5 Comments
For me, it was the curve fitting toolbox. I frequently used their nonlinear least squares solver lsqnonlin to fit monkey steering data in grad school.
Every realease I write down 2-3 new features on a post-it note to remind me to use them. For R2024a my pics were,
That the defaul data-type is the double array, nowadays n-dimensional, and that we get the matrix algebra using the standard operators. My students now almost exclusively program in python - and the mess of list-of-lists and tuples makes the second and third steps in scientific programming that much more annoying - having to explicitly crate arrays.
I agree with goc3: the documentation is excellent. On top of that I'd say the presence of many different useful data types; various toolboxes (of which I'd like to single out the Symbolic Math Toolbox for special praise); and plotting capabilities.
The high-level language aspect of MATLAB.
I find it particularly helpful in creating and manipulating arrays, complicated calculations, statistics, parameter studies, etc.
The official documentation on MathWorks.com is well written and many of its pages include helpful example code.