5:17 p.m.
Dear gretl friends,
does gretl have a built-in function to compute the PDF of a multivariate
normal. I couldn't find one which actually suprises me :-)
The corresponding matlab function is described here:
http://de.mathworks.com/help/stats/mvnpdf.html?requestedDomain=www.mathwo...
Best,
Artur
5:48 p.m.
On Mon, 15 Feb 2016, Artur T. wrote:
Dear gretl friends,
does gretl have a built-in function to compute the PDF of a multivariate
normal. I couldn't find one which actually suprises me :-)
The corresponding matlab function is described here:
http://de.mathworks.com/help/stats/mvnpdf.html?requestedDomain=www.mathwo...
Hmmm, we don't have one but clearly adding it wouldn't be a problem. Just
out of curiosity: what would you need such a function for?
-------------------------------------------------------
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
-------------------------------------------------------
6:26 p.m.
Am 15.02.2016 um 18:48 schrieb Riccardo (Jack) Lucchetti:
On Mon, 15 Feb 2016, Artur T. wrote:
> Dear gretl friends,
>
> does gretl have a built-in function to compute the PDF of a multivariate
> normal. I couldn't find one which actually suprises me :-)
> The corresponding matlab function is described here:
> http://de.mathworks.com/help/stats/mvnpdf.html?requestedDomain=www.mathwo...
>
Hmmm, we don't have one but clearly adding it wouldn't be a problem.
Just out of curiosity: what would you need such a function for?
I'm astonished that Jack didn't answer how easy it is to make one
yourself :-)
The use case strikes me as pretty obvious: parametric simulation.
cheers,
sven
10:01 p.m.
Am 15.02.2016 um 19:26 schrieb Sven Schreiber:
Am 15.02.2016 um 18:48 schrieb Riccardo (Jack) Lucchetti:
> On Mon, 15 Feb 2016, Artur T. wrote:
>
>> Dear gretl friends,
>>
>> does gretl have a built-in function to compute the PDF of a multivariate
>> normal. I couldn't find one which actually suprises me :-)
>> The corresponding matlab function is described here:
>> http://de.mathworks.com/help/stats/mvnpdf.html?requestedDomain=www.mathwo...
>>
>>
>
> Hmmm, we don't have one but clearly adding it wouldn't be a problem.
> Just out of curiosity: what would you need such a function for?
>
Actually, I am just playing around with simple AR-models estimated by
means of the
Metropolis-Hasting algorithm. The Matlab example I found makes use of it to
setup a normal prior.
I'm astonished that Jack didn't answer how easy it is to make
one
yourself :-)
Haha, Sven, actually I really expected it. It in principle it's
not very
difficult
to set this up. But I just was curious whether there already exists a
built-in command.
The use case strikes me as pretty obvious: parametric simulation.
cheers,
sven
Artur
10:52 p.m.
On Mon, 15 Feb 2016, Artur T. wrote:
Am 15.02.2016 um 19:26 schrieb Sven Schreiber:
> I'm astonished that Jack didn't answer how easy it is to make one
> yourself :-)
Haha, Sven, actually I really expected it.
OK, you win, dammit!
<hansl>
function matrix mnormal_pdf(matrix x, matrix m, matrix Sigma)
# x should be n x k (n k-dimensional vectors)
# m should be a k-vector of means
# Sigma should be a kxk pd matrix
scalar k = cols(x)
scalar n = rows(x)
if k != rows(m)*cols(m)
return mshape(NA, n, 1)
endif
scalar dS = det(Sigma)
if dS <= 0
return mshape(NA, n, 1)
endif
matrix U = x .- vec(m)'
matrix U = U .* (U/Sigma)
scalar k = (2*$pi)^k / sqrt(dS)
return k .* exp(-0.5*sumr(U))
end function
nulldata 10000
series x = normal()
series y = 2*x + normal()
matrix A = {5, 1; 1, 1}
matrix C = {y, x}
matrix m = zeros(2,1)
series f = mnormal_pdf(C, m, A)
</hansl>
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
10:53 a.m.
Thank you, Jack! It's nicely written as always!
Best,
Artur
OK, you win, dammit!
<hansl>
function matrix mnormal_pdf(matrix x, matrix m, matrix Sigma)
# x should be n x k (n k-dimensional vectors)
# m should be a k-vector of means
# Sigma should be a kxk pd matrix
scalar k = cols(x)
scalar n = rows(x)
if k != rows(m)*cols(m)
return mshape(NA, n, 1)
endif
scalar dS = det(Sigma)
if dS <= 0
return mshape(NA, n, 1)
endif
matrix U = x .- vec(m)'
matrix U = U .* (U/Sigma)
scalar k = (2*$pi)^k / sqrt(dS)
return k .* exp(-0.5*sumr(U))
end function
nulldata 10000
series x = normal()
series y = 2*x + normal()
matrix A = {5, 1; 1, 1}
matrix C = {y, x}
matrix m = zeros(2,1)
series f = mnormal_pdf(C, m, A)
</hansl>
-------------------------------------------------------
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
noon
On Tue, 16 Feb 2016, Artur T. wrote:
Thank you, Jack! It's nicely written as always!
Thanks. The only problem is (now that I look at it again) the normalising
constant is wrong :)
But I'm sure you can post a corrected version with very little effort
(hint, hint!)
-------------------------------------------------------
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
-------------------------------------------------------
12:03 p.m.
Actually, I am just working at it. Was wondering how the pdf can get
larger than unity :-D
I'll be back soon on this!
Artur
Am 16.02.2016 um 13:00 schrieb Riccardo (Jack) Lucchetti:
On Tue, 16 Feb 2016, Artur T. wrote:
> Thank you, Jack! It's nicely written as always!
Thanks. The only problem is (now that I look at it again) the
normalising constant is wrong :)
But I'm sure you can post a corrected version with very little effort
(hint, hint!)
-------------------------------------------------------
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
12:08 p.m.
Am 16.02.2016 um 13:03 schrieb Artur T.:
Actually, I am just working at it. Was wondering how the pdf can get
larger than unity :-D
I'll be back soon on this!
Actually I was a little surprised about Jack's suggestion also with
respect to what it does. I thought a function is wanted that also does
the RNG inside.
-s
3 p.m.
Ok, I changed the definition of the constant k. However, there is still
some difference to the outcome using MATLAB... Attached is also a
comparison --in the univarite case -- using gretl's pdf-function.
Actually, I am not sure whether these differences are due to numerical
issues or whether there is still some issue in the "mnormal_pdf"-function
<hansl>
clear
set echo off
set messages off
function matrix mnormal_pdf(matrix x, matrix m, matrix Sigma)
# x should be n x k (n k-dimensional vectors)
# m should be a k-vector of means
# Sigma should be a kxk pd matrix
scalar k = cols(x)
scalar n = rows(x)
if k != rows(m)*cols(m)
return mshape(NA, n, 1)
endif
scalar dS = det(Sigma)
if dS <= 0
return mshape(NA, n, 1)
endif
matrix U = x .- vec(m)'
matrix U = U .* (U/Sigma)
#scalar k = (2*$pi)^k / sqrt(dS) # OLD
scalar k = 1/(sqrt(dS)*(2*$pi)^k)
return k * exp(-0.5*sumr(U))
end function
# Bivariate Example
matrix mu = zeros(2,1)
matrix Sigma = {.25, .3; .3, 1}
matrix x = seq(-3,3.2,0.2)
matrix X = x' ~ x'
matrix f = mnormal_pdf(X, mu, Sigma)
f = f ~ x'
gnuplot 1 2 --matrix=f --with-lines --output=display
# CHECK: If Sigma is a scalar the problem reduces to the univariate #normal
matrix mu = zeros(1,1)
matrix Sigma = {1}
matrix f2 = mnormal_pdf(X[,1], mu, Sigma)
matrix f3 = pdf(N,X[,1])
f = f2 ~ f3 ~ x'
gnuplot 1 2 3 --matrix=f --with-lines --output=display
</hansl>
Actually I was a little surprised about Jack's suggestion also
with
respect to what it does. I thought a function is wanted that also does
the RNG inside.
No, it is really just about computing the pdf.
Artur
3:26 p.m.
On Tue, 16 Feb 2016, Artur T. wrote:
Ok, I changed the definition of the constant k. However, there is
still
some difference to the outcome using MATLAB...
Hehe. We're convergiung to the truth: here's the correct one --- I think!
<hansl>
function matrix mnormal_pdf(matrix x, matrix m, matrix Sigma)
# x should be n x k (n k-dimensional vectors)
# m should be a k-vector of means
# Sigma should be a kxk pd matrix
scalar k = cols(x)
scalar n = rows(x)
if k != rows(m)*cols(m)
return mshape(NA, n, 1)
endif
scalar dS = det(Sigma)
if dS <= 0
return mshape(NA, n, 1)
endif
matrix U = x .- vec(m)'
matrix U = U .* (U/Sigma)
scalar k = dS * (2*$pi)^k
return exp(-0.5*sumr(U)) / sqrt(k)
end function
</hansl>
> Actually I was a little surprised about Jack's suggestion
also with
> respect to what it does. I thought a function is wanted that also does
> the RNG inside.
No, it is really just about computing the pdf.
Generating a sample of n independent Gaussian pseurdo-rvs with mean m and
covariance Sigma is rather trivial, in fact:
<hansl>
n = 20000
m = {1, 2, 3}
Sigma = {4,1,-1;1,4,1;-1,1,4}
A = mnormal(n, 3) * cholesky(Sigma)' .+ m
eval meanc(A)
eval mcov(A)
</hansl>
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
3:39 p.m.
Thanks for your response, Jack.
Hehe. We're convergiung to the truth: here's the correct one
--- I think!
<hansl>
function matrix mnormal_pdf(matrix x, matrix m, matrix Sigma)
# x should be n x k (n k-dimensional vectors)
# m should be a k-vector of means
# Sigma should be a kxk pd matrix
scalar k = cols(x)
scalar n = rows(x)
if k != rows(m)*cols(m)
return mshape(NA, n, 1)
endif
scalar dS = det(Sigma)
if dS <= 0
return mshape(NA, n, 1)
endif
matrix U = x .- vec(m)'
matrix U = U .* (U/Sigma)
scalar k = dS * (2*$pi)^k
return exp(-0.5*sumr(U)) / sqrt(k)
end function
</hansl>
Ahh, ok, I see the issue. It works now, and also replicates gretl's
pdf-function in the univariate case as well.
>> Actually I was a little surprised about Jack's suggestion
also with
>> respect to what it does. I thought a function is wanted that also does
>> the RNG inside.
>
> No, it is really just about computing the pdf.
Generating a sample of n independent Gaussian pseurdo-rvs with mean m
and covariance Sigma is rather trivial, in fact:
<hansl>
n = 20000
m = {1, 2, 3}
Sigma = {4,1,-1;1,4,1;-1,1,4}
A = mnormal(n, 3) * cholesky(Sigma)' .+ m
eval meanc(A)
eval mcov(A)
</hansl>
Artur
5:54 p.m.
Am 16.02.2016 um 16:26 schrieb Riccardo (Jack) Lucchetti:
On Tue, 16 Feb 2016, Artur T. wrote:
>> Actually I was a little surprised about Jack's suggestion also with
>> respect to what it does. I thought a function is wanted that also does
>> the RNG inside.
>
> No, it is really just about computing the pdf.
Generating a sample of n independent Gaussian pseurdo-rvs with mean m
and covariance Sigma is rather trivial, in fact:
Yeah, I wasn't really paying attention, sorry.
-sven
6:27 p.m.
Jack,
Artur's code chokes for me when I run your function code:
gretl version 2016a-git
Current session: 2016-02-16 18:17
Invalid argument
Error executing script: halting
matrix x = seq(-3,3.2,0.2)
I wonder why? My -gretl-'s fully up to date!
C
On 16 February 2016 at 15:26, Riccardo (Jack) Lucchetti <
r.lucchetti(a)univpm.it> wrote:
On Tue, 16 Feb 2016, Artur T. wrote:
Ok, I changed the definition of the constant k. However, there is still
> some difference to the outcome using MATLAB...
>
Hehe. We're convergiung to the truth: here's the correct one --- I think!
<hansl>
function matrix mnormal_pdf(matrix x, matrix m, matrix Sigma)
# x should be n x k (n k-dimensional vectors)
# m should be a k-vector of means
# Sigma should be a kxk pd matrix
scalar k = cols(x)
scalar n = rows(x)
if k != rows(m)*cols(m)
return mshape(NA, n, 1)
endif
scalar dS = det(Sigma)
if dS <= 0
return mshape(NA, n, 1)
endif
matrix U = x .- vec(m)'
matrix U = U .* (U/Sigma)
scalar k = dS * (2*$pi)^k
return exp(-0.5*sumr(U)) / sqrt(k)
end function
</hansl>
Actually I was a little surprised about Jack's suggestion also with
>> respect to what it does. I thought a function is wanted that also does
>> the RNG inside.
>>
>
> No, it is really just about computing the pdf.
>
Generating a sample of n independent Gaussian pseurdo-rvs with mean m and
covariance Sigma is rather trivial, in fact:
<hansl>
n = 20000
m = {1, 2, 3}
Sigma = {4,1,-1;1,4,1;-1,1,4}
A = mnormal(n, 3) * cholesky(Sigma)' .+ m
eval meanc(A)
eval mcov(A)
</hansl>
-------------------------------------------------------
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
--
Clive Nicholas
"My colleagues in the social sciences talk a great deal about methodology.
I prefer to call it style." -- Freeman J. Dyson
6:50 p.m.
Clive,
The arguments to seq() all have to be integers. Try replacing that line with “matrix x =
seq(-30,32,2)./10”
PS
From: gretl-users-bounces(a)lists.wfu.edu [mailto:gretl-users-bounces@lists.wfu.edu] On
Behalf Of Clive Nicholas
Sent: Tuesday, February 16, 2016 1:27 PM
To: r.lucchetti(a)univpm.it; Gretl list
Subject: Re: [Gretl-users] PDF of a multivariate normal
Jack,
Artur's code chokes for me when I run your function code:
gretl version 2016a-git
Current session: 2016-02-16 18:17
Invalid argument
Error executing script: halting
matrix x = seq(-3,3.2,0.2)
I wonder why? My -gretl-'s fully up to date!
C
On 16 February 2016 at 15:26, Riccardo (Jack) Lucchetti
<r.lucchetti@univpm.it<mailto:r.lucchetti@univpm.it>> wrote:
On Tue, 16 Feb 2016, Artur T. wrote:
Ok, I changed the definition of the constant k. However, there is still
some difference to the outcome using MATLAB...
Hehe. We're convergiung to the truth: here's the correct one --- I think!
<hansl>
function matrix mnormal_pdf(matrix x, matrix m, matrix Sigma)
# x should be n x k (n k-dimensional vectors)
# m should be a k-vector of means
# Sigma should be a kxk pd matrix
scalar k = cols(x)
scalar n = rows(x)
if k != rows(m)*cols(m)
return mshape(NA, n, 1)
endif
scalar dS = det(Sigma)
if dS <= 0
return mshape(NA, n, 1)
endif
matrix U = x .- vec(m)'
matrix U = U .* (U/Sigma)
scalar k = dS * (2*$pi)^k
return exp(-0.5*sumr(U)) / sqrt(k)
end function
</hansl>
Actually I was a little surprised about Jack's suggestion also with
respect to what it does. I thought a function is wanted that also does
the RNG inside.
No, it is really just about computing the pdf.
Generating a sample of n independent Gaussian pseurdo-rvs with mean m and covariance Sigma
is rather trivial, in fact:
<hansl>
n = 20000
m = {1, 2, 3}
Sigma = {4,1,-1;1,4,1;-1,1,4}
A = mnormal(n, 3) * cholesky(Sigma)' .+ m
eval meanc(A)
eval mcov(A)
</hansl>
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti@univpm.it<mailto:r.lucchetti@univpm.it>
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
_______________________________________________
Gretl-users mailing list
Gretl-users@lists.wfu.edu<mailto:Gretl-users@lists.wfu.edu>
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
Clive Nicholas
"My colleagues in the social sciences talk a great deal about methodology. I prefer
to call it style." -- Freeman J. Dyson
7:01 p.m.
On Tue, 16 Feb 2016, Summers, Peter wrote:
Clive,
The arguments to seq() all have to be integers. Try replacing that
line with “matrix x = seq(-30,32,2)./10”
Actually, the seq() arguments no longer have to be integers, but
that's pretty new in git, and Clive is still on 2015a-git (so git
prior to the 2015a release).
Allin
From: gretl-users-bounces(a)lists.wfu.edu
[mailto:gretl-users-bounces@lists.wfu.edu] On Behalf Of Clive Nicholas
Sent: Tuesday, February 16, 2016 1:27 PM
To: r.lucchetti(a)univpm.it; Gretl list
Subject: Re: [Gretl-users] PDF of a multivariate normal
Jack,
Artur's code chokes for me when I run your function code:
gretl version 2016a-git
Current session: 2016-02-16 18:17
Invalid argument
Error executing script: halting
> matrix x = seq(-3,3.2,0.2)
I wonder why? My -gretl-'s fully up to date!
In fact it's not.
7:05 p.m.
-----Original Message-----
From: gretl-users-bounces(a)lists.wfu.edu [mailto:gretl-users-
bounces(a)lists.wfu.edu] On Behalf Of Allin Cottrell
Sent: Tuesday, February 16, 2016 2:02 PM
To: Gretl list
Subject: Re: [Gretl-users] PDF of a multivariate normal
On Tue, 16 Feb 2016, Summers, Peter wrote:
> Clive,
>
> The arguments to seq() all have to be integers. Try replacing that
> line with “matrix x = seq(-30,32,2)./10”
Actually, the seq() arguments no longer have to be integers, but that's pretty
new in git, and Clive is still on 2015a-git (so git prior to the 2015a release).
Allin
Cool. I'm on the build from 1/16/16, so I'm also a bit off the cutting edge.
PS
8:36 p.m.
Actually, I'm on 2016a-git and I would very much appreciate it if we were
told via the list when a new (mini-)version of -gretl- has been rolled out.
I never hear anything about them.
C
On 16 February 2016 at 19:01, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Tue, 16 Feb 2016, Summers, Peter wrote:
Clive,
>
> The arguments to seq() all have to be integers. Try replacing that line
> with “matrix x = seq(-30,32,2)./10”
>
Actually, the seq() arguments no longer have to be integers, but that's
pretty new in git, and Clive is still on 2015a-git (so git
prior to the 2015a release).
Allin
From: gretl-users-bounces(a)lists.wfu.edu [mailto:
> gretl-users-bounces(a)lists.wfu.edu] On Behalf Of Clive Nicholas
> Sent: Tuesday, February 16, 2016 1:27 PM
> To: r.lucchetti(a)univpm.it; Gretl list
> Subject: Re: [Gretl-users] PDF of a multivariate normal
>
> Jack,
> Artur's code chokes for me when I run your function code:
>
> gretl version 2016a-git
> Current session: 2016-02-16 18:17
>
> Invalid argument
>
> Error executing script: halting
>
>> matrix x = seq(-3,3.2,0.2)
>>
> I wonder why? My -gretl-'s fully up to date!
>
In fact it's not.
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
Clive Nicholas
"My colleagues in the social sciences talk a great deal about methodology.
I prefer to call it style." -- Freeman J. Dyson
9:18 p.m.
On Tue, 16 Feb 2016, Clive Nicholas wrote:
Actually, I'm on 2016a-git and I would very much appreciate it if
we were
told via the list when a new (mini-)version of -gretl- has been rolled out.
I never hear anything about them.
That's what gretl-announce is for. See
http://lists.wfu.edu/mailman/listinfo/gretl-announce
However, git development is pretty much continuous, so being up with
the latest release does not equal being up with the latest code.
Allin
9:35 p.m.
On Tue, 16 Feb 2016, Allin Cottrell wrote:
On Tue, 16 Feb 2016, Clive Nicholas wrote:
> Actually, I'm on 2016a-git and I would very much appreciate it if we were
> told via the list when a new (mini-)version of -gretl- has been rolled out.
> I never hear anything about them.
That's what gretl-announce is for. See
http://lists.wfu.edu/mailman/listinfo/gretl-announce
However, git development is pretty much continuous, so being up with the
latest release does not equal being up with the latest code.
And if you want to have an idea of what's happening in quasi-real time,
look here (click on "History")
https://sourceforge.net/p/gretl/git/ci/master/tree/
-------------------------------------------------------
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
-------------------------------------------------------
9:15 a.m.
Am 16.02.2016 um 22:18 schrieb Allin Cottrell:
On Tue, 16 Feb 2016, Clive Nicholas wrote:
> Actually, I'm on 2016a-git and I would very much appreciate it if we were
> told via the list when a new (mini-)version of -gretl- has been rolled
> out.
> I never hear anything about them.
That's what gretl-announce is for. See
http://lists.wfu.edu/mailman/listinfo/gretl-announce
Of course you're right in principle, but I tend to agree with Clive that
a copying the announcement on gretl-users wouldn't hurt. It seems to me
that gretl-announce is for those that don't want to receive any other news.
thanks,
sven
1:27 p.m.
On Wed, 17 Feb 2016, Sven Schreiber wrote:
Am 16.02.2016 um 22:18 schrieb Allin Cottrell:
> On Tue, 16 Feb 2016, Clive Nicholas wrote:
>
>> Actually, I'm on 2016a-git and I would very much appreciate it if we were
>> told via the list when a new (mini-)version of -gretl- has been rolled
>> out.
>> I never hear anything about them.
>
> That's what gretl-announce is for. See
> http://lists.wfu.edu/mailman/listinfo/gretl-announce
>
Of course you're right in principle, but I tend to agree with Clive that a
copying the announcement on gretl-users wouldn't hurt. It seems to me that
gretl-announce is for those that don't want to receive any other news.
Fair enough. We could send a short message with a link to the
ChangeLog to gretl-users.
Allin
7:49 p.m.
And this failed political scientist says that, in what is an election year
for some countries: "I vote for this."
C
On 17 February 2016 at 13:27, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Wed, 17 Feb 2016, Sven Schreiber wrote:
Am 16.02.2016 um 22:18 schrieb Allin Cottrell:
>
>> On Tue, 16 Feb 2016, Clive Nicholas wrote:
>>
>> Actually, I'm on 2016a-git and I would very much appreciate it if we were
>>> told via the list when a new (mini-)version of -gretl- has been rolled
>>> out.
>>> I never hear anything about them.
>>>
>>
>> That's what gretl-announce is for. See
>> http://lists.wfu.edu/mailman/listinfo/gretl-announce
>>
>>
> Of course you're right in principle, but I tend to agree with Clive that
> a copying the announcement on gretl-users wouldn't hurt. It seems to me
> that gretl-announce is for those that don't want to receive any other news.
>
Fair enough. We could send a short message with a link to the ChangeLog to
gretl-users.
Allin
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
Clive Nicholas
"My colleagues in the social sciences talk a great deal about methodology.
I prefer to call it style." -- Freeman J. Dyson
7:07 p.m.
On Tue, 16 Feb 2016, Summers, Peter wrote:
Clive,
The arguments to seq() all have to be integers. Try replacing that line
with “matrix x = seq(-30,32,2)./10”
Yes, this is true. However, in the development version (which is what
Artur is using) we relaxed the "integer" constraint and the code he posted
is perfectly legal. That'll be part of grel 2016b.
-------------------------------------------------------
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
-------------------------------------------------------
8:03 p.m.
I recognized this, but wasn't aware it's that new. However, when one runs the
linspace() in Matlab in this way, there is no need to set the upper-value to 3.2 but just
3 - which seems to be more natural to me - to obtain an evenly spaced vector from -3 to
3.
Artur
Am 16. Februar 2016 20:07:43 MEZ, schrieb "Riccardo (Jack) Lucchetti"
<r.lucchetti(a)univpm.it>:
On Tue, 16 Feb 2016, Summers, Peter wrote:
> Clive,
>
> The arguments to seq() all have to be integers. Try replacing that
line
> with “matrix x = seq(-30,32,2)./10”
Yes, this is true. However, in the development version (which is what
Artur is using) we relaxed the "integer" constraint and the code he
posted
is perfectly legal. That'll be part of grel 2016b.
-------------------------------------------------------
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
--
Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail gesendet.
2:27 a.m.
Thanks Peter, that sorted it. Do remember that this wasn't my code, so how
Artur's original code worked for him I've no clue.
C
On 16 February 2016 at 18:50, Summers, Peter <psummers(a)highpoint.edu> wrote:
Clive,
The arguments to seq() all have to be integers. Try replacing that line
with “matrix x = seq(-30,32,2)./10”
PS
*From:* gretl-users-bounces(a)lists.wfu.edu [mailto:
gretl-users-bounces(a)lists.wfu.edu] *On Behalf Of *Clive Nicholas
*Sent:* Tuesday, February 16, 2016 1:27 PM
*To:* r.lucchetti(a)univpm.it; Gretl list
*Subject:* Re: [Gretl-users] PDF of a multivariate normal
Jack,
Artur's code chokes for me when I run your function code:
gretl version 2016a-git
Current session: 2016-02-16 18:17
Invalid argument
Error executing script: halting
> matrix x = seq(-3,3.2,0.2)
I wonder why? My -gretl-'s fully up to date!
C
On 16 February 2016 at 15:26, Riccardo (Jack) Lucchetti <
r.lucchetti(a)univpm.it> wrote:
On Tue, 16 Feb 2016, Artur T. wrote:
Ok, I changed the definition of the constant k. However, there is still
some difference to the outcome using MATLAB...
Hehe. We're convergiung to the truth: here's the correct one --- I think!
<hansl>
function matrix mnormal_pdf(matrix x, matrix m, matrix Sigma)
# x should be n x k (n k-dimensional vectors)
# m should be a k-vector of means
# Sigma should be a kxk pd matrix
scalar k = cols(x)
scalar n = rows(x)
if k != rows(m)*cols(m)
return mshape(NA, n, 1)
endif
scalar dS = det(Sigma)
if dS <= 0
return mshape(NA, n, 1)
endif
matrix U = x .- vec(m)'
matrix U = U .* (U/Sigma)
scalar k = dS * (2*$pi)^k
return exp(-0.5*sumr(U)) / sqrt(k)
end function
</hansl>
Actually I was a little surprised about Jack's suggestion also with
respect to what it does. I thought a function is wanted that also does
the RNG inside.
No, it is really just about computing the pdf.
Generating a sample of n independent Gaussian pseurdo-rvs with mean m and
covariance Sigma is rather trivial, in fact:
<hansl>
n = 20000
m = {1, 2, 3}
Sigma = {4,1,-1;1,4,1;-1,1,4}
A = mnormal(n, 3) * cholesky(Sigma)' .+ m
eval meanc(A)
eval mcov(A)
</hansl>
-------------------------------------------------------
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
--
Clive Nicholas
"My colleagues in the social sciences talk a great deal about methodology.
I prefer to call it style." -- Freeman J. Dyson
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
Clive Nicholas
"My colleagues in the social sciences talk a great deal about methodology.
I prefer to call it style." -- Freeman J. Dyson
3:05 a.m.
Clive,
Artur's version of gretl is newer than yours or mine (though I've since updated).
The integer requirements for seq() have been removed.
PS
Sent from my iPad
On Feb 16, 2016, at 9:33 PM, Clive Nicholas
<clivelists@googlemail.com<mailto:clivelists@googlemail.com>> wrote:
Thanks Peter, that sorted it. Do remember that this wasn't my code, so how Artur's
original code worked for him I've no clue.
C
On 16 February 2016 at 18:50, Summers, Peter
<psummers@highpoint.edu<mailto:psummers@highpoint.edu>> wrote:
Clive,
The arguments to seq() all have to be integers. Try replacing that line with “matrix x =
seq(-30,32,2)./10”
PS
From: gretl-users-bounces@lists.wfu.edu<mailto:gretl-users-bounces@lists.wfu.edu>
[mailto:gretl-users-bounces@lists.wfu.edu<mailto:gretl-users-bounces@lists.wfu.edu>]
On Behalf Of Clive Nicholas
Sent: Tuesday, February 16, 2016 1:27 PM
To: r.lucchetti@univpm.it<mailto:r.lucchetti@univpm.it>; Gretl list
Subject: Re: [Gretl-users] PDF of a multivariate normal
Jack,
Artur's code chokes for me when I run your function code:
gretl version 2016a-git
Current session: 2016-02-16 18:17
Invalid argument
Error executing script: halting
matrix x = seq(-3,3.2,0.2)
I wonder why? My -gretl-'s fully
up to date!
C
On 16 February 2016 at 15:26, Riccardo (Jack) Lucchetti
<r.lucchetti@univpm.it<mailto:r.lucchetti@univpm.it>> wrote:
On Tue, 16 Feb 2016, Artur T. wrote:
Ok, I changed the definition of the constant k. However, there is still
some difference to the outcome using MATLAB...
Hehe. We're convergiung to the truth: here's the correct one --- I think!
<hansl>
function matrix mnormal_pdf(matrix x, matrix m, matrix Sigma)
# x should be n x k (n k-dimensional vectors)
# m should be a k-vector of means
# Sigma should be a kxk pd matrix
scalar k = cols(x)
scalar n = rows(x)
if k != rows(m)*cols(m)
return mshape(NA, n, 1)
endif
scalar dS = det(Sigma)
if dS <= 0
return mshape(NA, n, 1)
endif
matrix U = x .- vec(m)'
matrix U = U .* (U/Sigma)
scalar k = dS * (2*$pi)^k
return exp(-0.5*sumr(U)) / sqrt(k)
end function
</hansl>
Actually I was a little surprised about Jack's suggestion also with
respect to what it does. I thought a function is wanted that also does
the RNG inside.
No, it is really just about computing the pdf.
Generating a sample of n independent Gaussian pseurdo-rvs with mean m and covariance Sigma
is rather trivial, in fact:
<hansl>
n = 20000
m = {1, 2, 3}
Sigma = {4,1,-1;1,4,1;-1,1,4}
A = mnormal(n, 3) * cholesky(Sigma)' .+ m
eval meanc(A)
eval mcov(A)
</hansl>
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti@univpm.it<mailto:r.lucchetti@univpm.it>
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
_______________________________________________
Gretl-users mailing list
Gretl-users@lists.wfu.edu<mailto:Gretl-users@lists.wfu.edu>
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
Clive Nicholas
"My colleagues in the social sciences talk a great deal about methodology. I prefer
to call it style." -- Freeman J. Dyson
_______________________________________________
Gretl-users mailing list
Gretl-users@lists.wfu.edu<mailto:Gretl-users@lists.wfu.edu>
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
Clive Nicholas
"My colleagues in the social sciences talk a great deal about methodology. I prefer
to call it style." -- Freeman J. Dyson
_______________________________________________
Gretl-users mailing list
Gretl-users@lists.wfu.edu<mailto:Gretl-users@lists.wfu.edu>
http://lists.wfu.edu/mailman/listinfo/gretl-users
2990
days inactive
2992
days old
26 comments
7 participants
participants (7)
-
Allin Cottrell -
Artur T. -
Artur Tarassow
-
Clive Nicholas -
Riccardo (Jack) Lucchetti -
Summers, Peter -
Sven Schreiber