- New `allfinite`, `anynan` and `anymissing` functions to examine the elements of an input array
- The `quantile`, `prctile` and `iqr` functions are now part of core MATLAB (they were previously in the Statistics and Machine Learning Toolbox
- The `rms` (root mean square) function is now part of core MATLAB (it was previously in the Signal Processing toolbox)
- The `std` and `var` functions now optionally return the mean as second output argument
- New `writelines` function for writing plain text to a file
- New `pagemldivide`, `pagemrdivide`, and `pageinv` functions: Solve linear equations and calculate matrix inverses using pages of N-D arrays
- New `tensorprod` function: Calculate tensor products between two arrays
- The `round` function has a new `TieBreaker` name-value argument to specify how to break ties.
-`null` and `orth` functions have a new option to specify tolerance to treat singular values below a threshold as zero
- The `norm` function can now calculate Frobenius norm on N-D arrays
- The `equilibrate` function now has an option with values of `"vector"` or `"matrix"` to specify whether the output arguments are returned as vectors or matrices
- The `rand`, `randi`, and `randn` functions now support complex input and a `RandStream` object for the `"like"` syntax
- The `eps`, `flintmax`, `intmax`, `intmin`, `realmax` and `realmin` functions now accept the `"like"` syntax to return scalars based on a prototype object
-`qr` and `gsvd` have a new `"econ"` option for economy-size decompositions
- The syntax `R = qr(A)` now always returns `R` as an upper-triangular matrix, regardless of whether `A` is full or sparse (behaviour change)
# MATLAB 9.11 (R2021b)
# MATLAB 9.11 (R2021b)
- New `parfeval` syntax for parallelizing calls to a function
- New `parfeval` syntax for parallelizing calls to a function
