Is it mathematically guaranteed that this changed algorithm converges to the same value as the previous algorithm? It seems somewhat non-trivial.
mm
@russt17 from what I understood, this algorithm does not converge to the same exact value as the original but it is "close enough".
BigFish
@russt17 It is discussed in the previous slide. This will not lead to the exactly same solution, but it will converge within the error threshold.
rokislt10
As mentioned in class, the previous algorithm is an approximation in the first place, so even if the two algorithms convered to the same value, it wouldn't mean much. This algorithm is kind of giving an approximation of an approximation, which as the previous two commenters have said, is good enough.
Is it mathematically guaranteed that this changed algorithm converges to the same value as the previous algorithm? It seems somewhat non-trivial.
@russt17 from what I understood, this algorithm does not converge to the same exact value as the original but it is "close enough".
@russt17 It is discussed in the previous slide. This will not lead to the exactly same solution, but it will converge within the error threshold.
As mentioned in class, the previous algorithm is an approximation in the first place, so even if the two algorithms convered to the same value, it wouldn't mean much. This algorithm is kind of giving an approximation of an approximation, which as the previous two commenters have said, is good enough.