I am trying to sort this array of tuples by the 2nd element but if the elements being compared are equal, I want to sort them based off the 1st element. I've done research but can only find to sort by 1st element or by 2nd element only, instead of checking both
You can make a function that maps (x, y) on (Down y, x) so that it first on the second item in descending order, and then by the first item:
import Data.List(sortOn)
import Data.Ord(Down(Down))
sortTupleSndFirst = sortOn (\(x, y) -> (Down y, x))
Related
I am trying to make a List of Lists of Lists without values. If N_meses = 4 I want List =[[[A,B,C,D]]].
I get what I want ( List = [[[]]] ) but every lists have the same values as you can see in the print I attached. How can I change this code so every lists have a different "value"?
I am doing this
generate_table(Num_investigadores, Num_actividades, N_Meses, Tabela) :-
length(Row, Num_actividades),
length(X,N_Meses),
maplist(=(X), Row),
length(Tabela, Num_investigadores),
maplist(=(Row), Tabela).
The culprit is in essence the:
%% ...
maplist(=(X), Row),
%% ...
Here you basically defined a list X, and then you set with maplist/2 that all elements in Row are unified with that X. In the unification process. This thus means that all the elements of Row will in essence point to the same list.
Nevertheless, I think it would definitely help if you make the predicate less ambitious: implement helper predicates and let each predicate do a small number of things.
We can for example first design a predicate lengthlist/2 that is the "swapped" version of length/2, and thus has as first parameter the length and as second parameter the list, like:
lengthlist(N, L) :-
length(L, N).
Now we can construct a predicate that generates a 2d rectangular list, for example:
matrix(M, N, R) :-
lengthlist(M, R),
maplist(lengthlist(N), R).
here we thus first use lengthlist to construct a list with N elements, and then we use maplist/2 to call lengthlist(N, ...) on every element, such that every element is unified with a list of N elements. We thus construct a 2d list with M elements where every elements is a list of N elements.
Then finally we can construct a 3d tensor:
tensor3(L, M, N, T) :-
lengthlist(L, T),
maplist(matrix(M, N), T).
Here we thus construct an L×M×N tensor.
We can in fact generalize the above to construct a arbitrary deep cascade of lists that is "rectangular" (in the sense that for each dimension, the lists have the same number of elements), but I leave this as an exercise.
The sort or member functions in MAXIMA do not return the indicies of elements sorted or the index of element being searched for. Is there a sort function that returns indices? Thanks.
Sven
There aren't built-in functions for those tasks, but I think you can write something fairly easily:
member_index (x, l) :=
(sublist_indices (l, lambda ([y], x = y)),
if %% = [] then false else first (%%));
This member_index function returns the index of the first instance of x. If you want indices for all instances, just remove the if expression and return %% instead.
ranks (l) :=
block ([l1, l2],
makelist ([l[i], i], i, length (l)),
l1 : sort (%%, lambda ([x, y], orderlessp (x, y))),
l2 : makelist (0, length (l)),
for i thru length (l) do l2[l1[i][2]] : i,
l2);
This ranks function returns integer values when there are ties (i.e. same values). Sometimes it is preferable to return the average rank for ties; that take a little more code. There is a ranks function in share/amatrix/wilcoxon.mac which does average ties. You can load that function via load(wilcoxon);.
I have predicate which shows all numbers which are smaller than 10.
small([H|T],H):- H=<10.
small([_|T],X):-small(T,X).
It is quite simple. But how should I change it so that I could compare every next item to the first element?
For example, ?- small([4,2,3,9,1,0,12],X). would show X=2;X=3;X=1;X=0 ?
You can split the problem in two procedures, one that takes the first element and then applies the recursion over the remaining elements of the list and checks whether each item is less than or equal to the first element:
small([M|T],X):- small1(T, M, X).
small1([H|_], M ,H):- H =< M.
small1([_|T], M, X):- small1(T, M, X).
I am trying to solve the maximum sub array problem with a brute force approach i.e generating all the possible subarrays combinations. I got something that works but it's not satisfying at all because it produces way too many duplicated subarrays.
Does anyone knows a smart way to generate all the subarrays (in [[]] form) with a minimal number of duplicated elements ?
By the way, I'm new to Haskell. Here's my current solution:
import qualified Data.List as L
maximumSubList::[Integer]->[Integer]
maximumSubList x = head $ L.sortBy (\a b -> compare (sum b) (sum a)) $ L.nub $ slice x
where
-- slice will return all the "sub lists"
slice [] = []
slice x = (slice $ tail x) ++ (sliceLeft x) ++ (sliceRight x)
-- Create sub lists by removing "left" part
-- ex [1,2,3] -> [[1,2,3],[2,3],[3]]
sliceRight [] = []
sliceRight x = x : (sliceRight $ tail x)
-- Create sub lists by removing "right" part
-- ex [1,2,3] -> [[1,2,3],[1,2],[1]]
sliceLeft [] = []
sliceLeft x = x : (sliceLeft $ init x)
There are many useful functions for operating on lists in the standard Data.List module.
import Data.List
slice :: [a] -> [[a]]
slice = filter (not . null) . concatMap tails . inits
dave4420's answer is how to do what you want to do using smart, concise Haskell. I'm no Haskell expert, but I occasionally play around with it and find solving a problem like this to be an interesting distraction, and enjoy figuring out exactly why it works. Hopefully the following explanation will be helpful :)
The key property of dave4420's answer (which your answer doesn't have) is that the pair (startPos, endPos) is unique for each subarray it generates. Now, observe that two subarrays are distinct if either their startPos or endPos is different. Applying inits to the original array returns a list of subarrays that each have unique startPos, and the same endPos (equal to the number of elements in the array). Applying tails to each of these subarrays in turn produces another list of subarrays -- one list of subarrays is output per input subarray. Notice that tails does not disturb the distinctness between input subarrays because the subarrays output by invoking tails on a single input subarray all retain the same startPos: that is, if you have two subarrays with distinct startPoses, and put both of them through tails, each of the subarrays produced from the first input subarray will be distinct from each of the subarrays produced from the second one.
Additionally, each of the subarrays produced by the invocation of tails on a single subarray are distinct because, although they all share the same startPos, they all have distinct endPoses. Therefore all subarrays produced by (concatMap tails) . inits are distinct. It only remains to note that no subarray is missed out: for any subarray starting at position i and ending at position j, that subarray must appear as the j-i+1th list produced by applying tails to the i+1th list produced by inits. So in conclusion, every possible subarray appears exactly once!
I am trying to write a procedure order(List,Result) that has a List as input and returns a list Result of ordered pairs such that:
the first element of the ordered pair is the position of the pair in the list, and
the second element of the ordered pair is the element from List n the corresponding position.
Example:
if List = [a,b,c,d], the procedure order(List,Result) outputs the list:
Result = [(1,a), (2,b),(3,c),(4,d)].
I am struggling with the counter for the position of the pair in the list. I have made attempts such as:
increment(Accum,Total):-
Total is Accum + 1.
order([],[]).
order([Head|Tail],Result):-
order(Tail, NewTail),
NewCount is Count + 1,
increment(NewCount,Count),
Result = [(Count,Head)|NewTail].
Please help anyone?
The two clauses: NewCount is Count + 1 and increment(NewCount,Count) basically have the same meaning. You didn't make clear that Count is an input variable and it has a base case of 1, so Prolog didn't know where to start unifying values for it. For example, you should use Count as an input argument as follows (it doesn't change much if compared with your version):
order([],[], _).
order([Head|Tail],[(Count,Head)|NewTail], Count):-
NewCount is Count + 1,
order(Tail, NewTail, NewCount).
order(List, Result ):- order(List, Result, 1).
If you're OK with using findall/3 then this is probably the simplest solution:
order(List, Result) :-
findall(Index-Elem, nth1(Index, List, Elem), Result).
Note that here the key-value pairs are represented using the term -/2, which is how pairs are usually represented in Prolog, e.g. this is what keysort/2 expects.
order(List,Result) :-
findall((N,E),(
append(L0,[E|_],List),
length([_|L0],N)),
Result).