One of the common tasks in analyzing data sets is to perform error
propagation.
As a way to de-emphasize the "grunt work" of error analysis for the students,
I have written a program to do error calculations, weighted averages, tails of gaussians, and similar analysis chores.
This program is freely available for Windows machines under the GNU General Public License. If you are using
this program, please e-mail me so I can let you know about future upgrades.
The upper line on the display is for the value and the lower line is for the error.
You can toggle between these with the Tab key or by clicking with a mouse
If no error is entered, it assumes that the value has no uncertainty
When this was written, for simplicity the program did not include an operation stack or other operator precedence rules.
Operations like multiplication, division, etc.
are entered as a value (with uncertainty if appropriate), operator, value (with uncertainty) then equal sign; operations such as trig functions or exponentials operate
on the value in the display. This means that you cannot do something like 2 + 5 * 3 = and have it give 17.
Instead,
you should type 5 × 3 = + 2 =. If each of these values has an uncertainty of 0.1, then you
would enter 5 in the value box and 0.1 in the error box, then the × key, then
3 in value and 0.1 in uncertainty followed by =. Then type + and 2 with 0.1 uncertainty, and finally =.
Similarly to calculate the area of a circle
that has a radius of 2.0±0.1, you can't type πr2 directly.
Instead enter 2 into the value box 0.1 into the uncertainty
box, then type the X^2 key then × pi =.
In the history section on the right, you can click on any value and it will be copied to the display. The screen shot on this web page shows a couple of sample calculations.
To enter a value like Avogadros number, type 6.02 then the EE+ key and 23,
and the charge on the electron would be 1.602 then the EE- key and 19.
Results that have uncertainties greater than 25% will be colored red instead of green.
This is a warning to be careful about the validity of the conventional approximation of small errors. For uncertainties
larger than about 25%, it is very possible that a different technique (such as Monte Carlo approximation) would be
more appropriate for estimating the uncertainties.
The Advanced Options checkbox enables calculations such as tails of Gaussian Distributions,
and Weighted Averages. This also displays a column showing percent errors for all numbers.
