For , the gradient is , andAway from the origin, both and lie in the one-dimensional subspace perpendicular to , so they are parallel. More precisely,for some scalar , including . If , this can also be inverted to express as a multiple of . At a stationary point of with , that literal inverse wording is not valid; the geometric assertion is parallelism of the nonzero gradient directions.
Articles by others on the same topic
There are currently no matching articles.