+- HP Forums (https://www.hpmuseum.org/forum)
+-- Forum: HP Calculators (and very old HP Computers) (/forum-3.html)
+--- Forum: HP Prime (/forum-5.html)
+--- Thread: delcols bug (/thread-21096.html)
delcols bug - Albert Chan - 01-02-2024 07:08 PM
Below produce permanent of square matrix.
delcols does not work (tested on both HP emulators 2018/10/16, 2023/4/13)
I had to add mydelcols := delcols, and use mydelcols to make it work.
Code:
#cas
per(m) :=
BEGIN
LOCAL r, mydelcols := delcols;
IF len(m)<2 THEN RETURN m[1][1] END
r := map(x->sum(!x),m);
r := index(r,max(r)); /* most zeroes row */
r, m := m[r], delrows(m,r);
sum(when(r[k],r[k]*per(mydelcols(m,k)),0), k=1..len(r));
END;
#endLets debug with print(m); right after LOCAL r
> per([[0,1,2],[3,4,5],[6,7,8]]) → 144
Terminal:
m:[[0,1,2],[3,4,5],[6,7,8]]
m:[[3,5],[6,8]]
m:[[8]]
m:[[6]]
m:[[3,4],[6,7]]
m:[[7]]
m:[[6]]
This is using delcols directly inside sum:
Terminal:
m:[[0,1,2],[3,4,5],[6,7,8]]
m:[[4],[7]]
m:Error: Bad Argument Type
m:Error: Invalid Dimension
RE: delcols bug - jte - 01-02-2024 09:20 PM
Albert, thanks for bringing this up. I’ve filed a ticket for this in the bug tracker I’ve set up to help organize development. The programs given in the ticket are a bit shorter, namely
Code:
#cas
shortper(m) :=
BEGIN
sum(delcols(m,k),k=1..1);
END;
#endCode:
#cas
shortper(m) :=
BEGIN
sum(delcols(m,1),k=1..1);
END;
#endRE: delcols bug - jte - 01-05-2024 09:49 PM
Prof. Parisse has clarified what is going on here: delcols(-) is modifying its first argument (the matrix) inside the sum(-). For delcols(-)’ modification to its matrix argument to be observed from the outside, the matrix must be passed as a variable (not evaluated to a temporary).
One way of passing the matrix argument as a variable is to quote it (another is to use delcols(-) inside sum(-)…). A screenshot showing how quoting the matrix argument makes the modification visible afterwards follows:
If a temporary is passed, the change is not made to variables that contributed to the construction of the temporary, as shown by the following screenshot:
RE: delcols bug - parisse - 01-06-2024 10:45 AM
delcols does not auto-quote it's argument, sum does. If the first argument of delcols is an identifier of a variable containing a matrix, then the modified matrix is stored in the variable.
That should explain the issue.
RE: delcols bug - Albert Chan - 01-06-2024 11:55 AM
(01-06-2024 10:45 AM)parisse Wrote: delcols does not auto-quote it's argument, sum does.
Thanks!
OP sum line should avoid matrix to be modified, like this:
> sum(when(r[k],r[k]*per(delcols(m+0,k)),0), k=1..len(r));
Although this does not explain why OP hack work too.
I would have expected mydelcols := delcols line have no effect.
> mydelcols := delcols;
> sum(when(r[k],r[k]*per(mydelcols(m,k)),0), k=1..len(r));
Quote: If the first argument of delcols is an identifier of a variable containing a matrix,
then the modified matrix is stored in the variable.
Perhaps delcols/delrows possibly modify matrix behavior should be in documentation?
m2 := delcols(m,k) // m unchanged
m2 := delcols('m',k) // m changed to m2
If this is not yet in documentation (i.e. undefined behavior), perhaps we can change it?
I think m should not be modified, unless explicitly asked, say m := delcols('m',k)
RE: delcols bug - parisse - 01-08-2024 04:24 PM
If I decide to change something, then I'll certainly simplify, delrows/delcols shoud not modify the matrix.