Problem with diag() function

[Clyde Schechter]

Now, I know I have yet to imbibe my full daily dose of caffeine, and maybe I didn't get as many hours of sleep last night as I should have. But for the life of me, I do not see what I am doing wrong here:

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 A12 A13)
  35.49362   6.902328   10.190154   9.059019     4.90965 -14.830358   -33.51197   -29.69985 -10.174074   -.5877179     .6048743   .030278724  -.02317784
  6.902328  33.360325   2.3703847   2.409502    3.867615 -27.242607    -4.16133  -3.7759824 -3.7173665    2.631272    1.0363309     -.170952   -.0239322
 10.190154  2.3703847   19.771967  14.099613   -.3304263  -.8716252   -15.14319   -9.384815  .56452847  -.04007869    -.1524103    .02515032 -.004933141
  9.059019   2.409502   14.099613  28.527035  -3.2938886  -2.409359   -9.517553  -22.279425  2.6908784    .1336752    -.8775372   .013183976  -.04381609
   4.90965   3.867615   -.3304263 -3.2938886   18.270546  -2.937681   -1.941296   1.8324593  -16.78055   .16291936    .11271199    .05821532 -.007008217
-14.830358 -27.242607   -.8716252  -2.409359   -2.937681    51.1867    7.054405    8.733149   9.753474   2.0740674    -.3768104     .0627132   .02383808
 -33.51197   -4.16133   -15.14319  -9.517553   -1.941296   7.054405    46.93388   28.170256  .06202966    .4026062   -.17233193  -.071465515  .009773405
 -29.69985 -3.7759824   -9.384815 -22.279425   1.8324593   8.733149   28.170256    60.49548  -5.147311  -.12426873    -.1322009  -.027954483   .06627995
-10.174074 -3.7173665   .56452847  2.6908784   -16.78055   9.753474   .06202966   -5.147311   43.08675   .14523736    .10655335    .01136777   .05028607
 -.5877179   2.631272  -.04007869   .1336752   .16291936  2.0740674    .4026062  -.12426873  .14523736    7.743637    .10186014   .004620384 .0025797726
  .6048743  1.0363309   -.1524103  -.8775372   .11271199  -.3768104  -.17233193   -.1322009  .10655335   .10186014     .6255068 -.0016476278 .0015802446
.030278724   -.170952   .02515032 .013183976   .05821532   .0627132 -.071465515 -.027954483  .01136777  .004620384 -.0016476278   .014333568 .0008621671
-.02317784  -.0239322 -.004933141 -.04381609 -.007008217  .02383808  .009773405   .06627995  .05028607 .0025797726  .0015802446  .0008621671   .03381956
end

is my data set.

. about

Stata/MP 14.2 for Windows (64-bit x86-64)
Revision 16 Mar 2017
Copyright 1985-2015 StataCorp LLC

Total physical memory:     8269900 KB
Available physical memory: 3129912 KB

Single-user 2-core Stata perpetual license:
       Serial number:  REDACTED
         Licensed to:  Clyde Schechter
                       Albert Einstein College of Medicine

.
. count
  13

. assert r(N) == 13

. des, short

Contains data
  obs:            13                          
 vars:            13                          
 size:           676                          
Sorted by:
     Note: Dataset has changed since last saved.

. assert r(k) == 13

.
. mkmat A*, matrix(V)

.
. local r = rowsof(V)

. local c = colsof(V)

. assert `r' == `c'

.
.
. matrix D = diag(V)
conformability error
r(503);

It is hard for me to imagine there is a bug in a straightforward, and ancient function like diag().

Evidently the workaround is to just construct D element by element, and I have done that and can move forward. But I really don't get what I'm doing wrong here. Does anyone see it?

Added: Even more puzzling: although -diag()- chokes on this matrix, -syminv()- has no problem with it.

[Robert Picard]

The diag() matrix function creates a matrix from a vector, which explains the conformability error. You probably wanted vecdiag(V):

mat d = vecdiag(V)
mat list d
matrix D = diag(d)
mat list D

[Clyde Schechter]

Yes, thank you, Robert.