DUMP and LOAD files

These files are similar to PUNCH  and INSERT  files, but they record solution information in a manner that is more direct and more easily modified.  In particular, no distinction is made between columns and slacks.  Apart from the first and last line, each entry has the form

 

Columns

2-3

5-12

25-36

Contents

Key

Name

Value

 

as illustrated in Figure 3.  The keys LL, UL, BS and SB mean Lower Limit, Upper Limit, Basic and Superbasic respectively.

Notes on DUMP data

  1. A line is output for every variable, columns followed by slacks.

  2. Nonbasic variables between their bounds will be output with key LL and their current value.

Notes on LOAD data

  1. Before a LOAD file is read, all columns and slacks are made nonbasic at their smallest bound in absolute magnitude.  The basis is initially empty.

  2. BS causes Name  to become basic.

  3. SB causes Name  to become superbasic at the specified Value.

  4. LL or UL cause Name to be nonbasic at the specified Value.

  5. An entry will be ignored if Name  is already basic or superbasic.  (Thus, only the first BS or SB line takes effect for any given Name.)

  6. An SB line will not alter the status of Name  if the Superbasics limit has been reached, but the associated Value  will be retained if Name  is a Jacobian variable.

  7. (Partial basis) Let m be the number of rows in the problem.  If fewer than m variables are specified to be basic, a tentative basis list will be constructed by adding the requisite number of slacks, starting from the first row and taking those that were not previously specified to be basic or superbasic.  (If the resulting basis proves to be singular, the basis factorization routine will replace a number of basic variables by other slacks.)

  8. (Too many basics) If m variables have already been specified as basic, any further BS keys will be treated as though they were SB.  This feature may be useful for combining solutions to smaller problems.