Re: [R] optim or nlminb for minimization, which to believe?

Hi Harold, Hans is right. You can see this if you use BB::spg require(BB) opt2 - spg(startVal, fn) # this is fine opt3 - spg(startVal, fn,gradient) # this is not fine! opt3 - spg(startVal, fn, gradient) Gradient check details: max. relative difference in gradients= 0.001697913

[R] optim or nlminb for minimization, which to believe?

I have constructed the function mml2 (below) based on the likelihood function described in the minimal latex I have pasted below for anyone who wants to look at it. This function finds parameter estimates for a basic Rasch (IRT) model. Using the function without the gradient, using either

Re: [R] optim or nlminb for minimization, which to believe?

Your function named 'gradient' is not the correct gradient. Take as an example the following point x0, very near to the true minimum, x0 - c(-0.2517964, 0.4898680, -0.2517962, 0.4898681, 0.7500995) then you get gradient(x0) [1] -0.0372110470 0.0001816991 -0.0372102284