Linear combinations of Normal random variables

Linear combinations of Normal random variables

Spec 4.1 · Edexcel A-Level Further Maths: Revision notes

Studying this course? Keep your progress in one place.

Key facts

  • A linear combination of independent Normal variables is Normal.
  • aX±bY∼N(aμx±bμy, a2σx2+b2σy2)aX\pm bY\sim N(a\mu_x\pm b\mu_y,\ a^2\sigma_x^2+b^2\sigma_y^2): means combine as written, variances always add.
  • To compare PP and AA, test the difference: P(P>A)=P(P−A>0)\mathrm{P}(P>A)=\mathrm{P}(P-A>0).
  • 2X∼N(2μ,4σ2)2X\sim N(2\mu,4\sigma^2) but X1+X2∼N(2μ,2σ2)X_1+X_2\sim N(2\mu,2\sigma^2): a multiple is not the same as two copies.
  • The mean of nn copies is Xˉ∼N(μ,σ2n)\bar X\sim N\left(\mu,\frac{\sigma^2}{n}\right).

Combining Normal variables

A linear combination of independent Normal variables is Normal: means combine, variances always add.

If X∼N(μx,σx2)X\sim N(\mu_x,\sigma_x^2) and Y∼N(μy,σy2)Y\sim N(\mu_y,\sigma_y^2) are independent, any linear combination is also Normal. The result is stated without proof.

The means combine as the variables do, but the variances are added, each with its coefficient squared, even when the variables are subtracted. It extends to more variables and to a constant: aX+c∼N(aμ+c,a2σ2)aX+c\sim N(a\mu+c,a^2\sigma^2).

1001201401601802002202400.0050.010.0150.020.0250.030.0350.04xyAP
Two independent Normal variables, A∼N(150,122)A\sim N(150,12^2) and P∼N(200,152)P\sim N(200,15^2); any linear combination of them is also Normal.
  • Meanaμx±bμya\mu_x\pm b\mu_y
  • Variancea2σx2+b2σy2a^2\sigma_x^2+b^2\sigma_y^2 (always add)

X∼N(10,32)X\sim N(10,3^2) and Y∼N(4,22)Y\sim N(4,2^2) are independent. What is the distribution of X−YX-Y?

Sums and differences

Form the difference and test it against 00.

For independent A∼N(150,122)A\sim N(150,12^2) and P∼N(200,152)P\sim N(200,15^2), the total T=A+PT=A+P and the difference D=P−AD=P-A have the same variance, since variances add.

To compare two variables, form their difference and test against 00: write P(P>A)\mathrm{P}(P>A) as P(P−A>0)\mathrm{P}(P-A>0) first.

−40−20204060801001201400.0050.010.0150.020.025xyD = P − A
D=P−A∼N(50,369)D=P-A\sim N(50,369): the shaded area above 0 is P(P>A)=0.995\mathrm{P}(P>A)=0.995.

T = A + P

Mean:
350
Variance:
369
Distribution:
N(350, 369)

D = P − A

Mean:
50
Variance:
369
Distribution:
N(50, 369)

Worked example

Find P(P>A)\mathrm{P}(P>A).

To find P(P>A)\mathrm{P}(P>A) for independent Normal PP and AA, which variable do you test?

Multiples and copies

Both have mean 2μ2\mu, but 2X2X has variance 4σ24\sigma^2 and X1+X2X_1+X_2 only 2σ22\sigma^2.

For X∼N(μ,σ2)X\sim N(\mu,\sigma^2): 2X∼N(2μ,4σ2)2X\sim N(2\mu,4\sigma^2) is one rod doubled, but X1+X2∼N(2μ,2σ2)X_1+X_2\sim N(2\mu,2\sigma^2) is two independent rods. Two independent values partly cancel each other's variation, so 2X2X varies more.

For nn copies, the sum is N(nμ,nσ2)N(n\mu,n\sigma^2) and the mean is Xˉ∼N(μ,σ2n)\bar X\sim N\left(\mu,\frac{\sigma^2}{n}\right).

97.59898.59999.5100100.5101101.5102102.50.10.20.30.40.50.60.70.8xyX1 + X22X
Same mean, different spread: X1 + X2 (narrow) and 2X (wide).

Example: X∼N(50,0.42)X\sim N(50,0.4^2) gives X1+X2∼N(100,0.32)X_1+X_2\sim N(100,0.32) but 2X∼N(100,0.64)2X\sim N(100,0.64).

X∼N(5,3)X\sim N(5,3). What is the variance of X1+X2X_1+X_2 and of 2X2X?

Several variables

Add the means with their multiples, add the variances with squared multiples, then standardise.

Men M∼N(80,82)M\sim N(80,8^2) and women W∼N(65,62)W\sim N(65,6^2) are independent. For the total of three men and two women, add three men's variances and two women's.

State the distribution in full as N(mean,variance)N(\text{mean},\text{variance}) before standardising, and say the variables are assumed independent.

3003203403603804004204400.0050.010.0150.020.0250.03xyT
T∼N(370,264)T\sim N(370,264): the shaded area above 400 is P(T>400)=0.0324\mathrm{P}(T>400)=0.0324.
  • Sum of n copiesN(nμ, nσ2)N(n\mu,\ n\sigma^2)
  • Mean of n copiesXˉ∼N(μ, σ2n)\bar X\sim N\left(\mu,\ \dfrac{\sigma^2}{n}\right)

Worked example

Find P(T>400)\mathrm{P}(T>400) where T=M1+M2+M3+W1+W2T=M_1+M_2+M_3+W_1+W_2.

For Mˉ−W\bar M-W with Mˉ\bar M the mean of four men: mean 80−65=1580-65=15 and variance 644+36=52\frac{64}{4}+36=52.

Mean of four men Mˉ\bar M, each N(80,82)N(80,8^2). What is Var(Mˉ)\mathrm{Var}(\bar M)?

Try an exam question

The independent random variables X∼N(25,32)X\sim N(25,3^2) and Y∼N(20,22)Y\sim N(20,2^2). Find P(X>Y)\mathrm{P}(X>Y).

[4 marks]

That's the notes covered.

Carry on to the next subtopic.

Try a question

The total mass T=A+PT=A+P of one apple and one pear has distribution

1 mark, marked against the mark scheme

Revision notes