e04jy is an easy-to-use quasi-Newton algorithm for finding a minimum of a function Fx1,x2,,xn, subject to fixed upper and lower bounds of the independent variables x1,x2,,xn, using function values only.
It is intended for functions which are continuous and which have continuous first and second derivatives (although it will usually work even if the derivatives have occasional discontinuities).

Syntax

C#
public static void e04jy(
	int n,
	int ibound,
	E04..::..E04JY_FUNCT1 funct1,
	double[] bl,
	double[] bu,
	double[] x,
	out double f,
	out int nftotl,
	int[] bndinfo,
	out int freevar,
	double[] fdiff,
	out double phm,
	out int ifail
)
Visual Basic
Public Shared Sub e04jy ( _
	n As Integer, _
	ibound As Integer, _
	funct1 As E04..::..E04JY_FUNCT1, _
	bl As Double(), _
	bu As Double(), _
	x As Double(), _
	<OutAttribute> ByRef f As Double, _
	<OutAttribute> ByRef nftotl As Integer, _
	bndinfo As Integer(), _
	<OutAttribute> ByRef freevar As Integer, _
	fdiff As Double(), _
	<OutAttribute> ByRef phm As Double, _
	<OutAttribute> ByRef ifail As Integer _
)
Visual C++
public:
static void e04jy(
	int n, 
	int ibound, 
	E04..::..E04JY_FUNCT1^ funct1, 
	array<double>^ bl, 
	array<double>^ bu, 
	array<double>^ x, 
	[OutAttribute] double% f, 
	[OutAttribute] int% nftotl, 
	array<int>^ bndinfo, 
	[OutAttribute] int% freevar, 
	array<double>^ fdiff, 
	[OutAttribute] double% phm, 
	[OutAttribute] int% ifail
)
F#
static member e04jy : 
        n : int * 
        ibound : int * 
        funct1 : E04..::..E04JY_FUNCT1 * 
        bl : float[] * 
        bu : float[] * 
        x : float[] * 
        f : float byref * 
        nftotl : int byref * 
        bndinfo : int[] * 
        freevar : int byref * 
        fdiff : float[] * 
        phm : float byref * 
        ifail : int byref -> unit 

Parameters

n
Type: System..::..Int32
On entry: the number n of independent variables.
Constraint: n1.
ibound
Type: System..::..Int32
On entry: indicates whether the facility for dealing with bounds of special forms is to be used.
It must be set to one of the following values:
ibound=0
If you are supplying all the lj and uj individually.
ibound=1
If there are no bounds on any xj.
ibound=2
If all the bounds are of the form 0xj.
ibound=3
If l1=l2==ln and u1=u2==un.
funct1
Type: NagLibrary..::..E04..::..E04JY_FUNCT1
You must supply funct1 to calculate the value of the function Fx at any point x. It should be tested separately before being used with e04jy (see the E04 class).

A delegate of type E04JY_FUNCT1.

bl
Type: array<System..::..Double>[]()[][]
An array of size [n]
On entry: the lower bounds lj.
If ibound is set to 0, you must set bl[j-1] to lj, for j=1,2,,n. (If a lower bound is not specified for a particular xj, the corresponding bl[j-1] should be set to -106.)
If ibound is set to 3, you must set bl[0] to l1; e04jy will then set the remaining elements of bl equal to bl[0].
On exit: the lower bounds actually used by e04jy.
bu
Type: array<System..::..Double>[]()[][]
An array of size [n]
On entry: the upper bounds uj.
If ibound is set to 0, you must set bu[j-1] to uj, for j=1,2,,n. (If an upper bound is not specified for a particular xj, the corresponding bu[j-1] should be set to 106.)
If ibound is set to 3, you must set bu[0] to u1; e04jy will then set the remaining elements of bu equal to bu[0].
On exit: the upper bounds actually used by e04jy.
x
Type: array<System..::..Double>[]()[][]
An array of size [n]
On entry: x[j-1] must be set to an estimate of the jth component of the position of the minimum, for j=1,2,,n.
On exit: the lowest point found during the calculations. Thus, if ifail=0 on exit, x[j-1] is the jth component of the position of the minimum.
f
Type: System..::..Double%
On exit: the value of Fx corresponding to the final point stored in x.
nftotl
Type: System..::..Int32%
On exit: this is used to return the total number of function evaluations.
bndinfo
Type: array<System..::..Int32>[]()[][]
An array of size [n]
On exit: if ifail=03 or 5, bndinfo contain information about which variables are currently on their bounds and which are free. Specifically, if xi is:
fixed on its upper bound, bndinfo[i-1] is -1;
fixed on its lower bound, bndinfo[i-1] is -2;
effectively a constant (i.e., lj=uj), bndinfo[i-1] is -3;
free, bndinfo[i-1] gives its position in the sequence of free variables.
freevar
Type: System..::..Int32%
On exit:
contains the number of free variables (i.e., nz).
fdiff
Type: array<System..::..Double>[]()[][]
An array of size [n]
On exit: if ifail=03 or 5, fdiff[i-1] contains a finite-difference approximation to the ith element of the projected gradient vector gz, for i=1,2,,n 
phm
Type: System..::..Double%
On exit: contains an estimate of the condition number of the projected Hessian matrix (i.e., k)
ifail
Type: System..::..Int32%
On exit: ifail=0 unless the method detects an error or a warning has been flagged (see [Error Indicators and Warnings]).

Description

e04jy is applicable to problems of the form:
MinimizeFx1,x2,,xn  subject to  ljxjuj,  j=1,2,,n
when derivatives of Fx are unavailable.
Special provision is made for problems which actually have no bounds on the xj, problems which have only non-negativity bounds and problems in which l1=l2==ln and u1=u2==un. You must supply a method to calculate the value of Fx at any point x.
From a starting point you supplied there is generated, on the basis of estimates of the gradient and the curvature of Fx, a sequence of feasible points which is intended to converge to a local minimum of the constrained function. An attempt is made to verify that the final point is a minimum.
A typical iteration starts at the current point x where nz (say) variables are free from both their bounds. The projected gradient vector gz, whose elements are finite difference approximations to the derivatives of Fx with respect to the free variables, is known. A unit lower triangular matrix L and a diagonal matrix D (both of dimension nz), such that LDLT is a positive definite approximation of the matrix of second derivatives with respect to the free variables (i.e., the projected Hessian) are also held. The equations
LDLTpz=-gz
are solved to give a search direction pz, which is expanded to an n-vector p by an insertion of appropriate zero elements. Then α is found such that Fx+αp is approximately a minimum (subject to the fixed bounds) with respect to α; x is replaced by x+αp, and the matrices L and D are updated so as to be consistent with the change produced in the estimated gradient by the step αp. If any variable actually reaches a bound during the search along p, it is fixed and nz is reduced for the next iteration. Most iterations calculate gz using forward differences, but central differences are used when they seem necessary.
There are two sets of convergence criteria – a weaker and a stronger. Whenever the weaker criteria are satisfied, the Lagrange multipliers are estimated for all the active constraints. If any Lagrange multiplier estimate is significantly negative, then one of the variables associated with a negative Lagrange multiplier estimate is released from its bound and the next search direction is computed in the extended subspace (i.e., nz is increased). Otherwise minimization continues in the current subspace provided that this is practicable. When it is not, or when the stronger convergence criteria are already satisfied, then, if one or more Lagrange multiplier estimates are close to zero, a slight perturbation is made in the values of the corresponding variables in turn until a lower function value is obtained. The normal algorithm is then resumed from the perturbed point.
If a saddle point is suspected, a local search is carried out with a view to moving away from the saddle point. A local search is also performed when a point is found which is thought to be a constrained minimum.

References

Gill P E and Murray W (1976) Minimization subject to bounds on the variables NPL Report NAC 72 National Physical Laboratory

Error Indicators and Warnings

Note: e04jy may return useful information for one or more of the following detected errors or warnings.
Errors or warnings detected by the method:
Some error messages may refer to parameters that are dropped from this interface (IW, LIW, W, LW) In these cases, an error in another parameter has usually caused an incorrect value to be inferred.
ifail=1
On entry,n<1,
oribound<0,
oribound>3,
oribound=0 and bl[j-1]>bu[j-1] for some j,
oribound=3 and bl[0]>bu[0],
ifail=2
There have been 400×n function evaluations, yet the algorithm does not seem to be converging. The calculations can be restarted from the final point held in x. The error may also indicate that Fx has no minimum.
ifail=3
The conditions for a minimum have not all been met but a lower point could not be found and the algorithm has failed.
ifail=4
An overflow has occurred during the computation. This is an unlikely failure, but if it occurs you should restart at the latest point given in x.
ifail=5
ifail=6
ifail=7
ifail=8
There is some doubt about whether the point x found by e04jy is a minimum. The degree of confidence in the result decreases as ifail increases. Thus, when ifail=5 it is probable that the final x gives a good estimate of the position of a minimum, but when ifail=8 it is very unlikely that the method has found a minimum.
ifail=9
In the search for a minimum, the modulus of one of the variables has become very large 106. This indicates that there is a mistake in funct1, that your problem has no finite solution, or that the problem needs rescaling (see [Further Comments]).
ifail=10
The computed set of forward-difference intervals (stored in w[9×n],w[9×n+1],,w[10×n-1]) is such that x[i-1]+w[9×n+i-1]x[i-1] for some i.
This is an unlikely failure, but if it occurs you should attempt to select another starting point.
ifail=-9000
An error occured, see message report.
ifail=-8000
Negative dimension for array value
ifail=-6000
Invalid Parameters value
ifail= -4000
Invalid dimension for array value
If you are dissatisfied with the result (e.g., because ifail=567 or 8), it is worth restarting the calculations from a different starting point (not the point at which the failure occurred) in order to avoid the region which caused the failure. If persistent trouble occurs and the gradient can be calculated, it may be advisable to change to a method which uses gradients (see the E04 class).

Accuracy

A successful exit (ifail=0) is made from e04jy when (B1, B2 and B3) or B4 hold, and the local search confirms a minimum, where
  • B1αk×pk<xtol+ε×1.0+xk
  • B2Fk-Fk-1<xtol2+ε×1.0+Fk
  • B3gzk<ε1/3+xtol×1.0+Fk
  • B4gzk<0.01×ε.
(Quantities with superscript k are the values at the kth iteration of the quantities mentioned in [Description], xtol=100ε, ε is the machine precision and . denotes the Euclidean norm. The vector gz is returned in the array w.)
If ifail=0, then the vector in x on exit, xsol, is almost certainly an estimate of the position of the minimum, xtrue, to the accuracy specified by xtol.
If ifail=3 or 5, xsol may still be a good estimate of xtrue, but the following checks should be made. Let k denote an estimate of the condition number of the projected Hessian matrix at xsol. (The value of k is returned in w[n]). If
(i) the sequence Fxk converges to Fxsol at a superlinear or a fast linear rate,
(ii) gzxxol2<10.0×ε, and
(iii) k<1.0/gzxsol,
then it is almost certain that xsol is a close approximation to the position of a minimum. When (ii) is true, then usually Fxsol is a close approximation to Fxtrue.
When a successful exit is made then, for a computer with a mantissa of t decimals, one would expect to get about t/2-1 decimals accuracy in x and about t-1 decimals accuracy in F, provided the problem is reasonably well scaled.

Parallelism and Performance

None.

Further Comments

The number of iterations required depends on the number of variables, the behaviour of Fx and the distance of the starting point from the solution. The number of operations performed in an iteration of e04jy is roughly proportional to n2. In addition, each iteration makes at least m+1 calls of funct1, where m is the number of variables not fixed on bounds. So, unless Fx can be evaluated very quickly, the run time will be dominated by the time spent in funct1.
Ideally the problem should be scaled so that at the solution the value of Fx and the corresponding values of x1,x2,,xn are each in the range -1,+1, and so that at points a unit distance away from the solution, F is approximately a unit value greater than at the minimum. It is unlikely that you will be able to follow these recommendations very closely, but it is worth trying (by guesswork), as sensible scaling will reduce the difficulty of the minimization problem, so that e04jy will take less computer time.

Example

To minimize
F=x1+10x22+5x3-x42+x2-2x34+10x1-x44
subject to
-1x13,-2x20,-1x43,
starting from the initial guess 3,-1,0,1.

Example program (C#): e04jye.cs

Example program results: e04jye.r

See Also