11:08 a.m.
For anyone interested in working with state space modela via gretl:
please note that we have now replaced the documentation in chapter 30
of the Gretl User's Guide with an account of the new (bundle-style)
interface. We think the new code is now ready to roll, but if anyone
is able to do some testing before the next release -- which should be
in a week or so -- we'd be grateful.
The old Kalman interface is still available for now (but will likely
be removed in the release following the next one). And you can find
the old doc at
http://gretl.sourceforge.net/papers/kalman_old.pdf
Allin Cottrell
1:09 p.m.
Il 22/06/2016 13:08, Allin Cottrell ha scritto:
For anyone interested in working with state space modela via gretl:
please note that we have now replaced the documentation in chapter 30
of the Gretl User's Guide with an account of the new (bundle-style)
interface. We think the new code is now ready to roll, but if anyone
is able to do some testing before the next release -- which should be
in a week or so -- we'd be grateful.
The old Kalman interface is still available for now (but will likely
be removed in the release following the next one). And you can find
the old doc at
http://gretl.sourceforge.net/papers/kalman_old.pdf
Allin Cottrell
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
Dear Sven,
I've tried the example 30.2: I've copied and pasted (and run) the scipt
30.2.
I've recived the following error message: Data types not conformable for
operation
Actually the same message arised with other data.
Where is the problem?
Paolo
--
Paolo Chirico
Assistant Professor of Economic Statistics
Dept. of Economics and Statistics "Cognetti de Martiis"
Campus Luigi Einaudi, University of Turin
ITALY
1:15 p.m.
Dear Sven,
I've tried the example 30.2: I've copied and pasted (and run) the script
30.2.
I've recived the following error message: "Data types not conformable
for operation"
Actually the same message arised with other data.
Where is the problem?
Paolo
4:03 p.m.
On Tue, 28 Jun 2016, Paolo Chirico wrote:
Dear Sven,
I've tried the example 30.2: I've copied and pasted (and run) the script
30.2.
I've recived the following error message: "Data types not conformable for
operation"
Actually the same message arised with other data.
Where is the problem?
Do you mean the local level model example? I'm posting it here for
clarity:
<hansl>
nulldata 200
set seed 101010
setobs 1 1 --special-time-series
/* set the true variance parameters */
true_s1 = 0.5
true_s2 = 0.25
/* and simulate some data */
v = normal() * sqrt(true_s1)
w = normal() * sqrt(true_s2)
mu = 2 + cum(v)
y = mu + w
/* starting values for variance estimates */
s1 = 1
s2 = 1
/* state-space model set-up */
bundle kb = ksetup(y, 1, 1, s1)
kb.obsvar = s2
kb.diffuse = 1
/* ML estimation of variances */
mle ll = ERR ? NA : kb.llt
ERR = kfilter(&kb)
params kb.statevar kb.obsvar
end mle
/* compute the smoothed state */
ksmooth(&kb)
series muhat = kb.state
</hansl>
This should run ok on the snapshots. Which is the instruction that
triggers the error?
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
7:59 p.m.
Yes, the local level example.
This is the script output, when I run the local level script:
? nulldata 200
periodicity: 1, maxobs: 200
observations range: 1 to 200
? set seed 101010
Pseudo-random number generator seeded with 101010
? setobs 1 1 --special-time-series
Full data range: 1 - 200 (n = 200)
/* set the true variance parameters */
? true_s1 = 0.5
Generated scalar true_s1 = 0.5
? true_s2 = 0.25
Generated scalar true_s2 = 0.25
/* and simulate some data */
? v = normal() * sqrt(true_s1)
Generated series v (ID 2)
? w = normal() * sqrt(true_s2)
Generated series w (ID 3)
? mu = 2 + cum(v)
Generated series mu (ID 4)
? y = mu + w
Generated series y (ID 5)
/* starting values for variance estimates */
? s1 = 1
Generated scalar s1 = 1
? s2 = 1
Generated scalar s2 = 1
/* state-space model set-up */
? bundle kb = ksetup(y, 1, 1, s1)
Data types not conformable for operation
Error executing script: halting
bundle kb = ksetup(y, 1, 1, s1)
Where I'm wrong?
Paolo
Il 28/06/2016 18:03, Riccardo (Jack) Lucchetti ha scritto:
On Tue, 28 Jun 2016, Paolo Chirico wrote:
> Dear Sven,
> I've tried the example 30.2: I've copied and pasted (and run) the
> script 30.2.
> I've recived the following error message: "Data types not conformable
> for operation"
> Actually the same message arised with other data.
> Where is the problem?
Do you mean the local level model example? I'm posting it here for
clarity:
<hansl>
nulldata 200
set seed 101010
setobs 1 1 --special-time-series
/* set the true variance parameters */
true_s1 = 0.5
true_s2 = 0.25
/* and simulate some data */
v = normal() * sqrt(true_s1)
w = normal() * sqrt(true_s2)
mu = 2 + cum(v)
y = mu + w
/* starting values for variance estimates */
s1 = 1
s2 = 1
/* state-space model set-up */
bundle kb = ksetup(y, 1, 1, s1)
kb.obsvar = s2
kb.diffuse = 1
/* ML estimation of variances */
mle ll = ERR ? NA : kb.llt
ERR = kfilter(&kb)
params kb.statevar kb.obsvar
end mle
/* compute the smoothed state */
ksmooth(&kb)
series muhat = kb.state
</hansl>
This should run ok on the snapshots. Which is the instruction that
triggers the error?
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
Paolo Chirico
Assistant Professor of Economic Statistics
Dept. of Economics & Statistics "Cognetti de Martiis"
Campus Luigi Einaudi, University of Turin
ITALY
8:47 p.m.
On Tue, 28 Jun 2016, Paolo Chirico wrote:
Yes, the local level example.
This is the script output, when I run the local level script:
? nulldata 200
periodicity: 1, maxobs: 200
observations range: 1 to 200
? set seed 101010
Pseudo-random number generator seeded with 101010
? setobs 1 1 --special-time-series
Full data range: 1 - 200 (n = 200)
/* set the true variance parameters */
? true_s1 = 0.5
Generated scalar true_s1 = 0.5
? true_s2 = 0.25
Generated scalar true_s2 = 0.25
/* and simulate some data */
? v = normal() * sqrt(true_s1)
Generated series v (ID 2)
? w = normal() * sqrt(true_s2)
Generated series w (ID 3)
? mu = 2 + cum(v)
Generated series mu (ID 4)
? y = mu + w
Generated series y (ID 5)
/* starting values for variance estimates */
? s1 = 1
Generated scalar s1 = 1
? s2 = 1
Generated scalar s2 = 1
/* state-space model set-up */
? bundle kb = ksetup(y, 1, 1, s1)
Data types not conformable for operation
Error executing script: halting
> bundle kb = ksetup(y, 1, 1, s1)
Where I'm wrong?
I think your gretl version is not new enough. What is the "build
date" shown under /Help/About gretl? It should be within the last
week or so. (The latest snapshot was published today.)
Allin Cottrell
11:55 a.m.
I performed the example 30.3: "philips curve... with time-varying slope"
1) I was responded the following error (in the MLE step):
...
/* ML estimation of intercept and the two variances */
? mle LL = err ? NA : B.llt
? B.obsy = B.depvar - b0
? B.obsvar = s_obs^2
? B.statevar = s_state^2
? err = kfilter(&B)
? params b0 s_obs s_state
? end mle
Unexpected byte 0xe2
Syntax error
*** error in function at_each_step, line 1
b.obsymat = b.mX[b.t,]’
The formula 'err =
kfilter(&B)'
produced an error on execution
Error executing script: halting
end mle
2) How can I be sure that b.obsymat was varing if obymat in bundle
objects shows the initial value and not the last one?
Paolo
--
Paolo Chirico
Assistant Professor of Economic Statistics
Dept. of Economics and Statistics "Cognetti de Martiis"
Campus Luigi Einaudi, University of Turin
ITALY
12:16 p.m.
Am 05.07.2016 um 13:55 schrieb Paolo Chirico:
I performed the example 30.3: "philips curve... with
time-varying slope"
1) I was responded the following error (in the MLE step):
...
/* ML estimation of intercept and the two variances */
? mle LL = err ? NA : B.llt
? B.obsy = B.depvar - b0
? B.obsvar = s_obs^2
? B.statevar = s_state^2
? err = kfilter(&B)
? params b0 s_obs s_state
? end mle
Unexpected byte 0xe2
Syntax error
*** error in function at_each_step, line 1
> b.obsymat = b.mX[b.t,]’
The formula 'err = kfilter(&B)'
produced an error on execution
The 0xe2 byte thing sounds like a non-ASCII character sneaked into in
the file with the code. In your email the transposition sign after the ]
looks suspicious, but after sending via email anything can happen.
As always it could be useful to know your gretl version, platform, and
how the new Kalman code was installed, if not with gretl itself.
cheers,
sven
1:45 p.m.
On Tue, 5 Jul 2016, Sven Schreiber wrote:
Am 05.07.2016 um 13:55 schrieb Paolo Chirico:
> I performed the example 30.3: "philips curve... with time-varying slope"
> 1) I was responded the following error (in the MLE step):
> ...
> /* ML estimation of intercept and the two variances */
> ? mle LL = err ? NA : B.llt
> ? B.obsy = B.depvar - b0
> ? B.obsvar = s_obs^2
> ? B.statevar = s_state^2
> ? err = kfilter(&B)
> ? params b0 s_obs s_state
> ? end mle
> Unexpected byte 0xe2
> Syntax error
> *** error in function at_each_step, line 1
>> b.obsymat = b.mX[b.t,]’
> The formula 'err = kfilter(&B)'
> produced an error on execution
The 0xe2 byte thing sounds like a non-ASCII character sneaked into in the
file with the code. In your email the transposition sign after the ] looks
suspicious, but after sending via email anything can happen.
Ah, that's it: the transposition symbol in gretl is 0x27 (the ASCII
plain, upright single quote), but in Paolo's mail that's replaced with
"right single quotation mark" (character 0x92 in Windows code page
1252). On conversion to UTF-8 that becomes the 3-byte sequence 0xE2
0x80 0x99, which begins with 0xE2!
Allin
1:13 p.m.
On Tue, 5 Jul 2016, Paolo Chirico wrote:
I performed the example 30.3: "philips curve... with
time-varying slope"
1) I was responded the following error (in the MLE step):
...
/* ML estimation of intercept and the two variances */
? mle LL = err ? NA : B.llt
? B.obsy = B.depvar - b0
? B.obsvar = s_obs^2
? B.statevar = s_state^2
? err = kfilter(&B)
? params b0 s_obs s_state
? end mle
Unexpected byte 0xe2
Syntax error
*** error in function at_each_step, line 1
> b.obsymat = b.mX[b.t,]’
The formula 'err = kfilter(&B)'
produced an error on execution
Error executing script: halting
> end mle
Apparently corruption has occurred in copy-and-paste: there's no byte
0xe2 (which corresponds to a-circumflex in the ISO 8-bit encodings!)
in the original script. I've put a copy of the script for download at
http://users.wfu.edu/cottrell/tmp/kalman_phillips.inp
2) How can I be sure that b.obsymat was varing if obymat in bundle
objects
shows the initial value and not the last one?
You can modify the function at_each_step() as follows:
function void at_each_step(bundle *b)
printf "at_each_step, before: %g\n", b.obsymat[1]
b.obsymat = b.mX[b.t,]'
printf "at_each_step, after: %g\n", b.obsymat[1]
end function
This will show the value changing at each time-step.
Allin Cottrell
2846
days inactive
2859
days old
10 comments
5 participants
participants (5)
-
Allin Cottrell -
Paolo Chirico -
Paolo Chirico -
Riccardo (Jack) Lucchetti -
Sven Schreiber