Work on error propagation section in appendices #3
Description
It would be very nice to improve the section on error propagation from the point of view of theoretical computer science.
To date, I was not able to find a nice place where to learn error propagation in algorithms.
There are many places on the internet that discuss this, but most of them are taking the approach of a physicists that needs to run an experiment (https://nvlpubs.nist.gov/nistpubs/jres/70C/jresv70Cn4p263_A1b.pdf ). This is not really helpful for CS people, as the error propagation in CS is different.
We have some notes to share (here in the repository, commented in the code) as we are half-way through this part.
- How to go from relative error to absolute error and vice versa
- Propagation of error in function of one variable (and their proof) for relative and absolute cases. For example (and this is already commented in the
- Propagation of error in function of multiple variable
- The general case: the mean value theorem. ( https://www.onemathematicalcat.org/CalculusBook/MainBook/4_9MVT.pdf or any book in calcolus, https://homepage.divms.uiowa.edu/~atkinson/ftp/ENA_Materials/Overheads/sec_2-3.pdf, and much more..)
If you decide to start this issue and you want some clarification or help, just contact me (@Scinawa ) on the discord of the unitary.fund, the slack channel of QOSF, @Scinawa on twitter, or write me at ale@nus.edu.sg.