hi all,
we have a rather weird problem with recursive Mata structures and
pointers, which is even somwhat beyond the limits of what we can express in
English. Allow us to try it anyway.
The short version of our question is, whether there is a better way
to implement the "setpoint" function described below? We don't have any
experience with pointers in Mata, and our attempts so far might be a
complet dead end. Is so, just stop us from considering this any further. We
don't expect that you write the code for us.
The long version is as follows: Suppose we have defined the following
data structure:
------------------------------------------------xyz.mata---
...
struct listelem {
real scalar a
real scalar b
pointer(struct listelem scalar) scalar nextpoint
}
struct listelem scalar function setelem(real scalar s) {
struct listelem scalar e
e.a=s
e.b=s
return e;
}
...
-----------------------------------------------------------
This defines the structure "listelem", which implements an element of a
"linked list" of points (a,b) and a cunstructor-function "setelem".
Now suppose that we have an argument vector defined as -seq = 1, 2, 3, 4-, and
we want to define the structure "pointlist", which is a linked list of
"listelem". More specifically, the structure "pointlist" should look as
follows:
the first element: [(1,1) pointer to the second element]
the second element: [(2,2) pointer to the third element]
the third element: [(3,3) pointer to the fourth element]
the fourth element: [(4,4) null pointer ]
In order to circumvent our English language problem, here is the complete Java
code of what we want to achieve in Mata (you may skip this, if you don't
understand Java):
------------------------------------------java code-----------
public class listelem {
public int a;
public int b;
public listelem nextpoint;
public void setelem(int s){
a=s;
b=s;
}
}
public class pointlist {
public listelem first; //the first element of the list
public listelem last; //the last element of the list
//initialisation of the first element
pointlist(int k){
first = new listelem();
first.a = k;
first.b = k;
last = first;
}
public static pointlist setpoint(int[] seq){
pointlist v; //return linked list
int length; //length of the argument vector seq
v = new pointlist(seq[0]);
length = seq.length;
//filling out of the list
for(int i=1; i<length; i++){
listelem newelem = new listelem();
newelem.setelem(seq[i]);
v.last.nextpoint = newelem;
v.last = v.last.nextpoint;
}
return v;
}
-----------------------------------------------------------
Now let us turn to our attempts to implement this in Mata.
Basically we have defined the structure "pointlist" that implements a "linked
list" of points (a,b) and a function "setpoint" that fills out this list with
the values of the argument vector "seq". Below is how far we got. We have
indicated the line that causes trouble with a comment.
------------------------------------------------xyz.mata---
...
struct pointlist {
struct listelem scalar first
pointer(struct listelem scalar) scalar last
}
struct pointlist scalar function setpoint( real vector seq) {
struct pointlist scalar v
real scalar i
real scalar length
length = length(seq)
v.first.a = seq[1]
v.first.b = seq[1]
v.last = &v.first
liststruct(v)
for (i=2; i<=length; i++) {
struct listelem m
m= setelem(seq[i])
v.last.nextpoint= m // <- See error message below
v.last = &last.nextpoint
}
return(v)
}
...
-----------------------------------------------------------
Compiling this code we receive the following error message at the indicated
line:
------------------------------------------------------------
type mismatch: exp.exp: pointer found where struct expected
-------------------------------------------------------------
We somehow stick with this error message. Our main problem, however, concerns
the using of pointers. The pointer "last" should move along the list and
always point to the last element of the list (it means, it should allow to
set a new point at the end of the list). It does not work. How can we add a
new element at the end of the list?
Many regards
Uli and Magdalena
--
kohler@wz-berlin.de, luniak@wz-berlin.de
+49 (030) 25491-361
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/