Back home, I will tell wife that it took me 5 hours to understand 10 lines.

To multiply matrix I could imagine :

SQL> create table m1 as
  select 1 x, 1 y, 10 val from dual 
  union all select 1,2,20 from dual
  union all select 2,1,30 from dual
  union all select 2,2,40 from dual
  union all select 3,1,50 from dual
  union all select 3,2,60 from dual
  union all select 4,1,70 from dual
  union all select 4,2,80 from dual;
Table created.
SQL> create table m2 as
  select 1 x, 1 y, 100 val from dual 
  union all select 1,2,200 from dual
  union all select 1,3,300 from dual
  union all select 2,1,400 from dual
  union all select 2,2,500 from dual
  union all select 2,3,600 from dual;
Table created.
SQL> select * from m1 pivot (max(val) for y in (1,2))

         X          1          2
---------- ---------- ----------
         1         10         20
         2         30         40
         3         50         60
         4         70         80
SQL> select * from m2 pivot (max(val) for y in (1,2,3))

         X          1          2          3
---------- ---------- ---------- ----------
         1        100        200        300
         2        400        500        600
SQL> select * from (
select * from dual model
reference 
  m1 on (select * from m1) dimension by (x,y) measures (val)
reference 
  m2 on (select * from m2) dimension by (x,y) measures (val)
reference 
  dim1 on (select max(x) x,max(y) y, 0 dummy from m1) dimension by (dummy) measures (x,y)
reference 
  dim2 on (select max(x) x,max(y) y, 0 dummy from m2) dimension by (dummy) measures (x,y)
dimension by (1 x,1 y) measures (0 val)
rules iterate (100) until (dim2.x[0]=iteration_number+1)
(val[for x from 1 to dim1.x[0] increment 1, for y from 1 to dim2.y[0] increment 1]=
nvl(val[cv(),cv()],0)+m2.val[iteration_number+1,cv(y)]*m1.val[cv(x),iteration_number+1] ))
pivot (max(val) for y in (1,2,3))

         X          1          2          3
---------- ---------- ---------- ----------
         1       9000      12000      15000
         2      19000      26000      33000
         3      29000      40000      51000
         4      39000      54000      69000

However, I vote for this query over the lsc2, this one has a palatable goal.
is it possible to do an inverse of matrix using the pivot and model clause ? that would be usefull, awfully complicated so that you put it in the interview under the nose of the candidate and ask what the query does.

select * from (
select * from dual model dimension by (0 day, 1 week) 
measures (0 c)
(
c[for day from 0 to 6 increment 1, for week from 0 to 4 increment 1]=
  1+trunc(trunc(sysdate,'MM'),'IW')-trunc(sysdate,'MM')+cv(day)+7*cv(week),
c[any,any]=case when c[cv(),cv()] between 
  1 and extract(day from last_day(sysdate)) then c[cv(),cv()] end))
pivot (max(c) for day in (
  0 as mon,1 as tue,2 as wed,3 as thu,4 as fri,5 as sat,6 as sun))
order by week;

but is this really obscure?