Solvers
Rapid notes on solvers - algorithms for optimising a black-box
function. This is probably all in textbooks, but i don't have a
textbook. I also don't know what these are formally called -
optimization algorithms, or iterative somethingorothers.
- Newton's method, Newton-Raphson method, etc;
need to know the derivative, or an approximation to it, of the function
wrt each parameter, i think, although since there are applets to
do it, unless they're doing symbolic differentiation (which is not
entirely trivial), that's not true. Okay, yes, you can use finite
differences to approximate the derivatives - it's obvious that you can,
less obvious to my simple brain that it will work, but apparently, it's
kosher.
- BFGS. DFT.
- Impressive-sounding
names.
- Boom!
- Baby!
- Check out NumPy and Octave's python interface
- Numerical, sorry,Simple Recipes in Python
- ScientificPython