CELP (CELl Processing) is a functional programming package for Matlab.
It is often better to use FP instead of loops:
- it's more compact;
- it's more clear.
Matlab has all prerequisites to implement a Lisp-like FP:
- anonymous functions, @'s <-> lambdas;
- FEVAL <-> FUNCALL, EVAL, APPLY;
- cell arrays <-> lists.
Matlab has also several peculiarities:
- cell arrays may have arbitrary dimensions (not only 1d);
- functions may have several outputs
(btw you can use builtin DEAL to implement it in @'s);
- vectors can be converted to cells and back using
NUM2CELL and CELL2MAT.
Currently the package includes
CPAPPLY Analogue of Lisp's APPLY
CPBIND Substitution of argument
CPEVAL Evaluate multioutput function and return outputs in one cell array
CPFILTER Analogue of Lisp's FILTER
CPIF Conditional composition of functions
CPMAP Analogue of Lisp's MAPCAR
CPREDUCE Analogue of Lisp's REDUCE
(use 'help *' for more info and examples).
CPFILTER, CPMAP, CPREDUCE are equivalents of Lisp's routines. The latter
two are implemented in MEXs for speed.
CPIF and CPBIND are used for construction of new functions from
defined. They are implemented with the help of nested functions.
CPAPPLY, CPEVAL deal with multioutput functions. One general limitation of
M-language is that when you write
only one output of g(x) is substituted into f. Thus you need to use
[y,z] = g(x);
to substitute all outputs. Using CELP's functions you may code it as
The main disadvantage of CELP is speed - evaluation of anonymous functions
can be quite slow. Thus I do not recommend to use CELP with @'s in
intensive calculations (probably Nathan Thern's TMP_HANDLE may be of help
here - search for it on FEX).