File: find_invgampar.m

package info (click to toggle)
dynare 4.3.0-2
  • links: PTS, VCS
  • area: main
  • in suites: wheezy
  • size: 40,640 kB
  • sloc: fortran: 82,231; cpp: 72,734; ansic: 28,874; pascal: 13,241; sh: 4,300; objc: 3,281; yacc: 2,833; makefile: 1,288; lex: 1,162; python: 162; lisp: 54; xml: 8
file content (57 lines) | stat: -rw-r--r-- 2,641 bytes parent folder | download | duplicates (8)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
function [a, b, XLO, XUP] = find_invgampar(XLO, XUP, PLO, PUP, a0, b0);

% This function takes as inputs the bounds [XLO, XUP] in the support of the
% Inverse Gamma distribution (with parameters a and b), the probabilities of the
% bounds [PLO, PUP], and the initial values for ab=[a0, b0]
% and returns the estimates of a and b (as well as XLO and XUP)
% by solving the non-linear functions in invgampar(ab, XLO, XUP, PLO, PUP).\

%-----------------------------------------------------------------------------------
%------------------------ Inverse-Gamma distribution ------------------------------%
%--- p(x) = ( b^a/Gamma(a) ) x^(-a-1) exp(-b/x) for a>0 and b>0.
%---    where a is shape and b is scale parameter.
%--- E(x) = b/(a-1) for a>1;  var(x) = b^2/( (a-1)^2*(a-2) ) for a>2;
%--- Noninformative distribution: a,b -> 0.
%--- How to draw: (1) draw z from Gamma(a,b); (2) let x=1/z.
%-----------------------------------------------------------------------------------
%
% Copyright (C) 1997-2012 Tao Zha
%
% This free software: you can redistribute it and/or modify
% it under the terms of the GNU General Public License as published by
% the Free Software Foundation, either version 3 of the License, or
% (at your option) any later version.
%
% It is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
% GNU General Public License for more details.
%
% If you did not received a copy of the GNU General Public License
% with this software, see <http://www.gnu.org/licenses/>.
%

if XLO >= XUP;
    error('the lower bound needs to be smaller than the upper bound')
elseif XLO<= 0;
    error('the support for Gamma distribution needs to be non-negative');
end;

if a0 <= 0 || b0 <= 0;
    error('the values for a and b need to be positive');
end;

disp(' ')
disp('*************** Convergence results for inverse gamma density ***************')
options = optimset('Display', 'on','TolFun', 1.0e-10, 'TolX', 1.0e-10);
ab_values = fsolve('invgampar', [a0, b0], options, XLO, XUP, PLO, PUP);
a = ab_values(1); b = ab_values(2);

% Alternatively, it is possible to constrain the search for the values of a
% and b in the positive range (with 0 being the explicit lower bound) by
% using the lsqnonlin function (see below) instead of the fsolve.  The
% tradeoff is that lsqnonlin is typically slower than fsolve.
% LB_a = 0; LB_b = 0; UB_a = Inf; UB_b = Inf;
% ab_values = lsqnonlin('invgampar', [a0, b0], [LB_a, LB_b], [UB_a, UB_b],...
%     options, XLO, XUP, PLO, PUP);
% a = ab_values(1); b = ab_values(2);