I looked for the normal cone to a polyhedron, and I found this definition from https://sites.math.washington.edu/~rtr/papers/rtr169-VarAnalysis-RockWets.pdf: normal cone defintion

But this is confusing to me. it seems to me like the $\sum y_i a_i$ will yield a single value and not a vector. it was my understanding that the normal cone is a set of vectors.

Is my understanding wrong? Or am I misinterpreting what this book is saying for the Normal cone of a polyhedron?? thank.


Note that on the right side of your formula for $N_C(\bar{x})$, there are curly brackets indicating this is a set. It is a special set of linear combinations with all coefficients being nonnegative. It is a finitely generated convex cone.

PS: You have found one of the finest math books ever written: Rockafellar-Wets's Variational Analysis - Have fun with it!

  • $\begingroup$ I see, the $a_i$'s are actually vectors, NOT elements of a single vector. thank you @max_zorn. $\endgroup$ – jaja May 6 '18 at 6:55
  • $\begingroup$ Although, technically $a_i$ is a row vector. So, I imagine they should actually all be transposed in order for the result of $\sum y_i a_i$ to be a column vector. i.e. I think it should be $\sum y_i a_i^T$. do you think so too @max_zorn? $\endgroup$ – jaja May 6 '18 at 18:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.