LUMS, PHY100, Formula Sheet for Uncertainties and Linear Fitting

Probability distribution functions:

FIG. 1: The probability distribution functions associated with measurements.

Propagation of uncertainties: If a quantity q = q(x, y, z) is measured using some input

variables x, y and z which are measured with uncertainties ux , uy and uz , respectively, then
uq can also be find out using the Taylor series approximation given as,
s 2  2  2
q q q
uq = ux + uy + uz .
x y z

Some examples are:

q
q = x y, uq = u2x + u2y ,
q
q = xy, uq = (yux )2 + (xuy )2 ,
s 
2  2
x ux xuy
q = , uq = + ,
y y y2
p
q = xm , uq = (mxm1 ux )2
P
i xi
Mean (or average): x = N
qP
2
i (xi x)
Standard uncertainty in the mean: ux = N 1
.

1
 LUMS, PHY100, Formula Sheet for Uncertainties and Linear Fitting

P
Weighted average: xw = Pwi xi
wi
q
N i wi (xi x)2
P
Standard uncertainty in the weighted average: uxw = P
(N 1) i wi
.

Slope (m) and intercept (c) with equal weights:

PN
N N
P PN PN
i yi (xi x) i xi y i i xi i yi
m = PN or m = (1)
i (xi x)
2 D
PN PN
yi N
PN
x2i
P
i i i xi i xi yi
c = y mx or c= , (2)
D
X X
where, D = N x2i ( xi )2 . (3)
i i

Uncertainty in the slope m and intercept c is given as,

r
N
um = uy ,
D
rP
i xi
uc = uy .
D
Slope m and intercept c with unequal weights

The weights are reciprocal squares of the total uncertainty (uTotal ),

1
wi = . (4)
u2total
The mathematical relationships for slope (m) and intercept (c) are,
i wi i wi (xi yi ) i (wi xi ) i (wi yi )
m = , (5)

i (wi x2i ) i (wi yi ) i (wi xi ) i (wi xi yi )

c = , (6)

where x is the independent variable, y is the dependent variable and w is the weight.
The expressions for the uncertainties in m and c are,
r
i w i
um = , (7)

r
i (wi x2i )
uc = , (8)

where, = i wi i (wx2i ) (i wi xi )2 (9)