The Student Room Group

Probability question

Hi, I am stuck on this question.

The life time T of a particular brand of light bulb is modelled as follows. There is a probability p of the light-bulb blowing immediately (so that T= 0); given that the light bulb does not blow immediately, the probability of it having lifetime τ or less is 1-e^(-λτ)

(i) Write down the cumulative distribution function F(t) of T

I am totally confused, especially as it just says write down. Any help would be appreciated
Reply 1
Original post by Mathstat
Hi, I am stuck on this question.

The life time T of a particular brand of light bulb is modelled as follows. There is a probability p of the light-bulb blowing immediately (so that T= 0); given that the light bulb does not blow immediately, the probability of it having lifetime τ or less is 1-e^(-λτ)

(i) Write down the cumulative distribution function F(t) of T

I am totally confused, especially as it just says write down. Any help would be appreciated


by definition

F(t) </= P(T</=t)
Reply 2
Original post by TeeEm
by definition

F(t) </= P(T</=t)


So it's just 1-e^(-λτ)?
Reply 3
Original post by Mathstat
So it's just 1-e^(-λτ)?


I have not done exponential distributions for a while but that is what I made it
Reply 4
Original post by TeeEm
I have not done exponential distributions for a while but that is what I made it


But where does the lightbulb exploding instantly with probability p come into it?
Reply 5
Original post by Mathstat
But where does the lightbulb exploding instantly with probability p come into it?


I do not what the question is asking, but I would imagine P is merely notation.

As far as your question is concerned it makes sense
the exponential distribution is defined between 0 and inf

F(0) = 0
F(inf) =1

and by differentiation you can find the PDF , mean and all sorts.
Reply 6
Original post by TeeEm
I do not what the question is asking, but I would imagine P is merely notation.

As far as your question is concerned it makes sense
the exponential distribution is defined between 0 and inf

F(0) = 0
F(inf) =1

and by differentiation you can find the PDF , mean and all sorts.


Is this not correct (this is working I just asked my friend for).

P(T<τ|T=/=0)= P(T<τ union T=/=0)/P(T=/=0)

The LHS = 1-e^(-λτ), so P(T<τ union T=/= 0) = P(T<τ)= p(1-e^(-λτ)

Thanks for your time
Original post by Mathstat
But where does the lightbulb exploding instantly with probability p come into it?


I beg to differ with TeeEm (WADR). Although the question says "write down", implying no working is required, I think you do need some in this case:

P(Tt)=P(T=0)+P(0<Tt)P(T\leq t) = P(T=0) + P(0 < T \leq t)
Edit: For t > 0.

And you're told in the question P(0<Tt    T0)P(0<T\leq t\; |\; T \not= 0)

Unusual question for A-level. Is it A-level?

Can you take it from there?
(edited 8 years ago)
Reply 8
Original post by Mathstat
Is this not correct (this is working I just asked my friend for).

P(T<τ|T=/=0)= P(T<τ union T=/=0)/P(T=/=0)

The LHS = 1-e^(-λτ), so P(T<τ union T=/= 0) = P(T<τ)= p(1-e^(-λτ)

Thanks for your time


Original post by ghostwalker
I beg to differ with TeeEm (WADR). Although the question says "write down", implying no working is required, I think you do need some in this case:

P(Tt)=P(T=0)+P(0<Tt)P(T\leq t) = P(T=0) + P(0 < T \leq t)

And you're told in the question P(0<Tt    T0)P(0<T\leq t\; |\; T \not= 0)

Unusual question for A-level. Is it A-level?

Can you take it from there?


I will leave you in the capable hands of ghostwalker, as stats is not my forte ...
All the best.
Reply 9
Original post by ghostwalker
I beg to differ with TeeEm (WADR). Although the question says "write down", implying no working is required, I think you do need some in this case:

P(Tt)=P(T=0)+P(0<Tt)P(T\leq t) = P(T=0) + P(0 < T \leq t)
Edit: For t > 0.

And you're told in the question P(0<Tt    T0)P(0<T\leq t\; |\; T \not= 0)

Unusual question for A-level. Is it A-level?

Can you take it from there?


I am going into my second year of uni and this is on the holiday sheets.

So is the answer (1-p)(1-e^(-λτ)) ?

What is P(0<=T<τ union T=/=0)? Is it just 1-e^(-λτ)?

Thanks for your help
Original post by Mathstat
I am going into my second year of uni and this is on the holiday sheets.

So is the answer (1-p)(1-e^(-λτ)) ?


For t>0,

P(Tt)=P(T=0)+P(0<Tt)P(T \leq t) = P(T=0) + P(0 < T \leq t)

=P(T=0)+P(0<Tt    T0)P(T0)= P(T=0) + P(0 < T \leq t\; | \; T\not= 0)P(T\not=0)

=p+(1eλt)(1p)= p + (1-e^{-\lambda t})(1-p)




What is P(0<=T<τ union T=/=0)? Is it just 1-e^(-λτ)?

Thanks for your help


P(0<=T<τ union T=/=0). I think you mean intersect there, rather than union.

P(0T<tT0)=P(0<T<t)P(0\leq T<t \cap T \not= 0) = P(0 < T < t)

=P(0<Tt    T0)P(T0)= P(0 < T \leq t\; | \; T\not= 0)P(T\not=0)

=(1eλt)(1p)= (1-e^{-\lambda t})(1-p)

I've not been consistent with the < or <= at the upper end as you've used it slightly differently, but it's not important as long as there are no discontinuities, which there aren't for t > 0.

So you cdf will be as I previously wrote for t>0, and "p" for t=0
Edit: The case t=0 is actually covered by the formula - ho hum.
(edited 8 years ago)
Original post by ghostwalker
For t>0,

P(Tt)=P(T=0)+P(0<Tt)P(T \leq t) = P(T=0) + P(0 < T \leq t)

=P(T=0)+P(0<Tt    T0)P(T0)= P(T=0) + P(0 < T \leq t\; | \; T\not= 0)P(T\not=0)



For clarity, by the law of total probability that second line is really:


=P(T=0)+P(0<Tt    T0)P(T0)+P(0<Tt    T=0)P(T=0)= P(T=0) + P(0 < T \leq t\; | \; T\not= 0)P(T\not=0)+P(0 < T \leq t\; | \; T= 0)P(T=0)

But those final terms come to 0, since T can't simultaneously be >0 and =0.

Hence as I originally wrote.
(edited 8 years ago)
Reply 12
Original post by ghostwalker
For t>0,

P(Tt)=P(T=0)+P(0<Tt)P(T \leq t) = P(T=0) + P(0 < T \leq t)

=P(T=0)+P(0<Tt    T0)P(T0)= P(T=0) + P(0 < T \leq t\; | \; T\not= 0)P(T\not=0)

=p+(1eλt)(1p)= p + (1-e^{-\lambda t})(1-p)




P(0<=T<τ union T=/=0). I think you mean intersect there, rather than union.

P(0T<tT0)=P(0<T<t)P(0\leq T<t \cap T \not= 0) = P(0 < T < t)

=P(0<Tt    T0)P(T0)= P(0 < T \leq t\; | \; T\not= 0)P(T\not=0)

=(1eλt)(1p)= (1-e^{-\lambda t})(1-p)

I've not been consistent with the < or <= at the upper end as you've used it slightly differently, but it's not important as long as there are no discontinuities, which there aren't for t > 0.

So you cdf will be as I previously wrote for t>0, and "p" for t=0


Thanks you so much :biggrin:. All my friends were stuck on this as well.

Quick Reply

Latest