Calculate numerical gradient

Sorry for flooding the forum, but I am currently discovering the newer versions of Gauss and some questions arise.

I would like to understand the difference between the commands gradp and gradMT. If I understand well, the only difference is that the latter allows structure input, while the former only a vector input of the parameters. Is that correct? Or are there any other differences, such as different computation methods, etc?

2 Answers



0



The main difference, and the reason it was created, is to allow the parameter vector input. It does also have a slightly different calculation for the distance it adds to the parameters. i.e. the h in f(x+h).

aptech

1,773


0



Thanks for the quick reply. I tried gradp and it seemed to work fine. However, I should compare the two for a function for which the analytical gradient is known, so I can compare the numerical approximations.

Your Answer

2 Answers

0

The main difference, and the reason it was created, is to allow the parameter vector input. It does also have a slightly different calculation for the distance it adds to the parameters. i.e. the h in f(x+h).

0

Thanks for the quick reply. I tried gradp and it seemed to work fine. However, I should compare the two for a function for which the analytical gradient is known, so I can compare the numerical approximations.


You must login to post answers.

Have a Specific Question?

Get a real answer from a real person

Need Support?

Get help from our friendly experts.