havBpNet:J 

All Packages  Class Hierarchy  This Package  Previous  Next  Index

havBpNet:J Home Page

Class havNN.havSystem.havSys 

java.lang.Object
   |
   +----havNN.havSystem.havSys
public final class havSys
extends Object
The havBpsys class contains various enumerations of special types or general utility functions used in various hav.Software products.

Version:
1.1(java) 1/1/98

Variable Index

o havCUBE
Value = 3.
o havFALSE
Value = 0.
o havHERMITE
Value = 4.
o havHIDDEN
Value = 1.
o havHTAN
Value = 1.
o havINPUT
Value = 0.
o havLINEAR
Value = 2.
o havOFF
Value = 1.
o havON
Value = 0.
o havOUTPUT
Value = 2.
o havQUAD
Value = 4.
o havSIGMOID
Value = 0.
o havSIN
Value = 3.
o havSQUARED
Value = 2.
o havTRUE
Value = 1.

Constructor Index

o havSys()

Method Index

o havexp(double)
General utility used for data and response UN-scaling - assumes f = 1.
o havexp(double, double)
General utility used for data and response UN-scaling.
o havlog(double)
General utility used to for data and response scalling - assumes f1.
o havlog(double, double)
General utility used for data and response scalling.
o mk1d(long, int[], int[])
This general utility function will calculate a 1d index value from an Nd specification.
o mkNd(int, int, int[], int[])
This utility function will calculate an Nd index from a 1d index specification.
o readDouble(StreamTokenizer)
General utility for reading a Double value from a file.
o readInt(StreamTokenizer)
General utility for reading an Integer value from a file.
o sign_of(double)
This utility function will return the sign of the input as +/- 1 or 0 if the input is 0.

Variables

o havFALSE 
 public static final int havFALSE
Value = 0.

o havTRUE 
 public static final int havTRUE
Value = 1.

o havON 
 public static final int havON
Value = 0.

o havOFF 
 public static final int havOFF
Value = 1.

o havINPUT 
 public static final int havINPUT
Value = 0.

o havHIDDEN 
 public static final int havHIDDEN
Value = 1.

o havOUTPUT 
 public static final int havOUTPUT
Value = 2.

o havSIGMOID 
 public static final int havSIGMOID
Value = 0.

o havHTAN 
 public static final int havHTAN
Value = 1.

o havLINEAR 
 public static final int havLINEAR
Value = 2.

o havSIN 
 public static final int havSIN
Value = 3.

o havHERMITE 
 public static final int havHERMITE
Value = 4.

o havSQUARED 
 public static final int havSQUARED
Value = 2.

o havCUBE 
 public static final int havCUBE
Value = 3.

o havQUAD 
 public static final int havQUAD
Value = 4.

Constructors

o havSys 
 public havSys()

Methods

o havlog 
 public static double havlog(double x,
                             double f)
General utility used for data and response scalling. Full implementation of the so-called mu-law which is useful when the input- and expected-data values are both positive and negative.

The basic formula for havlog is ...

x' = sign_of(x) * log( (|x|/f) + 1.0)

Parameters:
x - The number to be scaled
f - The OPTIONAL shaping factor - will affect the steepness of the resulting curve.
o havlog 
 public static double havlog(double x)
General utility used to for data and response scalling - assumes f = 1. Full implementation of the so-called mu-law which is useful when the input- and expected-data values are both positive and negative.

The basic formula for havlog is ...

x' = sign_of(x) * log( (|x|/1.0) + 1.0)

Parameters:
x - The number to be scaled
o havexp 
 public static double havexp(double x,
                             double f)
General utility used for data and response UN-scaling. Full implementation of the so-called mu-law which is useful when the input- and expected-data values are both positive and negative.

The basic formula for havexp to un-scale a value, x', that had previously been scaled using havlog is ...

x = sign_of(x') * (exp(|x'|) - 1.0) * f

Parameters:
x - The number to be un-scaled
f - The OPTIONAL shaping factor - will affect the steepness of the resulting curve.
o havexp 
 public static double havexp(double x)
General utility used for data and response UN-scaling - assumes f = 1. Full implementation of the so-called mu-law which is useful when the input- and expected-data values are both positive and negative.

The basic formula for havexp to un-scale a value, x', that had previously been scaled using havlog is ...

x = sign_of(x') * (exp(|x'|) - 1.0) * 1.0

Parameters:
x - The number to be un-scaled
o mk1d 
 public static int mk1d(long numdims,
                        int dims[],
                        int in[])
This general utility function will calculate a 1d index value from an Nd specification.

Assumptions: It is assumed that the lower bounds of all dimensions are 0. Examples: 

 Imagine that we have a vector of data defined as double a[70] ... but suppose that 
 we view it logically as a 3d array specified as ...
 

       a[7,5,2]   i.e.   0   1   2   3   4   5   6       35  36  37  38  39  40  41
                         7   8   9  10  11  12  13       42  43  44  45  46  47  48
                        14  15  16  17  18  19  20       49  50  51  52  53  54  55
                        21  22  23  24  25  26  27       56  57  58  58  60  61  62 
                        28  29  30  31  32  33  34       63  64  65  66  67  68  69
 

                           a(0,0,0) to a(6,4,0)             a(0,0,1) to a(6,4,1)
 

  ...to get the 1d index of the logical position (3,2,1) - element 52 above - we can use ...
 

       long dims[] = {7, 5, 2};  // max dims of the logical array
       long in[3];
          . . .
       in[0] = 3;  // identify the logical position for which we
       in[1] = 2;  // want the 1d index
       in[2] = 1;
         . . .
      1d_index = mk1d(3, dims, in);  // get the 1d index into a 
                                     // of the logical index (3,4,2)
 

      Note that the 1d index returned in the example above is...
 

         1d_index = 3 + (2*7) + (1*7*5)
                  = 3 + 14 + 35 
                  = 52

Parameters:
numdims - The number of indicies (i.e. N)
dims - A list of the size of each of the N dimensions
in - The Nd index specification to be converted to a 1d index
o mkNd 
 public static void mkNd(int index1D,
                         int numdims,
                         int dims[],
                         int indexND[])
This utility function will calculate an Nd index from a 1d index specification. This is the inverse of the mk1D utility function.

Parameters:
index1D - is the 1D index to be converted
numdims - is the number of demensions for the resulting Nd index
dims - is a dimension vector with (numdims) elements - these are the upper limit for dimensions - lower limits for all dimensions are assumed to be 0
indexND - is a caller supplied vector with at least (numdims) elements into which the resulting Nd index will be placed for return to the caller.
o sign_of 
 public static double sign_of(double x)
This utility function will return the sign of the input as +/- 1 or 0 if the input is 0. Maintained for compatability with c++ version.

o readInt 
 public static int readInt(StreamTokenizer fileIn) throws IOException
General utility for reading an Integer value from a file.

Throws: IOException
Exception is thrown if any IO Error occurs.
o readDouble 
 public static double readDouble(StreamTokenizer fileIn) throws IOException
General utility for reading a Double value from a file.

Throws: IOException
Exception is thrown if any IO Error occurs. 
All Packages  Class Hierarchy  This Package  Previous  Next  Index

Copyright © 1998 by hav.Software. All Rights Reserved.