2 - 331statistics05b - Quickie Statistics Summary

  • Upload
    alkimia

  • View
    219

  • Download
    0

Embed Size (px)

Citation preview

  • 8/13/2019 2 - 331statistics05b - Quickie Statistics Summary

    1/3

  • 8/13/2019 2 - 331statistics05b - Quickie Statistics Summary

    2/3

    PHYSICS 331 (revised 28 March 2005) p.2 of 3

    1

    x2=

    1

    n2

    1

    N

    In general, the quantity reported for a measurement will be x x . If the

    parent distribution is a normal(Gaussian BR pp. 27-30) distribution, the mostcommon case, then the true mean will be within the range x x 68% of thetime. Stated in other terms, if the set of Nmeasurements is repeated, 68% of the

    time the mean obtained in the second set will lie within the range x x .

    Propagation of Errors

    When the quantity being measured, let's call it u, is some combination of

    independent quantities which we will call x, y, z,...,u=u(x,y,z,...),there is a simple

    general rule for calculating the uncertainty in u,u , given the

    uncertainitiesx ,y ,z ,...in x, y, z,.... This is

    u2=

    u

    x

    x ,y ,z ,

    2

    x2+ u

    y

    x,y ,z ,

    2

    y2+ u

    z

    x,y ,z ,

    2

    z2+...

    Two examples of this rule are of particular interest. The first is the

    situation in which uis the sum or difference of the quantities x, y, z,..., forexample, u= x+ y z. The partial derivatives of uwith respect to x, yand z areeither +1 or 1 so the expression for the uncertainty of u reduces to (BR p.42)

    u2 = x

    2+ y

    2+ z

    2

    The square of the uncertainty of uis the sum of the squares of the uncertainties ofx, y,and z.

    The second example is that in which u can be expressed as the product

    and/or quotient of x, y, and z, for example, u=xy

    z. It is a simple matter to show

    that the general expression reduces in this case to (BR p. 43)

    u

    u

    2

    =x

    x

    2

    +y

    y

    2

    +z

    z

    2

  • 8/13/2019 2 - 331statistics05b - Quickie Statistics Summary

    3/3

    PHYSICS 331 (revised 28 March 2005) p.3 of 3

    Thus the square of thefractional(or relative) uncertainly in u,u

    u, is the sum of

    the squares of thefractional(or relative) uncertainties in x, y, and z.

    For both of the examples above, if the uncertainty in one of the quantitiesx, y, or zis several times that of the uncertainties in the other quantities itdominates the uncertainty in u. For example, if the uncertainty is xis twice that

    in yand z, x = 2y = 2z , the uncertainty in xcontributes 82% of the

    uncertainty in u.