# What is the name of the process I'm describing here ... Mr Magoo Brute Force Descent?

Imagine you have:

• a function that works on some static data
• the function has one parameter p with a known bounding range [0, 100]
• you want to minimise the output of the function.

You set up run through of the function on the data, with lets say 10 iterations, p is changed by 100/10=10 each run.

You find a minimum at some p

You now run through the range [p-5, p+5], but now p is changed by 1 each run...

This process is recursive...

Therefore, rinse and repeat until the resolution of the parameters and/or the minimum reaches the resolution of the system/original data.

Voila, you have at least a local minimum - but global is not guaranteed...

What is this process called ?

Myopic Brute Force Descent ?

p.s. I realise it's certainly not a good process, possibly the best name for it is simply 'bad' :)

• Is the observed function noisy or deterministic? (i.e. if I give it the same input, do I always get exactly the same output?) Commented Sep 18, 2015 at 9:11
• Hi Glen, output is exactly the same ... I hoped saying it was 'static' might account for that, not sure of the terminology though :) Commented Sep 18, 2015 at 9:30
• @Glen_b it's my experience that questions like this that involve ad-hoc approaches to problems never get any attention at math (as in tumbleweed badge territory). This forum also deals with machine learning, which is a bit more ad-hoc than purely stats, and also includes algorithms like gradient descent. I thought it the best fit, which forum do you think it should fit into ? Commented Sep 18, 2015 at 10:47
• ML certainly uses gradient descent (as does stats), but I don't know that gradient descent is ML per se (it's not really that empty a term, is it?). Both use matrix multiplication as well, but that doesn't make a matrix-multiplication question on topic either. Commented Sep 18, 2015 at 10:55
• I've only learned gradient descent in a ML course, that's all - it seems like a very multi-disciplinary field, lot's of folk learning stats and not realising it :D ... Anyhoo, if not here, where ? Commented Sep 18, 2015 at 11:01