Add tests for cgfam and lbfgsb. Fix lbfgsb example

nonlinear.pd

@@ -1505,8 +1505,8 @@ It uses the limited memory BFGS method.
 	# The SIAM 100-Digit Challenge problem 4
 	# see http://www-m8.ma.tum.de/m3/bornemann/challengebook/
 	# result: -3.30686864747523728007611377089851565716648236
+	use PDL::NiceSlice;
 	use PDL::Opt::NonLinear;
-	use PDL::Stat::Distributions;
 
 	$x = (random(2,500)-0.5)*2;
 	$gx = zeroes(2,500);

t/1.t

@@ -2,6 +2,7 @@ use strict;
 use warnings;
 use PDL::LiteF;
 use PDL::Opt::NonLinear;
+use PDL::NiceSlice;
 use Test::More;
 
 sub approx_ok {
@@ -26,21 +27,21 @@ my $gtol = pdl(1e-8);
 #$stepmx = pdl(0.5);
 my $maxit = pdl(long, 50);
 sub min_func{
-	my ($fx, $x) = @_;
-	$fx .= rosen($x);
+    my ($fx, $x) = @_;
+    $fx .= rosen($x);
 }
 sub grad_func{
-	my ($gx, $x) = @_;
-	$gx .= rosen_grad($x);
+    my ($gx, $x) = @_;
+    $gx .= rosen_grad($x);
 }
 sub hess_func{
-	my ($hx, $x) = @_;
-	$hx .= rosen_hess($x);
+    my ($hx, $x) = @_;
+    $hx .= rosen_hess($x);
 }
-tensoropt($fx, $gx, $hx, $x, 
-	  1,$maxit,15,1,2,1,
-	  ones(5),0.5,$xtol,$gtol,2,6,
-	  \&min_func, \&grad_func, \&hess_func);
+tensoropt($fx, $gx, $hx, $x,
+      1,$maxit,15,1,2,1,
+      ones(5),0.5,$xtol,$gtol,2,6,
+      \&min_func, \&grad_func, \&hess_func);
 
 approx_ok $x,$res,0.001,'tensoropt';
 
@@ -70,4 +71,41 @@ lbfgs($fx, $gx, $x, $diag, $diagco, $m, $maxit, $maxfc, $eps, $xtol, $gtol,
                        $print,$info,\&fg_func,\&fdiag);
 approx_ok $x,$res,0.0001,'lbfgs';
 
+$x = pdl '[0.49823058 0.98093641 0.63151156 0.66477157 0.60801367]';
+$gx = zeroes(5);
+$fx = pdl(0);
+
+my $bounds =  zeroes(5,2);
+$bounds(,0).= -5;
+$bounds(,1).= 5;
+my $tbounds = zeroes(5);
+$tbounds .= 2;
+$gtol = pdl(0.9);
+my $pgtol = pdl(1e-10);
+my $factr = pdl(100);
+$print = pdl(long, [0,0]);
+$maxit = pdl(long,100);
+$info = pdl(long,0);
+$m = pdl(long,10);
+my $iv = zeroes(long,44);
+my $v = zeroes(29);
+sub fg_func{
+	my ($f, $g, $x) = @_;
+	$f .= rosen($x);
+	$g .= rosen_grad($x);
+	return 0;
+}
+lbfgsb($fx, $gx, $x, $m, $bounds, $tbounds, $maxit, $factr, $pgtol, $gtol,
+	$print, $info,$iv, $v,\&fg_func);
+approx_ok $x,$res,0.0001,'lbfgsb';
+
+$x = pdl '[0.49823058 0.98093641 0.63151156 0.66477157 0.60801367]';
+$gx = rosen_grad($x);
+$fx = rosen($x);
+$print = zeroes(2);
+$maxit = pdl(long, 200);
+$info = pdl(long,0);
+cgfam($fx, $gx, $x, $maxit, $eps, $xtol, $gtol,$print,$info,1,\&fg_func);
+approx_ok $x,$res,0.0001,'cgfam';
+
 done_testing;