You’re only as good as your weakest link adaptability of your strongest skill.

Nay loops, Yay MATLAB bsxfun/NumPy Broadcasting

Interesting Indulgences:

• 6D Permute : MATLAB, NumPy

• Kill recursions by tracing.

Exciting MATLAB:

• Vectorized computations on ragged pattern data (including cell arrays) with bsxfun's masking capability.
• Sum over blocks : [1], [2], [3], [4].
• Dump meshgrid/ndgrid for bsxfun: [1] [2] [3]
• Avoid anonymous functions, even with bsxfun : [1] runtime numbers.
• Index into diagonal elements efficiently with bsxfun: [1]
• Function-inline for insane speedups:[1].
• Avoid sub2ind with its raw version for a noticeable speedup! [1]
• Replace repmat with bsxfun : [1] [2] [3] [4]
• Even tic/toc need warming up before benchmarking
• Efficient euclidean distances calculation: Original source, Explanation
• Dump kron for bsxfun!
• Alternatives to nchoosek(1:N,2)
• Accumarray on 2D matrix

Exciting NumPy:

• Heavy Broadcasting based vectorization

• Faster alternative to np.any in np.einsum