Class details Probability I

9.07 2/10/2004 Reminder: HW 1 due on Friday. HW2 is now on the web. Its due Friday of next week. Readings in Probability now on the web. Reminder: Office hours today, 3-4 pm

Probability and gambling

De Mere: Which is more likely, rolling at least one 6 in 4 rolls of a single die, or rolling at least one double 6 in 24 rolls of a pair of dice? De Mere reasoned they should be the same:
Chance of one 6 in one roll = 1/6 Average number in 4 rolls = 4(1/6) = 2/3 Chance of one double 6 in one roll = 1/36 Average number in 24 rolls = 24(1/36) = 2/3

Basic definitions
Random experiment = observing the outcome of a chance event. Elementary outcome = any possible result of the random experiment = Oi Sample space = the set of all elementary outcomes.

Why, then, did it seem like he lost more often with the second gamble? He asked his friend Pascal, and Pascal & Fermat worked out the theory of probability.

Example sample spaces

Tossing a single coin:
{H, T}

Sample space for a pair of dice

Tossing two coins:

{HH, TH, HT, TT}

One roll of a single die:

Each pair is an elementary outcome.

Fair coin or die

For a fair coin or die, the elementary outcomes have equal probability
P(H) = P(T) = 0.5 P(1 spot) = P(2) = P(3) = P(4) = P(5) = P(6) = 1/6

Properties of probabilities
P(Oi) 0
Negative probabilities are meaningless

The total probability of the sample space must equal 1.

If you roll the die, one of the elementary outcomes must occur.

Of course, the coin or die might not be fair

P = .15 .10 .25 .15 .15 .20

How do we decide what these probabilities are?

1. Probability = events relative frequency in the population.
Look at every member of the population, and record the relative frequency of each event. Often you simply cant do this.

Events
An event is a set of elementary outcomes. The probability of an event is the sum of the probabilities of the elementary outcomes. E.G. tossing a pair of dice:

2. Estimate probability based on the relative frequency in a (large) sample.

Not perfect, but feasible.

3. Classical probability theory: probability based on an assumption that the game is fair.
E.G. heads and tails equally likely. Similarly, might otherwise have a theoretical model for the probabilities.

Event A: Dice sum to 3

Event B: Dice sum to 6

Event C: White die = 1

Event D: Black die = 1

Combining events
E AND F: both event E and event F occur E OR F: either event E occurs, or event F does, or both NOT E: event E does not occur

C OR D: W=1 OR B=1

The addition rule

P(W=1) = 6/36 P(B=1) = 6/36 P(W=1 or B=1) P(W=1) + P(B=1) P(E or F) = P(E) + P(F) P(E and F)
Subtract the region of overlap, so you dont count it twice.

A or B: Dice sum to 3, or sum to 6

Mutually exclusive events

Events E and F are mutually exclusive if the two events could not have both occurred.
P(E and F) = 0. The events have no elementary outcomes in common. (Theres no overlap in our sample space diagram.)

Another example: P(sum=7) = ?

If E and F are mutually exclusive,

P(E or F) = P(E) + P(F)

The elementary outcomes are mutually exclusive.

P(any Oi) = P(O1)+P(O2)+P(ON) = 1 P(sum=7) = 6/36.

Another example: P(B=5 or 4) = ?

P(sum=7 or (B=5 or 4)) = ?

P(sum=7 and (B=5 or 4)) = P({2, 5}, {3, 4}) = 2/36 P(sum=7 or (B=5 or 4)) = 6/36 + 12/36 2/36 = 16/36

P(B=5 or 4) = 6/36 + 6/36 = 12/36.

P(sum=7 or (B=5 or 4))

De Mere revisited
Wanted to know what is the probability of getting at least one 6 in 4 tosses of a die. P(1st=6 or 2nd=6 or 3rd=6 or 4th=6) P(1st=6) = P(2nd=6) = P(3rd=6)=P(4th=6) = 1/6 Are these events mutually exclusive?
No, you could get a 6 on both the 1st & 2nd tosses, for example. So De Mere was incorrect. P 4(1/6)

P(A or B) = P(A) + P(B) P(A and B) = 6/36 + 12/36 2/36 = 16/36.

The addition formula, continued

P(E or F) = P(E) + P(F) P(E and F) You will probably rarely use this formula except for simple cases! It gets complicated quickly if you want to compute P(E or F or G or ) Example: P(at least one 6 in 3 tosses of a die)=?

2 tosses of a die

1st toss (W)

2nd toss (B)

3 tosses of a die

3 tosses of a die

3rd=1

3rd=2

3rd=3

3rd=1

3rd=2

3rd=3

3rd=4

3rd=5

3rd=6

3rd=4

3rd=5

3rd=6

P(one 6 in 3 rolls) = P(1st=6) + P(2nd=6) + P(3rd=6) +

P(one 6 in 3 rolls) = ... P(6 in 1st & 2nd)

3 tosses of a die

3 tosses of a die

3rd=1

3rd=2

3rd=3

3rd=1

3rd=2

3rd=3

3rd=4

3rd=5

3rd=6

3rd=4

3rd=5

3rd=6

P(one 6 in 3 rolls) = ... P(6 in 1st & 3rd)

P(one 6 in 3 rolls) = ... P(6 in 2nd & 3rd)

3 tosses of a die
P(at least one 6 in 3 tosses) = P(6 in 1st) + P(6 in 2nd) + P(6 in 3rd) P(6 in 1st & 2nd) P(6 in 1st & 3rd) P(6 in 2nd & 3rd) + P(6 in 1st & 2nd & 3rd) = 1/6+1/6+1/6-1/36-1/36-1/36+1/216 = 91/216 Phew... It only gets worse from here. De Mere probably doesnt want to calculate P(at least one 6 in 4 tosses) this way. Luckily there are other ways to go about this.

3 tosses of a die Venn diagram

A = 6 in 1st B = 6 in 2nd or

P(A + B + C) = P(A) + P(B) + P(C)

C = 6 in 3rd

3 tosses of a die Venn diagram

A and B=AB A B P(A + B + C) = P(A) + P(B) + P(C) P(AB)

3 tosses of a die Venn diagram

P(A + B + C) = P(A) + P(B) + P(C) P(AB) P(AC)

3 tosses of a die Venn diagram

3 tosses of a die Venn diagram

P(A + B + C) = P(A) + P(B) + P(C) P(AB) P(AC) P(BC)

P(A + B + C) = P(A) + P(B) + P(C) P(AB) P(AC) P(BC) + P(ABC)

3 tosses of a die Venn diagram

Venn diagrams, II
Just as with sample space diagrams, lack of overlap means two events are mutually exclusive. Consider the event A, but not A and B) = A-AB. Are the events B, and AAB mutually exclusive? Yes.

P(A + B + C) = P(A) + P(B) + P(C) P(AB) P(AC) P(BC) + P(ABC)

Conditional probability
The probability that event A will occur, given that event C has already occurred. P(A|C) P(dice sum to 3) = P({1,2},{2,1}) = 2/36. Suppose we have already tossed the black die, and got a 2. Given that this has already occurred, what is the probability that the dice will sum to 3?

P(A|C) = P(sum to 3|B=2)

10

Another formula
P(A|C) = P(A and C)/P(C) E.G. P(sum to 3|B=2) = P(B=2 & sum to 3)/P(B=2)

The formula in action

P(A & C) = 1/36 P(C) = 1/6

P(A|C) = (1/36) / (1/6) = 1/6.

Rearranging to get the multiplication rule

P(E|F) = P(E and F)/P(F) Multiplication rule: P(E and F) = P(F) P(E|F) Another example: What is the probability that the sum=6 and the white die came up a 3?

P(sum=6 and W=3) = P(sum=6) P(W=3|sum=6)

P(sum=6)=5/36. P(W=3|sum=6)=1/5. P(sum=6 & W=3)=1/36.

11

Some notes
P(E|E) = P(E and E)/P(E) = P(E)/P(E) = 1
Once an event occurs, its certain.

P(W=1|B=1)

If E and F are mutually exclusive, P(E|F) = P(E and F)/P(F) = 0/P(F) = 0

Once F has occurred, E is impossible, because the two are mutually exclusive.

Swapping E & F -> P(F|E) = P(E and F)/P(E) -> P(F|E) P(E) = P(E and F) = P(E|F) P(F) -> P(E) P(F|E) = P(F) P(E|F) Another example: what is P(W=1 | B=1)?

P(W=1|B=1) = 1/6 = P(W=1).

Independent events
Two events E and F are independent if the occurrence of one has no effect on the probability of the other. E.G. the roll of one die has no effect on the roll of another (unless theyre glued together or something). If E and F are independent, this is equivalent to saying that P(E|F) = P(E), and P(F|E) = P(F) Special multiplication rule for independent events: P(E and F) = P(E) P(F)

A last rule (an easy one)

P(not E) = 1-P(E)
e.g. P(failed to roll a 6)

P(roll a 6) = 1/6 P(dont roll a 6) = 5/6

12

Probability rules
Addition rule:

Now were ready to solve De Meres problem (the easy way)

What is the probability of getting at least one 6 on 4 rolls of a die? Remember how icky the addition rule got for P(A or B or C or D)? Problems like this are often easier to solve in reverse. Find the probability of the event NOT happening.
But sometimes figuring out what the inverse is can be tricky

P(E or F) = P(E) + P(F) P(E and F) P(E or F) = P(E) + P(F) P(E and F) = P(E|F) P(F) = P(F|E) P(E) P(E and F) = P(E) P(F) P(not E) = 1 P(E)

Addition rule for E & F mutually exclusive: Multiplication rule: Multiplication rule, independent E & F: Inverse rule:

What is the probability of getting at least one 6 one 4 rolls of a die?

P(not E) = P(roll 4 times and dont roll a 6) P(dont roll a 6 on one roll) = 5/6 We know rolls are independent, so P(dont roll a 6 on any roll) = P(no 6 on 1st) P(no 6 on 2nd) P(no 6 on 3rd) P(no 6 on 4th) = (5/6)4 = 0.482 P(E) = 1 P(not E) = 1-0.482 = 0.518

What is the probability of getting at least one pair of 6s on 24 rolls of a pair of dice?
Again, solve the problem in reverse. P(not E) = P(no pair of 6s on any of 24 rolls) P(pair of 6s on a single roll) = 1/36. P(no pair of 6s on a single roll) = 35/36. P(not E) = (35/36)24 = 0.509 P(E) = 1 P(not E) = 1 0.509 = 0.491 De Mere was right this event is less likely to occur than rolling at least one 6 in 4 throws. (Its a pretty small difference he must have gambled a lot and paid close attention to the results!)

13

P(loaded dice sum to 3)=? P({1,2,3,4,5,6}) = {.15, .10, .25, .15, .15, .20}

Another way to solve problems with fair dice & coins

Probability of event made up of equally probable elementary outcomes = (# of outcomes that are part of the event)/(total number of outcomes) E.G. P(at least one 6 in 3 throws) How many total outcomes? How many outcomes with at least one 6?

P(sum to 3) = 0.15*0.10 + 0.10*0.15 = 0.03

A useful visualization
Box diagrams:

Total number of outcomes in 3 throws of a die

6 possibilities for each throw
6 6 2nd 6 3rd 1st

Number of possibilities in 1st throw.

In 2nd throw

In

3rd

throw

Total number of outcomes from 3 throws = 6*6*6 = 216.

To get total number of possible outcomes, multiply the numbers in the 3 boxes.

14

Number of outcomes including at least one 6

First, assume the 1st throw is a 6:
1 1st 6 2nd 6 3rd These can be anything they like. 6 possibilities each.

Number of outcomes including at least one 6

Next, assume the 2nd throw is a 6:
5 1st 1 2nd 6 3rd This can be anything. 6 possibilities. One possibility its a 6

Only one possibility its a 6

5 possibilities we already counted the outcomes where this is a 6.

Number of outcomes including at least one 6

Finally, assume the 3rd throw is a 6:
5 1st 5 2nd 1 3rd One possibility a 6

Another fairly easy and reliable way to solve problems like this
Probability of event made up of equally probable elementary outcomes = (# of outcomes that are part of the event)/(total number of outcomes) E.G. P(at least one 6 in 3 throws) How many total outcomes?
6*6*6 = 216

5 possibilities each we already counted the outcomes where these are 6s.

How many outcomes with at least one 6?

1*6*6 + 5*1*6 + 5*5*1 = 36+30+25 = 91

P(E) = 91/216.

15

Check: do the problem the other way

P(no 6 in 3 throws) = (5/6)3 P(at least one 6 in 3 throws) = 1 (5/6)3 = 216/216 125/216 = 91/216 It worked!

16