Transcript Functional Programming

Rahman Lavaee Mashhadi
Mohammad Shadravan
Conditional expressions
 LISP was the first language to contain a conditional
expression
 In Fortran and Pascal we have to drop from expression
level to statement level
 (cond (p1 e1) … (pn en))
 ( if p e1 e2) = (cond (p e1) (t e2))
example :( defun sg(x)
(cond((plusp x) 1)
((zerop x) 0)
((minusp x) -1)) )
Logical connectives
 Logical connectives are evaluated conditionally
 (or x y) = (if x t y)
 Their operands are evaluated sequentially
(or (eq (car L) ‘key) (null L) )
wrong
 Strict interpretation vs. sequential interpretation
Iteration
 Iteration is done by recursion
 Analogous to while-loop
(defun plus-red (a)
(if (null a) 0
(plus (car a) (plus-red (cdr a)) )) )
 There is no “bounds” , no “explicit indexing” , no
“control variable”
 It obeys the zero-one-infinity principle
Nested Loops
 Example :
Cartesian product
(defun all-pairs (M N)
(if (null M) nil
(append (distl (car M) N)
(all-pairs (cdr M ) N )) ))
(defun distl (x N)
(if (null N) nil
(cons (list x (car N))
(distl x (cdr N)) )) )
Hierarchical structures
 Are difficult to handle iteratively
example: equal function
 eq only handles atoms
 initial states
 If x and y are both atoms (equal x y) = (eq x y)
 If exactly one of x and y is atom (equal x y) = nil
(and (atom x) (atom y) (eq x y))
 use car and cdr to write equal recursively
Equivalency of recursion and
iteration
 it may be seemed that recursion is more powerful than
iteration
 in theory these are equivalent
 As we said iteration can be done by recursion
 by maintaining a stack of activation records we can
convert a recursive program to an iterative one.
Functional arguments and
abstraction
 Suppress details of loop control and recursion
example: applying a function to all elements of list
(defun mapcar (f x)
(if (null x)
nil
(cons (f (car x)) (mapcar f (cdr x)) )) )
(defun reduce (f a x)
(if (null x)
a
(f (car x) (reduce f a (cdr x) )) ) )
Functional arguments and
combination of programs
 Functional arguments simplify the combination of
already implemented programs
example : inner product of two lists
(defun ip (u v)
(reduce ‘plus 0 (mapcar2 ‘times u v)) )
Anonymous functions
 It is inconvenient to give a name to every function we
want to pass it only once
(defun consval (x) (cons val x))
 An obvious solution is just to pass the function’s body
(mapcar ‘(cons val x) L)
 It is ambiguous : we do not know which names are
parameters and which ones are global
 “lambda expressions” come in hand
(mapcar ‘(lambda (x) (cons val x)) L)
Functionals
 A functional is a function that has either(or both) of
the following:
 One or more functions as arguments
 A function as its result
 Examples:
 Mapcar, reduce ,…
 Converting a binary function to a unary one
Functional can replace lambda
expressions
 Specifying parameters of a function with lambda
expressions can be done by functionals
example:
(lambda (y) (f x y))
 define “bu” function to fix one parameter of “f”
(defun bu (f x) (function (lambda (y) (f x y)) ))
 (lambda (y) (f x y)) ~ (bu f x)
 (lambda (x y) (f y x)) ~ (rev f)
Combining functions
 Flexibility of combining functions can be increased by
using the uniform style.
 Take a function , and return a function
((map ‘add1) L) = (mapcar ‘add1 L)
 This form is called ‘combining forms’
 Combining existing programs to acomplish new tasks
example: (set ‘vec-dbl (map (bu ‘times 2)) )
(vec-dbl ‘(1 4 3 2))
(2 8 6 4)
Backus style for functional
programming
 alternate functional style of programming
 Based on ‘combining form’
 Programming at a higher level of abstraction
 Simple algebraic manipulation
combining, reversing, mapping functions
 Function-level vs. object-level(manipulate function
rather objects)
Backus notation
Name
Lisp
Backus
Application
Mapping
Reduction
Composition
Binding
Constant
Lists
Built-in-functions
Selectors
(f x)
(map f)
(red f a)
(comp f g)
(bu f k)
(const k)
(a b c d)
plus, times, …
cdr, car, cadr, …
f:x
αf
/f
fog
(bu f k)
k©
(a,b,c,d)
+,×,…
tail, 1 , 2
Variable-free programming
 Variables (i.e. formal parameters and algol scope rules)
complicate manipulating programs
 One of the goals of this notation was to eliminate
variables
 Elimination of lambda expressions by using bu and rev
functionals
Example: inner product function
Trans:<<…ui…> , <…vi…>> = <….<ui,vi>…>
(α ×): (trans: <u,v>)=<u1v1,…,unvn>
Ip: <u,v> = (/+): ((α ×): (trans: <u,v>))
 Variable elimination
Ip= (/+)o (α ×)o trans
Name structures
 Value bindings are established in two ways:
 Property lists



Established by pseudo functions such as ‘set’ and ‘defun’
example:
(set ‘text ‘(to be or not to be))
They are global
low level execution with property lists
example:
(putprop ‘add1 ‘(lambda (x) (+ x 1)) ‘expr)
 Actual formal correspondence
 Like argument passing in other languages
example :
(defun times (x y) (* x y))
(times 3 FACT)
the formal x will be bound to atom 3 and the formal y will be bound
to the value of actual FACT
Temporary binding




Binding names in a local context like in algol’s blocks
is done by actual
Prevents duplicate calculation
Obeys abstraction principle
(defun roots (a b c)
(list
(/ (+ (- b) (sqrt (- (expt b 2) (* 4 a c)) ))
(* 2 a))
(/ (+ (- b) (sqrt (- (expt b 2) (* 4 a c)) ))
(* 2 a)) ))
solution
(defun roots-aux (d)
(list
(/ (+ (- b) d) (* 2 a))
(/ (+ (- b) d) (* 2 a)) ))
(defun roots (a b c)
roots-aux (sqrt (- ( expt b 2)
(times 4 a c)) )) )
Let function
 Function calling with let
(let ((n1 e1) … (nm em)) E)
 Abbreviation for a function definition and application
(defun QQQQQ (n1 … nm) E)
(QQQQQ e1 …. em)
 Allows a number of local names to be bound to values
at one time
 The second argument is evaluated in the resulting
environment and is returned as the value of let
 Like alogol block , Ada declare block
Dynamic scoping
 review
 Dynamic scoping: functions are called in the
environment of its caller
 Static scoping: functions are called in the environment
of its definition
 LISP uses dynamic scoping
example : roots-aux had access to the names ‘a’ and ‘b’
 The ‘let’ function needs dynamic scoping
Dynamic scoping complicates
functional arguments

(defun twice (func val) (func (func val)) )
(twice ‘add1 5) returns 7
(twice ‘(lambda (x) (times 2 x)) 3) returns 12


func
val
DL
3
x
DL
3
a
(times 2 x)
Functional argument problem
(set ‘val 2)
2
(twice ‘(lambda (x) (times val x)) 3)
 What is the return value of this expression?
we expect it to be 12
but it will return 27!
why?
Cause :Collision of variables
val
2
func
val
DL
3
x
3
DL
(times val x)
What’s the solution?
 What should we do?
 use different names
In a large program it would be impractical to ensure
that all of the names are distinct
 change the scoping rule
‘function’ primitive
 Use ‘function’ primitive to bind the lambda expression
to its environment of definition
(twice (function (lambda (x) (times val x)) ) 3 )
this will return 12
val
2
func
val
x
3
3
(times val x)
Syntactic structures
 Convenience for programmers is not the main
intention of the S-Expression
 ‘cond’ syntax was improved for convenience of
programmers
(cond (p1 e1) (p2 e2) … (pn en))
(cond p1 e1 p2 e2 … pn en)
 ‘define’ function was used for global binding
(define ((n1 e1) … (nk ek)) )
Syntactic structures
 (define ( (add1 (lambda (x) (+ x 1)) )) )
 Lots of idiotic single parentheses
 ‘defun’ was declared
(defun (add1 x) (+ x 1))
 (quote val) was improved to ‘val
 (set (qutoe val) 2) was improved (setq val 2)
List representation of programs
 Facilitates program manipulation’
 Balances reduction of readability
 Simple to write LISP programs that generate other
LISP programs
 Separating parts of a function
 (car A) is the name of the function A
 (cdr A) is the arguments of the function A
 Obeys Elegance Principle : writing LISP interpreter in
LISP