pbuk
Science Advisor
Homework Helper
Gold Member
- 4,966
- 3,217
That does not appear to be the case.valenumr said:Right, I understood that.
valenumr said:But also in that case you can just pick h to be 2^-24 (or maybe -23?) For all x, for 32-bit floats.
- We are in Python, there aren't any 32-bit floats.
- A 32 bit float has ## \epsilon = 2^{-24}##, what is going to happen to your accuracy when say ## x = 0.9 ##?
- If you use a step of ## h = 2^{-24} ## at ## x = 2^{-25} ## the rocket explodes.
- I could go on but this is not a good use of anyone's time: if you have something to contribute then show us the math.