Porisma

Worked solution

Follow the reroll rule and translate a difference into an absolute value

An unbiased ordinary die is thrown. If its score is not 3, the result is that score. If it is 3, the die is thrown again and the result is the absolute difference of the two scores. Let R denote the result. Show that P(R=0)=1/36P(R=0)=1/36 and P(R=1)=2/9P(R=1)=2/9, and find the distribution of R. marks

Find

(i) the mean result, marks

(ii) the probability that the difference between (R−1)2(R-1)^2 and 3 is at most the mean from (i). marks

[2006/PJC/II/29(OR)]

CJC · Review problems · Question 6

Answer

P(R=0)=1/36P(R=0)=1/36, P(R=1)=2/9P(R=1)=2/9; full distribution above; E(R)=13/4E(R)=13/4; (ii)1/2.

Zero and one: retain the two show-that calculations

Let X1,X2X_1,X_2 denote the first and possible second throws. P(R=0)=P(X1=3,X2=3)=(1/6)(1/6)=1/36P(R=0)=P(X_1=3,X_2=3)=(1/6)(1/6)=1/36. For result 1, P(R=1)=P(X1=1)+P(X1=3,X2=4)+P(X1=3,X2=2)=1/6+(1/6)(1/6)+(1/6)(1/6)=2/9P(R=1)=P(X_1=1)+P(X_1=3,X_2=4)+P(X_1=3,X_2=2)=1/6+(1/6)(1/6)+(1/6)(1/6)=2/9. The source’s last subscript is printed X3X_3; it must be X2X_2, since there is no third throw.

How did this part go?

Retain all other source result calculations

P(R=2)=P(X1=2)+P(X1=3,X2=5)+P(X1=3,X2=1)=1/6+(1/6)(1/6)+(1/6)(1/6)=2/9P(R=2)=P(X_1=2)+P(X_1=3,X_2=5)+P(X_1=3,X_2=1)=1/6+(1/6)(1/6)+(1/6)(1/6)=2/9. The final subscript here is also corrected from the source’s X3X_3 to X2X_2.

P(R=3)=P(X1=3,X2=6)=(1/6)(1/6)=1/36P(R=3)=P(X_1=3,X_2=6)=(1/6)(1/6)=1/36.

P(R=4)=P(X1=4)=1/6P(R=4)=P(X_1=4)=1/6; P(R=5)=P(X1=5)=1/6P(R=5)=P(X_1=5)=1/6; P(R=6)=P(X1=6)=1/6P(R=6)=P(X_1=6)=1/6.

How did this part go?

Full source distribution

rP(R=r)P(R=r)
01/361/36
12/92/9
22/92/9
31/361/36
41/61/6
51/61/6
61/61/6
How did this part go?

(i) Retain all seven mean terms

E(R)=0(1/36)+1(2/9)+2(2/9)+3(1/36)+4(1/6)+5(1/6)+6(1/6)=13/4E(R)=0(1/36)+1(2/9)+2(2/9)+3(1/36)+4(1/6)+5(1/6)+6(1/6)=13/4.

How did this part go?

(ii) Retain every source inequality stage

“Difference between” is interpreted as absolute difference, as in the source working. Therefore

P ⁣(∣(R−1)2−3∣≤134)=P ⁣(−134≤(R−1)2−3≤134).\begin{aligned}&P\!\left(|(R-1)^2-3|\le\frac{13}4\right)\\ & =P\!\left(-\frac{13}4\le(R-1)^2-3\le\frac{13}4\right).\end{aligned}

Add 3 to get P(−1/4≤(R−1)2≤25/4)P(-1/4\le(R-1)^2\le25/4). Since a square is nonnegative, this is P((R−1)2≤25/4)=P(∣R−1∣≤5/2)P((R-1)^2\le25/4)=P(|R-1|\le5/2).

This is P(−5/2≤R−1≤5/2)=P(−3/2≤R≤7/2)P(-5/2\le R-1\le5/2)=P(-3/2\le R\le7/2). Using the integer support, it becomes P(0≤R≤3)=1/36+2/9+2/9+1/36=1/2P(0\le R\le3)=1/36+2/9+2/9+1/36=1/2.

How did this part go?

Common mistakes

An initial 3 is not automatically result 3. The difference in part (ii) is an absolute difference.

Solution notes

All initial and numbered asks/marks, each source event/product, full seven-value table, complete expectation and all inequality stages retained. Source X3 subscripts in two reroll cases corrected explicitly to X2, native originals unchanged.

Read the hint