Compute the three-dimensional Chebyshev polynomial tensor product approximation on [1, 3] using 11 points in each dimension.

Compute the three-dimensional Chebyshev polynomial tensor product approximation on [1, 3] using 11 points in each dimension. Use 41 uniformly distributed points in each dimension to compute the L and L norms of the error. Then compute the complete polynomial approximation. How much accuracy do you lose? What is the relative computational cost of the two approximations?