I'm confused about one point in the powell method with discarding the direction of largest decrease (NR in C++, 2nd ed).

The statement is made "The basic idea...is still to take PN - P0 as the new direction..."

Later on, an extrapolated point "somewhat further along the proposed new direction" is defined as 2*P_N - P_0. But this is NOT in the same direction as PN - P0. Wouldn't a better choice be 2*(PN - P0 )? Or am I missing something.:confused:

Thanks

Hi,

The sentence you are worried about describes one of the exceptions to the basic idea, as refered to at the beginning of the paragraph.

Saul Teukolsky

Oh - for some reason I was thinking that the extrapolated point was used as the new direction (if the criteria were met). Now a closer look at the code shows that that is not the case.

Thanks for your help