfrac{a}{b} is a rational number for some relatively prime positive integers 1 \le a < b. Suppose that n is a positive integer such that \dfrac{1}{n+1} \le \dfrac{a}{b} < \dfrac{1}{n}. Then, note that n\dfrac{a}{b}<1 and hence \{n\alpha\} = n \dfrac{a}{b} \ge \dfrac{n}{n+1} > \dfrac{3}{5} if n \ge 2.

Hence, it must be the case that \{ \alpha\} > \dfrac{1}{2}. But then, since \{\alpha\} < \dfrac{3}{5}, we must have \{2\alpha\} < \dfrac{1}{5}. We can apply the same argument to the sequence \{2\alpha\},\{4\alpha\},\{6 \alpha\},\{8\alpha\},\dots and deduce that one of the elements of the sequence is > \dfrac{3}{5}. \blacksquare

According to the lemma, we must have that \{\alpha\} = \left\{ \dfrac{10}{v} \right \} = \dfrac{1}{2}. This gives us solution v=\dfrac{20}{2n+1} meter/second for some non-negative integer n. For example, v=20, v=\dfrac{20}{3}, v=4, etc. will work.

This question appears as Brazilian Mathematical Olympiad 2000 Question 4 (first question on Day 2). Even though it looks like a physics problem, this problem involves very beautiful mathematics of rational and irrational numbers. What a surprise!

My solution above seems not so complete though since I assume that x_0=0; that is, the car starts at point 0 (or complete, depending on how we interpret the question — it does not ask for all possible values of v, right?) I am not sure how the solution will change. Do you have any idea how to resolve the case? Please let me know in the comment below. I hope you enjoy this problem as much as I do.

Update (21 Dec 2022). A more general version of the lemma would be \{\alpha+\beta\}, \{2\alpha+\beta\}, \{3\alpha+\beta\}, \dots are all < \dfrac{3}{5}. Here, \alpha = \dfrac{10}{v} and \beta = \dfrac{y_0}{v} where 0 \le y_0 < 10. I believe that using similar argument, we can say the condition will hold iff \{\alpha\} = \dfrac{1}{2} and additionally, it must be the case that \{\beta \} < \dfrac{1}{10}. Note that this condition is already sufficient, but I am not sure if this is also a necessary condition.

Note that from condition on \alpha, we must have v=\dfrac{20}{2n+1} for some non-negative integer n. The condition on \beta states that \left \{ \dfrac{y_0(2n+1)}{20} \right \} < \dfrac{1}{10}. Suppose that y_0=4, then we have \left \{ \dfrac{2n+1}{5} \right \} < \dfrac{1}{10}. This is true for example for n=2,7,12,\dots (integers that are 2 modulo 5).

Transitive filtration on (Q,<)

This a bonus question for HW 2 Basic Modal Logic course 2016/2017.

Show that transitive fitration on (\mathbb{Q},<) is a finite list of clusters, maybe interspersed by some irreflexive singletons, no two of which may be adjacent.

What I did is to consider it in four steps.

Finiteness and transitivity

This follows from filtration property (upon finite \Sigma) and the property that R has as a transitive filtration.

Linearity

For all a \ne b, we have Rab \vee Rba.

Proof. WLOG, a<b. Then, we show that for all \diamond \phi \in \Sigma, we have that if b \Vdash \phi \vee \diamond \phi, then a \Vdash \diamond \phi. If b \Vdash \phi, then because a<b, we have a \Vdash \diamond \phi. If b \Vdash \diamond \phi, then there exists some c>b, such that c \Vdash \phi, and thus becaue a<b<c, a \Vdash \diamond \phi\blacksquare

Moreover, we basically have < \subseteq R.

Properties of irreflexive points

If \neg Raa, then there must be some \diamond \phi \in \Sigma, such that a is the greatest number satisfying \phi.

Proof. If \neg Raa, it means that for some \diamond \phi \in \Sigma, we have a \Vdash \phi \vee \diamond \phi but a \nVdash \diamond \phi. Thus, a \Vdash \phi and a \nVdash \diamond \phi. \blacksquare

I also claim no two irreflexive points can be adjacent.

Proof. Suppose that a<b are irreflexive points w.r.t. formula \alpha and \beta, respectively and suppose that they are adjacent in the finite list of filtration. Take any number c \in (a,b) \cap \mathbb{Q}. Clearly, Rac. But a and b are adjacent in the filtration, then we must have Rbc for all such c. This means that for all \diamond \phi \in \Sigma, particularly \diamond \beta, we have c \vDash \beta \vee \diamond \beta implies b \vDash \diamond \beta. But, we have b \nvDash \diamond \beta. \blacksquare

Properties of non-simple cluster

Now, we prove that for any reflexive point, it belongs to a non-simple cluster.

Proof. If there is no \diamond \phi \in \Sigma, then all the points belong to the same cluster. If there exist \diamond \phi, then for reflexive points a, Raa means for all \diamond \phi \in \Sigma, a \vDash \phi \vee \diamond \phi implies a \vDash \Diamond \phi. This means, a must not be the greatest number satisfying formula \phi for any \diamond \phi \in \Sigma. Because there is only finite number of singletons (due to the finiteness of \Sigma, we can choose rational number a_1 >a such that all the numbers in (a,a_1) \cap \mathbb{Q} are all reflexive (not a singleton). Now pick any x \in (a,a_1) \cap \mathbb{Q}. Clearly Rax. We show Rxa now. Because it means for all \diamond \phi \in \Sigma, a \vDash \diamond \phi implies x \vDash \diamond \phi, and because a is not a singleton, we then just need to check a \vDash \diamond \phi implies x \vDash \diamond \phi. Suppose a \vDash \diamond \phi but x \nvDash \diamond \phi. This means that there’s a singleton for \phi \in (a,x) which is a contradiction. Thus a belongs to a cluster which also contains at least the whole (a,a_1) \cap \mathbb{Q}.


Because every point in filtration is either reflexive or irreflexive, these properties show that the result of filtration is a finite list of clusters, may be interspersed by some irreflexive singletons, no two of which are adjacent.


This is approximately what I wrote in my homework. I have not got any feedback yet at the time I am writing this. I hope it got almost full marks (like 8 or more).