Error " <istmt>: 3499 mm_root() not found" when finding the root using mm_root command

[Mohsen Jafarian]

Dear Statalisters,


I have an unbalanced panel data, where among many other variables, I have: p(current price), eps1, dps, eps2 and x(implied cost of capital). All variables, except x, are known and carry a value. An excerpt of first 20 observations is shown below.

	p	hvz1_eps	dps	hvz2_eps
1.	.	.	.	.
2.	.	.1247489	.	.
3.	.	.	.	.
4.	.	.	.	.
5.	1.91	.	0	.
6.	3.72	.	0	.1975022
7.	4.7	.2124444	0	.2172921
8.	5.37	.2103094	.19	.2131406
9.	6.72	.2066778	.23	.2100024
10.	6.57	.2006544	.22	.
11.	7.01	.	.22	.
12.	.	.	.	.
13.	.	.	.	.2131791
14.	.	.208495	.	.
15.	.	.	.	.
16.	5.3	.2281662	.12	.2314164
17.	5.39	.2250118	.27	.2278917
18.	6.1	.2258492	.4	.2257041
19.	6.71	.2245243	.32	.2224477
20.	6.91	.2251943	.4

My purpose is to find x using PEG model (by Easton (2004) by following formula:

I've searched previous posts and found this back in 2010: https://www.stata.com/statalist/arch.../msg01113.html

I made an effort to change the code to meet my setting:

     gen x=.
mata
v=J(1,1,.)
st_view(v,.,"x p hvz1_eps dps hvz2_eps")
function y(x,a,b,c,d) {
return((.2131406+(x*0.19)-.2103094)/x^2)
}
for (i=1;i<=rows(v);i++) {
r=mm_root(x=.,&y(),0,1,1e-9,1000,v[i,2],
v[i,3],v[i,4],v[i,5])
v[i,1]=x
}
end

To follow previous post, my intuition was that I should use the first non-missing observation, so I used observation number 8 in the code (in bold). The code is supposed to find the value of x between lower and upper bound of 0 and 1. Moremata is installed from SSC. But the code ends up with error:
Code:

 <istmt>:  3499  mm_root() not found

I am unfamiliar with Mata and it is kind of complicated for me. So I will be grateful if you guide me through this.

Thanks in advance for your kind attention.

Best,
Mohsen

[Hua Peng (StataCorp)]

First step is to determine if mm_root() is installed on your system, type

mata: mata which mm_root

If it is indeed installed, you should see the output like:

. mata: mata which mm_root
  mm_root():  lmoremata

If otherwise, ssc install moremata again. Please report back if mm_rrot() is available and error persists.

[Mohsen Jafarian]

Dear Hua,

Although I had installed moremata at the first place, it seems that it is only after restarting application that Stata can access mm_root. I did so and

<istmt>: 3499 mm_root() not found

disappeared.

But the original code only generates missing value for x variable when I run if for the first time. Output is shown below:

. gen x=.
(6,448 missing values generated)

. mata
------------------------------------------------- mata (type end to exit) ------------------------------
: v=J(1,1,.)

: st_view(v,.,"x p hvz1_eps dps hvz2_eps")

: function y(x,a,b,c,d) {
> return((.2131406+(x*0.19)-.2103094)/x^2)
> }
note: argument a unused
note: argument b unused
note: argument c unused
note: argument d unused

: for (i=1;i<=rows(v);i++) {
> r=mm_root(x=.,&y(),0,1,1e-9,1000,v[i,2],
> v[i,3],v[i,4],v[i,5])
> v[i,1]=x
> }

: end
--------------------------------------------------------------------------------------------------------

Running the same code again brought following errors:

. gen x=.
(6,448 missing values generated)

. mata
------------------------------------------------- mata (type end to exit) ------------------------------
: v=J(1,1,.)

: st_view(v,.,"x p hvz1_eps dps hvz2_eps")

: function y(x,a,b,c,d) {
y() already exists
r(3000);

: return((.2131406+(x*0.19)-.2103094)/x^2)
'return' found where almost anything else expected
r(3000);

: }
expression invalid
r(3000);

: for (i=1;i<=rows(v);i++) {
> r=mm_root(x=.,&y(),0,1,1e-9,1000,v[i,2],
> v[i,3],v[i,4],v[i,5])
> v[i,1]=x
> }

: end
--------------------------------------------------------------------------------------------------------

I really appreciate advises on this matter.

Best Regards,
Mohsen

[Sergio Correia]

You made three "mistakes":

1. You typed mata instead of mata: (note the colon!), so the code kept running even after the first mistake.
2. You created the function -y-, which will keep existing until you either type i) clear mata, ii) mata: mata drop y() , or iii) restart stata. For good practice, you can start your do file with "clear all" or "clear mata"
3. " the original code only generates missing value for x variable when I run if for the first time". That's because the second time you ran the code the variables (r, v, etc.) already existed. Unless you wrap that code inside a function, all the variables created will be global (you can see which variables currently exist with "mata: mata describe". Note that typing "clear mata" above will also solve this problem.

Best,
S

[Mohsen Jafarian]

Hi Sergio,

Thank you for truly helpful comments. I followed your guideline and now the code runs smoothly.

I share the code for future reference of those who need (usual disclaimer applies).

clear mata
gen xx=.
mata:
v=J(1,1,.)
st_view(v,.,"x p eps1 dps eps2")
function y(x,a,b,c,d) {
return((-(x^2)+((d+(x*c)-b)/a)))
}
for (i=1;i<=rows(v);i++) {
r=mm_root(x=.,&y(),0,1,1e-9,1000,v[i,2],
v[i,3],v[i,4],v[i,5])
v[i,1]=x
}
end

Best,
Mohsen