I am trying to write a procedure stream-weighted-tuples that takes a weight procedure and any number of streams, to generate a stream of tuples.
For instance,
(stream-weighted-tuples
(lambda (t) (+ (car t) (cadr t) (caddr t))
integers
integers
integers)
should generate a stream of (1 1 1) (1 1 2) (1 2 1) (2 1 1) (1 1 3) (1 2 2) (1 3 1) (2 1 2) (2 2 1) (3 1 1) (1 1 4) (1 2 3) ....
I was inspired by exercise 3.70 in SICP, which is about writing a procedure weighted-pairs that takes two streams and a weight procedure to generate a stream of pairs in the order according to weight.
So basically, this is a generalization of the weighted-pairs procedure to take more than two streams.
I wrote the following version:
(define (stream-weighted-tuples weight . streams)
(cond ((null? streams)
(error "No streams given -- STREM-WEIGHTED-TUPLES"))
((null? (cdr streams))
(stream-map list (car streams)))
(else
(let ((s (car streams))
(rest (cdr streams)))
(if (stream-null? s)
the-empty-stream ; {} x S = {}
(stream-merge-weighted
weight
(stream-map (lambda (tuple)
(cons (stream-car s) tuple))
(apply stream-weighted-tuples
(lambda (tuple) ; partial weight
(weight (cons (stream-car s)
tuple)))
rest))
(apply stream-weighted-tuples
weight
(stream-cdr s)
rest)))))))
(which apparently doesn’t work).
The idea was to merge 1. the stream resulting from consing the first element of the first stream with each element of the (recursively constructed) stream that consists of the tuples from the rest of the given streams, and 2. the stream that consists of the tuples of the stream-cdr of the first stream and the rest of the given streams.
This doesn’t work because generating stream 2 won’t halt in the case of an infinite stream (due to the same reason pairs procedure from exercise 3.68 doesn’t work).
This is stream-merge-weighted I implemented:
(define (stream-merge-weighted weight s1 s2)
(cond ((stream-null? s1) s2)
((stream-null? s2) s1)
(else
(let* ((s1car (stream-car s1))
(s2car (stream-car s2))
(s1car-weight (weight s1car))
(s2car-weight (weight s2car)))
(cond ((<= s1car-weight s2car-weight)
(cons-stream
s1car
(stream-merge-weighted weight
(stream-cdr s1)
s2)))
((> s1car-weight s2car-weight)
(cons-stream
s2car
(stream-merge-weighted weight
s1
(stream-cdr s2)))))))))
Is there a way to recursively construct this problem?
EDIT: What I meant by “tuple” in this question is actually a Scheme list that semantically represents a tuple.
For reference, I leave my implementation of stream primitives (which is identical to the ones in SICP):
(define-syntax cons-stream
(syntax-rules ()
((_ a b)
(cons a (delay b)))))
(define (stream-car stream)
(car stream))
(define (stream-cdr stream)
(force (cdr stream)))
(define (stream-null? stream)
(null? stream))
(define the-empty-stream '())
(define (stream-map proc s)
(if (stream-null? s)
the-empty-stream
(cons-stream
(proc (stream-car s))
(stream-map proc (stream-cdr s)))))
I managed to solve it using a recursion structure similar to that of the pairs procedure shown in SICP.
The following stream-weighted-tuples works with recursively constructed stream of tuples of the cdr of streams (rest-tuples), and then uses the same recursion strategy to “pair” (unlike the original pair procedure, the other stream is already a stream of tuples) it with the car of streams (combine-with-rest).
(define (stream-weighted-tuples weight . streams)
(if (null? streams)
(cons-stream '() the-empty-stream)
(let* ((stream (car streams))
(rest (cdr streams))
(rest-tuples
(apply stream-weighted-tuples
(lambda (tuple) ; partial weight
(weight (cons (stream-car stream) tuple)))
rest)))
(let combine-with-rest ((stream stream)
(rest-tuples rest-tuples))
(if (or (stream-null? stream)
(stream-null? rest-tuples))
the-empty-stream ; {} x S = {}
(let ((s-car (stream-car stream))
(s-cdr (stream-cdr stream))
(r-car (stream-car rest-tuples))
(r-cdr (stream-cdr rest-tuples)))
(cons-stream (cons s-car r-car)
(stream-merge-weighted
weight
(stream-map (lambda (t) (cons s-car t))
r-cdr)
(stream-merge-weighted
weight
(stream-map (lambda (x) (cons x r-car))
s-cdr)
(combine-with-rest s-cdr r-cdr))))))))))
Note, that the weight procedure requires the cdr of tuples to be ordered independent to the cars, i.e., W(a, b, c) >= W(x, y, z) only if W(b, c) >= W(y, z).
Related
The problem is to:
Write a function (encode L) that takes a list of atoms L and run-length encodes the list such that the output is a list of pairs of the form (value length) where the first element is a value and the second is the number of times that value occurs in the list being encoded.
For example:
(encode '(1 1 2 4 4 8 8 8)) ---> '((1 2) (2 1) (4 2) (8 3))
This is the code I have so far:
(define (encode lst)
(cond
((null? lst) '())
(else ((append (list (car lst) (count lst 1))
(encode (cdr lst)))))))
(define (count lst n)
(cond
((null? lst) n)
((equal? (car lst) (car (cdr lst)))
(count (cdr lst) (+ n 1)))
(else (n)))))
So I know this won't work because I can't really think of a way to count the number of a specific atom in a list effectively as I would iterate down the list. Also, Saving the previous (value length) pair before moving on to counting the next unique atom in the list. Basically, my main problem is coming up with a way to keep a count of the amount of atoms I see in the list to create my (value length) pairs.
You need a helper function that has the count as additional argument. You check the first two elements against each other and recurse by increasing the count on the rest if it's a match or by consing a match and resetting count to 1 in the recursive call.
Here is a sketch where you need to implement the <??> parts:
(define (encode lst)
(define (helper lst count)
(cond ((null? lst) <??>)
((null? (cdr lst)) <??>))
((equal? (car lst) (cadr lst)) <??>)
(else (helper <??> <??>))))
(helper lst 1))
;; tests
(encode '()) ; ==> ()
(encode '(1)) ; ==> ((1 1))
(encode '(1 1)) ; ==> ((1 2))
(encode '(1 2 2 3 3 3 3)) ; ==> ((1 1) (2 2) (3 4))
Using a named let expression
This technique of using a recursive helper procedure with state variables is so common in Scheme that there's a special let form which allows you to express the pattern a bit nicer
(define (encode lst)
(let helper ((lst lst) (count 1))
(cond ((null? lst) <??>)
((null? (cdr lst)) <??>))
((equal? (car lst) (cadr lst)) <??>)
(else (helper <??> <??>)))))
Comments on the code in your question: It has excess parentheses..
((append ....)) means call (append ....) then call that result as if it is a function. Since append makes lists that will fail miserably like ERROR: application: expected a function, got a list.
(n) means call n as a function.. Remember + is just a variable, like n. No difference between function and other values in Scheme and when you put an expression like (if (< v 3) + -) it needs to evaluate to a function if you wrap it with parentheses to call it ((if (< v 3) + -) 5 3); ==> 8 or 2
I'm using the beginning language with list abbreviations for DrRacket and want to make a powerset recursively but cannot figure out how to do it. I currently have this much
(define
(powerset aL)
(cond
[(empty? aL) (list)]
any help would be good.
What's in a powerset? A set's subsets!
An empty set is any set's subset,
so powerset of empty set's not empty.
Its (only) element it is an empty set:
(define
(powerset aL)
(cond
[(empty? aL) (list empty)]
[else
As for non-empty sets, there is a choice,
for each set's element, whether to be
or not to be included in subset
which is a member of a powerset.
We thus include both choices when combining
first element with smaller powerset,
that, which we get recursively applying
the same procedure to the rest of input:
(combine (first aL)
(powerset (rest aL)))]))
(define
(combine a r) ; `r` for Recursive Result
(cond
[(empty? r) empty] ; nothing to combine `a` with
[else
(cons (cons a (first r)) ; Both add `a` and
(cons (first r) ; don't add, to first subset in `r`
(combine ; and do the same
a ; with
(rest r))))])) ; the rest of `r`
"There are no answers, only choices". Rather,
the choices made, are what the answer's made of.
In Racket,
#lang racket
(define (power-set xs)
(cond
[(empty? xs) (list empty)] ; the empty set has only empty as subset
[(cons? xs) (define x (first xs)) ; a constructed list has a first element
(define ys (rest xs)) ; and a list of the remaining elements
;; There are two types of subsets of xs, thouse that
;; contain x and those without x.
(define with-out-x ; the power sets without x
(power-set ys))
(define with-x ; to get the power sets with x we
(cons-all x with-out-x)) ; we add x to the power sets without x
(append with-out-x with-x)])) ; Now both kind of subsets are returned.
(define (cons-all x xss)
; xss is a list of lists
; cons x onto all the lists in xss
(cond
[(empty? xss) empty]
[(cons? xss) (cons (cons x (first xss)) ; cons x to the first sublist
(cons-all x (rest xss)))])) ; and to the rest of the sublists
To test:
(power-set '(a b c))
Here's yet another implementation, after a couple of tests it appears to be faster than Chris' answer for larger lists. It was tested using standard Racket:
(define (powerset aL)
(if (empty? aL)
'(())
(let ((rst (powerset (rest aL))))
(append (map (lambda (x) (cons (first aL) x))
rst)
rst))))
Here's my implementation of power set (though I only tested it using standard Racket language, not Beginning Student):
(define (powerset lst)
(if (null? lst)
'(())
(append-map (lambda (x)
(list x (cons (car lst) x)))
(powerset (cdr lst)))))
(Thanks to samth for reminding me that flatmap is called append-map in Racket!)
You can just use side effect:
(define res '())
(define
(pow raw leaf)
(cond
[(empty? raw) (set! res (cons leaf res))
res]
[else (pow (cdr raw) leaf)
(pow (cdr raw) (cons (car raw) leaf))]))
(pow '(1 2 3) '())
"define a procedure 'reduce-per-key' which a procedure reducef and a list of associations in which each key is paired with a list. The output is a list of the same structure except that each key is now associated with the result of applying reducef to its associated list"
I've already written 'map-per-key' and 'group-by-key' :
(define (map-per-key mapf lls)
(cond
[(null? lls) '()]
[else (append (mapf (car lls))(map-per-key mapf (cdr lls)))]))
(define (addval kv lls)
(cond
[(null? lls) (list (list (car kv)(cdr kv)))]
[(eq? (caar lls) (car kv))
(cons (list (car kv) (cons (cadr kv) (cadar lls)))(cdr lls))]
[else (cons (car lls)(addval kv (cdr lls)))]))
(define (group-by-key lls)
(cond
[(null? lls) '()]
[else (addval (car lls) (group-by-key (cdr lls)))]))
how would I write the next step, 'reduce-per-key' ? I'm also having trouble determining if it calls for two arguments or three.
so far, I've come up with:
(define (reduce-per-key reducef lls)
(let loop ((val (car lls))
(lls (cdr lls)))
(if (null? lls) val
(loop (reducef val (car lls)) (cdr lls)))))
however, with a test case such as:
(reduce-per-key
(lambda (kv) (list (car kv) (length (cadr kv))))
(group-by-key
(map-per-key (lambda (kv) (list kv kv kv)) xs)))
I receive an incorrect argument count, but when I try to write it with three arguments, I also receive this error. Anyone know what I'm doing wrong?
Your solution is a lot more complicated than it needs to be, and has several errors. In fact, the correct answer is simple enough to make unnecessary the definition of new helper procedures. Try working on this skeleton of a solution, just fill-in the blanks:
(define (reduce-per-key reducef lls)
(if (null? lls) ; If the association list is empty, we're done
<???> ; and we can return the empty list.
(cons (cons <???> ; Otherwise, build a new association with the same key
<???>) ; and the result of mapping `reducef` on the key's value
(reduce-per-key <???> <???>)))) ; pass reducef, advance the recursion
Remember that there's a built-in procedure for mapping a function over a list. Test it like this:
(reduce-per-key (lambda (x) (* x x))
'((x 1 2) (y 3) (z 4 5 6)))
> '((x 1 4) (y 9) (z 16 25 36))
Notice that each association is composed of a key (the car part) and a list as its value (the cdr part). For example:
(define an-association '(x 3 6 9))
(car an-association)
> 'x ; the key
(cdr an-association)
> '(3 6 9) ; the value, it's a list
As a final thought, the name reduce-per-key is a bit misleading, map-per-key would be a lot more appropriate as this procedure can be easily expressed using map ... but that's left as an exercise for the reader.
UPDATE:
Now that you've found a solution, I can suggest a more concise alternative using map:
(define (reduce-per-key reducef lls)
(map (lambda (e) (cons (car e) (map reducef (cdr e))))
lls))
Given a recursive function in scheme how do I change that function to tail recursive, and then how would I implement it using streams? Are there patterns and rules that you follow when changing any function in this way?
Take this function as an example which creates a list of numbers from 2-m (this is not tail recursive?)
Code:
(define listupto
(lambda (m)
(if (= m 2)
'(2)
(append (listupto (- m 1)) (list m)))))
I'll start off by explaining your example. It is definitely not tail recursive. Think of how this function executes. Each time you append you must first go back and make the recursive call until you hit the base case, and then you pull your way back up.
This is what a trace of you function would look like:
(listupto 4)
| (append (listupto(3)) '4)
|| (append (append (listupto(2)) '(3)) '(4))
||| (append (append '(2) '(3)) '(4))
|| (append '(2 3) '(4))
| '(2 3 4)
'(2 3 4)
Notice the V-pattern you see pulling in and then out of the recursive calls. The goal of tail recursion is to build all of the calls together, and only make one execution. What you need to do is pass an accumulator along with your function, this way you can only make one append when your function reaches the base case.
Here is the tail recursive version of your function:
(define listupto-tail
(lambda (m)
(listupto m '())))
# Now with the new accumulator parameter!
(define listupto
(lambda (m accu)
(if (= m 2)
(append '(2) accu)
(listupto (- m 1) (append (list m) accu)))))
If we see this trace, it will look like this:
(listupto 4)
| (listupto (3) '(4)) # m appended with the accu, which is the empty list currently
|| (listupto (2) '(3 4)) # m appended with accu, which is now a list with 4
||| (append '(2) '(3 4))
'(2 3 4)
Notice how the pattern is different, and we don't have to traverse back through the recursive calls. This saves us pointless executions. Tail recursion can be a difficult concept to grasp I suggest taking a look here. Chapter 5 has some helpful sections in it.
Generally to switch to a tail recursive form you transform the code so that it takes an accumulator parameter which builds the result up and is used as the final return value. This is generally a helper function which your main function delegates too.
Something of the form:
(define listupto
(lambda (m)
(listupto-helper m '())))
(define listupto-helper
(lambda (m l)
(if (= m 2)
(append '(2) l)
(listupto-helper (- m 1) (append (list m) l)))))
As the comments point out, the helper function can be replaced with a named let which is apparently (haven't done much/enough Scheme!) more idiomatic (and as the comments suggest cons is much better than creating a list and appending.
(define listupto
(lambda (n)
(let loop ((m n) (l '()))
(if (= m 2)
(append '(2) l)
(loop (- m 1) (cons m l))))))
You also ask about streams. You can find a SICP styled streams used e.g. here or here which have a from-By stream builder defined:
;;;; Stream Implementation
(define (head s) (car s))
(define (tail s) ((cdr s)))
(define-syntax s-cons
(syntax-rules ()
((s-cons h t) (cons h (lambda () t)))))
;;;; Stream Utility Functions
(define (from-By x s)
(s-cons x (from-By (+ x s) s)))
Such streams creation relies on macros, and they must be accessed by special means:
(define (take n s)
(cond ; avoid needless tail forcing for n == 1 !
((= n 1) (list (head s))) ; head is already forced
((> n 1) (cons (head s) (take (- n 1) (tail s))))
(else '())))
(define (drop n s)
(cond
((> n 0) (drop (- n 1) (tail s)))
(else s)))
But they aren't persistent, i.e. take and drop recalculate them on each access. One way to make streams persistent is to have a tailing closure surgically altering the last cons cell on access:
(1 . <closure>)
(1 . (2 . <closure>))
....
like this:
(define (make-stream next this state)
(let ((tcell (list (this state)))) ; tail sentinel cons cell
(letrec ((g (lambda ()
(set! state (next state))
(set-cdr! tcell (cons (this state) g))
(set! tcell (cdr tcell))
tcell)))
(set-cdr! tcell g)
tcell)))
(define (head s) (car s))
(define (tail s)
(if (or (pair? (cdr s))
(null? (cdr s)))
(cdr s)
((cdr s))))
We can now use it like this
(define a (make-stream (lambda (i) (+ i 1)) (lambda (i) i) 1))
;Value: a
a
;Value 13: (1 . #[compound-procedure 14])
(take 3 a)
;Value 15: (1 2 3)
a
;Value 13: (1 2 3 . #[compound-procedure 14])
(define b (drop 4 a))
;Value: b
b
;Value 16: (5 . #[compound-procedure 14])
a
;Value 13: (1 2 3 4 5 . #[compound-procedure 14])
(take 4 a)
;Value 17: (1 2 3 4)
a
;Value 13: (1 2 3 4 5 . #[compound-procedure 14])
Now, what does (make-stream (lambda (i) (list (cadr i) (+ (car i) (cadr i)))) car (list 0 1)) define?
update: in Daniel Friedman's 1994 slides "The Joys of Scheme, Cont'd" we find simpler implementation of these "memoized streams" (as they are called there), making the tail function itself store the forced stream in the tail sentinel, as
(define (tail s)
(if (or (pair? (cdr s))
(null? (cdr s)))
(cdr s)
(let ((n ((cdr s))))
(set-cdr! s n)
(cdr s))))
;; can be used as e.g. (https://ideone.com/v6pzDt)
(define fibs
(let next-fib ((a 0) (b 1))
(s-cons a (next-fib b (+ a b)))))
Here's a tail recursive form -
(define (listupto n)
(let run
((m 0)
(return identity))
(if (> m n)
(return null)
(run (add1 m)
(lambda (r) (return (cons m r)))))))
(listupto 9)
; '(0 1 2 3 4 5 6 7 8 9)
And here it is as a stream -
(define (listupto n)
(let run
((m 0))
(if (> m n)
empty-stream
(stream-cons m
(run (add1 m))))))
(stream->list (listupto 9))
; '(0 1 2 3 4 5 6 7 8 9)
I'm new to Scheme (via Racket) and (to a lesser extent) functional programming, and could use some advise on the pros and cons of accumulation via variables vs recursion. For the purposes of this example, I'm trying to calculate a moving average. So, for a list '(1 2 3 4 5), the 3 period moving average would be '(1 2 2 3 4). The idea is that any numbers before the period are not yet part of the calculation, and once we reach the period length in the set, we start averaging the subset of the list according the chosen period.
So, my first attempt looked something like this:
(define (avg lst)
(cond
[(null? lst) '()]
[(/ (apply + lst) (length lst))]))
(define (make-averager period)
(let ([prev '()])
(lambda (i)
(set! prev (cons i prev))
(cond
[(< (length prev) period) i]
[else (avg (take prev period))]))))
(map (make-averager 3) '(1 2 3 4 5))
> '(1 2 2 3 4)
This works. And I like the use of map. It seems composible and open to refactoring. I could see in the future having cousins like:
(map (make-bollinger 5) '(1 2 3 4 5))
(map (make-std-deviation 2) '(1 2 3 4 5))
etc.
But, it's not in the spirit of Scheme (right?) because I'm accumulating with side effects. So I rewrote it to look like this:
(define (moving-average l period)
(let loop ([l l] [acc '()])
(if (null? l)
l
(let* ([acc (cons (car l) acc)]
[next
(cond
[(< (length acc) period) (car acc)]
[else (avg (take acc period))])])
(cons next (loop (cdr l) acc))))))
(moving-average '(1 2 3 4 5) 3)
> '(1 2 2 3 4)
Now, this version is more difficult to grok at first glance. So I have a couple questions:
Is there a more elegant way to express the recursive version using some of the built in iteration constructs of racket (like for/fold)? Is it even tail recursive as written?
Is there any way to write the first version without the use of an accumulator variable?
Is this type of problem part of a larger pattern for which there are accepted best practices, especially in Scheme?
It's a little strange to me that you're starting before the first of the list but stopping sharply at the end of it. That is, you're taking the first element by itself and the first two elements by themselves, but you don't do the same for the last element or the last two elements.
That's somewhat orthogonal to the solution for the problem. I don't think the accumulator is making your life any easier here, and I would write the solution without it:
#lang racket
(require rackunit)
;; given a list of numbers and a period,
;; return a list of the averages of all
;; consecutive sequences of 'period'
;; numbers taken from the list.
(define ((moving-average period) l)
(cond [(< (length l) period) empty]
[else (cons (mean (take l period))
((moving-average period) (rest l)))]))
;; compute the mean of a list of numbers
(define (mean l)
(/ (apply + l) (length l)))
(check-equal? (mean '(4 4 1)) 3)
(check-equal? ((moving-average 3) '(1 3 2 7 6)) '(2 4 5))
Well, as a general rule, you want to separate the manner in which you recurse and/or iterate from the content of the iteration steps. You mention fold in your question, and this points in the right step: you want some form of higher-order function that will handle the list traversal mechanics, and call a function you supply with the values in the window.
I cooked this up in three minutes; it's probably wrong in many ways, but it should give you an idea:
;;;
;;; Traverse a list from left to right and call fn with the "windows"
;;; of the list. fn will be called like this:
;;;
;;; (fn prev cur next accum)
;;;
;;; where cur is the "current" element, prev and next are the
;;; predecessor and successor of cur, and accum either init or the
;;; accumulated result from the preceeding call to fn (like
;;; fold-left).
;;;
;;; The left-edge and right-edge arguments specify the values to use
;;; as the predecessor of the first element of the list and the
;;; successor of the last.
;;;
;;; If the list is empty, returns init.
;;;
(define (windowed-traversal fn left-end right-end init list)
(if (null? list)
init
(windowed-traversal fn
(car list)
right-end
(fn left-end
(car list)
(if (null? (cdr list))
right-end
(second list))
init)
(cdr list))))
(define (moving-average list)
(reverse!
(windowed-traversal (lambda (prev cur next list-accum)
(cons (avg (filter true? (list prev cur next)))
list-accum))
#f
#f
'()
list)))
Alternately, you could define a function that converts a list into n-element windows and then map average over the windows.
(define (partition lst default size)
(define (iter lst len result)
(if (< len 3)
(reverse result)
(iter (rest lst)
(- len 1)
(cons (take lst 3) result))))
(iter (cons default (cons default lst))
(+ (length lst) 2)
empty))
(define (avg lst)
(cond
[(null? lst) 0]
[(/ (apply + lst) (length lst))]))
(map avg (partition (list 1 2 3 4 5) 0 3))
Also notice that the partition function is tail-recursive, so it doesn't eat up stack space -- this is the point of result and the reverse call. I explicitly keep track of the length of the list to avoid either repeatedly calling length (which would lead to O(N^2) runtime) or hacking together a at-least-size-3 function. If you don't care about tail recursion, the following variant of partition should work:
(define (partition lst default size)
(define (iter lst len)
(if (< len 3)
empty
(cons (take lst 3)
(iter (rest lst)
(- len 1)))))
(iter (cons default (cons default lst))
(+ (length lst) 2)))
Final comment - using '() as the default value for an empty list could be dangerous if you don't explicitly check for it. If your numbers are greater than 0, 0 (or -1) would probably work better as a default value - they won't kill whatever code is using the value, but are easy to check for and can't appear as a legitimate average