Error Analysis. Experimental Error –The uncertainty obtained in a measurement of an experiment...
-
Upload
natalie-ingledue -
Category
Documents
-
view
223 -
download
1
Transcript of Error Analysis. Experimental Error –The uncertainty obtained in a measurement of an experiment...
Error Analysis
Experimental Error
• Experimental Error – The uncertainty obtained in a
measurement of an experiment– Results can from systematic and/or random
errors• Blunders• Human Error• Instrument Limitations
– Relates to the degree of confidence in an answer
– Propagation of uncertainties must be calculated and taken into account
Experimental Error
It is impossible to make an exact measurement. Therefore, all experimental results are wrong. Just how wrong they are depends on the kinds of errors that were made in the experiment.
As a science student you must be careful to learn how good your results are, and to report them in a way that indicates your confidence in your answers.
Types of Errors
• Systematic Errors– These are errors caused by the way in
which the experiment was conducted. In other words, they are caused by the design of the system or arise from flaws in equipment or experimental design or observer
– Sometimes referred to as determinate errors
– Reproducible with precision– Can be discovered and corrected
Systematic Error
The cloth tape measure that you use to measure the length of an object had been stretched out from years of use. (As a result, all of your length measurements were too small)
The electronic scale you use reads 0.05 g too high for all your mass measurements (because it was improperly zeroed at the beginning of your experiment).
Examples:
Detection of Systematic Errors
• Analyze samples of known composition– Use standard Reference material– Develop a calibration curve
• Analyze “blank” samples– Verify that the instrument will give a zero
result• Obtain results for a sample using
multiple instruments– Verifies the accuracy of the instrument
How to Eliminate Systematic Errors
How would you measure the distance between two parallel vertical lines
- most would pull out a ruler, align one end with one bar, read of the distance.
- You should put ruler down randomly (as perpendicular as you can). Note where each mark hits the ruler, then subtract the two readings. Repeat a number of times and average the result.
- Minimize the number of human operations you can
Elimination of systematic error can best be accomplished by a well planned and well executed experimental procedure
Types of Errors• Random Errors
– Sometimes referred to as indeterminate errors or noise errors
– Arises from things that cannot be controlled
• Variations in how an individual or individuals read the measurements
• Instrumentation noise– Always present and cannot always be
corrected for, but can be treated statistically
– The important property of random error is that it adds variability to the data but does not affect average performance for the data
Random Errors
Examples:
You measure the mass of a ring three times using the same balance and get slightly different values: 12.74 g, 12.72 g, 12.75 g
The meter stick that is used for measuring, slips a little when measuring the object
Accuracy or Precision
• PrecisionReproducibility of resultsSeveral measurements afford the same
results Is a measure of exactness
• AccuracyHow close a result is to the “true”
value“True” values contain errors since they
too were measured Is a measure of rightness
Accuracy vs Precision
= Accuracy Precision
3 NO NO
7.18281828 NO YES
3.14 YES NO
3.1415926 YES YES
Calculating Errors
Terminology Significant Figures – minimum number of
digits required to express a value in scientific notation without loss of accuracy
Absolute Uncertainty – margin of uncertainty associated with “a” measurement
Relative Uncertainty – compares the size of the absolute uncertainty with the size of its associated measurement (a percent)
Propagation of Uncertainty – The calculation to determine the uncertainty that results from multiple measurements
Significant Figures
How to determine which digits are Significant Write the number as a power of 10 Zero’s are significant and must be included when
they occur• In the middle of a number• At the end of a number on the right hand side of
the decimal point– This implies that you know the value of a measurement
accurately to a specific decimal point
The significant figures (digits) in a measurement include all digits that can be known precisely, plus a last digit that is an estimate.
Significant Figures
Let’s look at 123.45
1.2345x102
Scientific Notation
We have 5 significant digits
Let’s look at 0.000123
1.23x10-4
We have 3 significant digits
Significant Figures
Determine the number of significant digits in the following numbers:
Significant Figures
142.7 1.427x102
4 significant digits
142.70 1.4270x102
5 significant digits
Significant Figures
0.000006302 6.302x10-6
4 significant digits
0.003050 3.050x10-3
4 significant digits
Significant Figures
9.250x104 9.250x104
4 significant digits
10.003x104 1.0003x105
5 significant digits
Significant Figures
9000 9x103
1 significant digit
9000. 9.000x103
4 significant digits
Significant Figures
The last significant digit in a measured quantity is the first digit of uncertainty
Significant Figures
Certain values 1 degree of uncertainty
True expression
Absorbance 0.23 0.234 0.234 ± 0.001
% Transmittance 58 58.3 58.3 ± 0.1
Determine the significant figures from the diagram below
Significant Figures
When adding or subtracting, the last digit retained is set by the first doubtful number.
When multiplying or dividing, the number of significant digits you use is simply the number of significant figures as is in the term with the fewest significant digits.
Adding Significant Digits
4503+34.90+550= 5090
3 is the first doubtful number
0 is the first doubtful number
5 is the first doubtful number
The 87.9 are the doubtful numbers
3 significant digits
Via Calculator: 5087.9
Adding Significant Digits
2456.2345+23.21=
23400.00+111.49=
23400+111.49=
234000-2340=
2479.44
23511.49
23500
232000
2479.4445
23511.49
23511.49
231660
Multiplying Significant Digits
2.7812x1.7= 4.72804
Rounded to 4.7 because 1.7 only has 2 significant digits
4.7
Multiplying Significant Digits
14.200x3.2400=
1.00x150.03=
1200x1.234=
45.35.2345=
48.008
150.03
1480.8
8.654121…
48.008
150
1500
8.65
Rounding
Rounding is the process of reducing the number of significant digits in a number. The result of rounding is a "shorter" number having fewer non-zero digits yet similar in magnitude. The result is less precise but easier to use. There are several slightly different rules for rounding.
Rounding
Common method• This method is commonly used, for example
in accounting.• Decide which is the last digit to keep. • Increase it by 1 if the next digit is 5 or more
(this is called rounding up) • Leave it the same if the next digit is 4 or less
(this is called rounding down) • Example: 7.146 rounded to hundredths is
7.15 (because the next digit [6] is 5 or more).
RoundingThis method is also known as statistician's rounding . It is identical to the common method of rounding except when the digit(s) following to rounding digit start with a five and have no non-zero digits after it. The new algorithm is:
Decide which is the last digit to keep. Increase it by 1 if the next digit is 6 or more, or a 5 followed by one or
more non-zero digits. Leave it the same if the next digit is 4 or less Round up or down to the nearest even digit if the next digit is a five
followed (if followed at all) only by zeroes. That is, increase the rounded digit if it is currently odd; leave it if it is already even.
Examples:
7.016 rounded to hundredths is 7.02 (because the next digit (6) is 6 or more)
7.013 rounded to hundredths is 7.01 (because the next digit (3) is 4 or less)
7.015 rounded to hundredths is 7.02 (because the next digit is 5, and the hundredths digit (1) is odd)
7.045 rounded to hundredths is 7.04 (because the next digit is 5, and the hundredths digit (4) is even)
7.04501 rounded to hundredths is 7.05 (because the next digit is 5, but it is followed by non-zero digits)
Increasing Precision with Multiple Measurements
One way to increase your confidence in experimental data is to repeat the same experiment many times.
When dealing with repeated measurements, there are three important statistical quantities
Mean (or average)
Standard Deviation
Standard Error
Mean
What is it:
An estimate of the true value of the measurement
Statistical Interpretation:
The central value
Symbol:
x
Standard Deviation
What is it:
A measure of the spread in the data
Statistical Interpretation:
You can be reasonably sure (about 70% sure) that if you repeat the same experiment one more time, that the next measurement will be less than one standard deviation away from the average
Symbol:
Use you calculator or computer to determine the Standard Deviation.
Standard Error
What is it:
An estimate in the uncertainty in the average of the measurements
Statistical Interpretation:
You can be reasonably sure (about 70% sure) that if you repeat the entire experiment again with the same number of repetitions, the average value from the new experiment will be less than one standard deviation away from the average value of this experiment
Symbol:M
N
Standard Error Example
Measurements: 0.32, 0.54, 0.44, 0.29, 0.48
Calculate the Mean:
0.41
Calculate the Standard Deviation:
0.09
Calculate the Standard Error:
0.04
MN
Therefore: 0.41±0.04
Use this technique to determine the uncertainty if you do not know the uncertainty of a measurement, but have multiple measurements of the value.
Propagation of Uncertainty
• Since measurements commonly will contain random errors that lead to a degree of uncertainty, arithmetic operations that are performed using multiple measurements must take into account this propagation of errors when reporting uncertainty values
Systematic Errors
Errors calculated from data are Random Errors
Errors from the instrument are called System Errors (usually labeled on instrument or told by instructor as a percent)
15.23 0.05 0.17
random systemk k k
22
15.23 0.18
15.2 0.2
ran systemk k k
or
Error Propagation
There are 3 different ways of calculating or estimating the uncertainty in calculated results
Significant digits (The easy way out)Useful when a more extensive uncertainty analysis is not needed.
Error Propagation (Not as bad as it looks)Useful for limited number or single measurements
Statistical Methods (When you have lots of numbersUseful for many measurements
Dependent Error Propagation
Adding and Subtracting
2 2 2...e x y z
Multiplying and Dividing
2 2 2
...x y z
e Ex y z
Average
v v
Dependent (approx)
(121)+(52)-(73)
(121)*(52)*(73)
2 2 2
1 2 312 5 7 12 5 7
12 5 7
420 249
If the Average is 25, then 25 5
2 2 212 5 7 1 2 3
10 4
Propagation of Errors Basic Rule
If x and y have independent random errors and , then error in z=x+y is
2 2z x y
x y
3 0.1
4 0.2
x
y
3 4
7
z
2 20.1 0.2
0.223
z
7 0.2Therefore we have
Adding and Subtracting
Adding and Subtracting
1.76 (0.03) + 1.89 (0.02) – 0.59 (0.02) =
Z=1.76+1.89-0.59=3.06
2 2 2
3
0.03 0.02 0.02
1.7 10
0.041231056
z
Therefore Z=3.06 0.04
Propagation of Errors Basic Rule
If x and y have independent random errors and , then error in z=xy is 22
x yz z
x y
x y
3 0.1
4 0.2
x
y
3 4
12
z
2 20.1 0.2
123 4
0.72211102555
z
12 0.7Therefore we have
Multiplying and Dividing
Multiplying and Dividing
[1.76(0.03) x 1.89(0.02)] / 0.59(0.02) =
Z=
1.76 1.89
5.6379661020.59
2 2 21.76 1.89 0.03 0.02 0.02
0.59 1.76 1.89 0.59
0.222083034
z
Therefore z=5.6 0.2
Putting it Together
x=200 2Y=50 2z=40 2
xq
y z
x, y, z are independent, find q
Let d=y-z 2 250 40 2 2
10 2 2
10 3
d
200
1020
xq
d
2 22 3
20200 10
20 0.0901
6
q
Therefore q=20 6
What about Functions of 1 Variable
Find error for with s=20.023V s
We cannot use because
s, s, s are not independent
2 2 2s s s
z zs s s
What to the rescue???
Calculus
V=s3
Let’s take the derivative of V with respect to s 23
dVs
ds
2
2
3
3 2 0.02
0.24
dV s ds
Therefore the value for V is
V=80.2
32
8
V
Think of dV and ds as a small change (error) in V and s
x=100 6 then find V when V x
A function of one variable… CALCULUS
1
2
1
21
2
26
2 1000.3
V x
x
dVx
dxdx
dVx
Therefore V=10.0 0.3
100
10
V x
What about a Function with a Constant?
You measure the diameter of a circle to be 20.02
Determine the area of the circle2
2
2
2
1
4
A r
d
d
Calculus
1
21
21
2 0.0220.06
dAd
dd
dA d dd
2
21
A r
The area is 3.14 0.06
If q=f(x1, x2, x3, …xn)
22 2
1 21 2
... nn
q q qq x x x
x x x
then
Let q=x1+x2
2 2
1 21 2
2 2
1 2
2 2
1 1
1 1
q qq x x
x x
x x
x x
2 2
1 1q x x Previous rule
PROOF
If q=f(x1, x2, x3, …xn)
Let q=x1*x2
2 2
1 1
1 2
x xq q
x x
Previous rule
PROOF
2 2
1 21 2
2 2
2 1 1 2
2 22 22 1 1 2
2 22 2
1 22 21 2
2 2
1 2
1 2
q qq x x
x x
x x x x
x x x x
q qx x
x x
x xq
x x
21
qx
x
The Atwood Machine consists of two masses M and m attached to the ends of a light, frictionless pulley. When the masses are released, the mass M is show to accelerate down with an acceleration:
M ma g
M m
Suppose the M and m are measured as M=100 1g and m=50 1 g. Find the uncertainty in a
2
2
1 1
2
M m M mag
M M m
mg
M m
2
2
1 1
2
M m M mag
m M m
Mg
M m
The Partial Derivatives are:
2 2
2 2
2 2
2 22 22
2 2 2 2
2
2 2
2
2 9.850 1 100 1
100 50
0.1
a aq M m
M m
mg MgM m
M m M m
gm M M m
M m
100 509.8
100 503.3
M ma g
M m
Therefore a=(3.3 0.1) m/s2
Uncertainty
Focal Length
pqf
p q
Determine the focal length plus uncertainty when p=100±2 cm and q=30±1 cm
2
2
2
1q p q pqf
p p q
q
p q
2
2
2
1p p q pqf
q p q
p
p q
Focal Length
2 2
2 24 4
2
2 24 4
2
30 2 100 1
100 30
0.6112
f ff p q
p q
q p p q
p q
100 30
100 3023.076
23.1
f
The focal length is (23.1±0.6) cm or (23±1) cm
Ugly Trig Problem
2
cos 4
xq
x y
Determine q and error is x=10±2, y=7±1, Ø=400±30
2
cos 4 2
cos 4
yq
x x y
2
2 cos 4
cos 4
xq
y x y
2
4 2 sin 4
cos 4
x yq
x y
=-0.732
=0.963
=9.813
Still the Ugly Trig Problem
2
2 20.732 2 0.963 1 9.813 3
180
3.3
1.8165
q
Therefore q=3.5±2
Trig should be in
radians
Max-Min Technique
If you do not have a calculus background, then you can use this technique to determine the uncertainties in a complicated equation.
1) Determine the actual value.2) Make the largest possible value.3) Make the smallest possible value.4) Average the difference between the actual and the largest
value and the actual and the smallest value.5) This average is the uncertainty.
Ugly Trig Problem again
2
cos
x yq
Determine q and error is x=10±2, y=7±1, Ø=400±30
2
cos
10 2 7
cos 40
24.8027
x yq
max
1
2
cos
2
cos
30.0812
2 8
43
x yq
min
2
cos
2
cos
20.034
8 6
37
x yq
max
min
24.8028 30.0812 5.2784
24.8028 20.034 4.7688
q q
q q
Ave=5.0236
Therefore: 25±5
Using Percent Errors
Two Simple Rules:
1. When Adding or Subtracting add the Absolute Errors like you would normally do, then convert to Percent (Relative) Error.
2. When Multiplying or Dividing add the Percentage Errors
Observed Value - True ValuPerc
e
Tru =
e Ve
ant Er
lueror 100
Percent Error Exmple
4.52 0.02 , 2.0 0.2 , 3.0 0.6w x y
2znd yF wxi
For wx: 0.02 0.20.004425 0.1 0.104425 10.44%
4.52 2.0
For y2:0.6 0.6
0.2 0.2 0.4 40%3.0 3.0
To find Error for z, we need to convert Percent Errors to Absolute
2
0.910.4 444%
40%
4.52 2.0
3.63.0
2 218 0.9 3.6 18 3.79.0 180.9 9.0 3.6 4
Therefore: