Skip to content

Commit

Permalink
Typos etc.
Browse files Browse the repository at this point in the history
  • Loading branch information
worc4021 committed Nov 30, 2024
1 parent ea258b0 commit e048dc2
Showing 1 changed file with 21 additions and 21 deletions.
42 changes: 21 additions & 21 deletions uno/solvers/MA27/MA27Solver.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -30,9 +30,9 @@ namespace uno


enum eICNTL {
LP = 0, // sed by the subroutines as the output stream for error messages. If it is set to zero these messages will be suppressed. The default value is 6.
MP, //used by the subroutines as the output stream for diagnostic printing and for warning messages. If it is set to zero then messages are suppressed. The default value is 6.
LDIAG, // used by the subroutines to control diagnostic printing. If ICNTL(3) is equal to zero (the default), no diagnostic printing will be produced, a value of 1 will print scalar parameters (both in argument lists and in the control and information arrays) and a few entries of array parameters on entry and successful exit from each subroutine while ICNTL(3) equal to 2 will print all parameter values on entry and successful exit.
LP = 0, // Used by the subroutines as the output stream for error messages. If it is set to zero these messages will be suppressed. The default value is 6.
MP, // Used by the subroutines as the output stream for diagnostic printing and for warning messages. If it is set to zero then messages are suppressed. The default value is 6.
LDIAG, // Used by the subroutines to control diagnostic printing. If ICNTL(3) is equal to zero (the default), no diagnostic printing will be produced, a value of 1 will print scalar parameters (both in argument lists and in the control and information arrays) and a few entries of array parameters on entry and successful exit from each subroutine while ICNTL(3) equal to 2 will print all parameter values on entry and successful exit.
/* The entries ICNTL(4) to ICNTL(25) are not of interest to the general user and are discussed more fully by Duff and Reid (AERE R-10533, 1982) under the internal names IOVFLO, NEMIN and IFRLVL
*/
IOVFLO,
Expand Down Expand Up @@ -65,29 +65,29 @@ enum eICNTL {
};

enum eCNTL {
U = 0, // used by the subroutine to control numerical pivoting. Values greater than 0.5 are treated as 0.5 and less than –0.5 as –0.5. Its default value is 0.1. If U is positive, numerical pivoting will be performed. If U is non-positive, no pivoting will be performed, the subroutine will fail if a zero pivot is encountered, and a flag (see section 2.3) will be set if not all pivots are of the same sign; the factorization will continue after a sign change is detected if U is zero but will exit immediately if U is less than zero. If the system is definite, then setting U to zero will decrease the factorization time while still providing a stable decomposition. For problems requiring greater than average numerical care a higher value than the default would be advisable.
FRATIO, // given the default value of 1.0 by MA27I/ID. If MA27A/AD encounters a row of the reduced matrix with a proportion of entries greater than FRATIO, the row is treated as full. FRATIO is not altered by MA27.
PIVTOL, // given the default value of 0.0 by MA27I/ID. MA27B/BD will not accept an entry with absolute value less than PIVTOL as a 1×1 pivot or the off-diagonal entry of a 2×2 pivot. PIVTOL is not altered by MA27.
U = 0, // Used by the subroutine to control numerical pivoting. Values greater than 0.5 are treated as 0.5 and less than –0.5 as –0.5. Its default value is 0.1. If U is positive, numerical pivoting will be performed. If U is non-positive, no pivoting will be performed, the subroutine will fail if a zero pivot is encountered, and a flag (see section 2.3) will be set if not all pivots are of the same sign; the factorization will continue after a sign change is detected if U is zero but will exit immediately if U is less than zero. If the system is definite, then setting U to zero will decrease the factorization time while still providing a stable decomposition. For problems requiring greater than average numerical care a higher value than the default would be advisable.
FRATIO, // Given the default value of 1.0 by MA27I/ID. If MA27A/AD encounters a row of the reduced matrix with a proportion of entries greater than FRATIO, the row is treated as full. FRATIO is not altered by MA27.
PIVTOL, // Given the default value of 0.0 by MA27I/ID. MA27B/BD will not accept an entry with absolute value less than PIVTOL as a 1×1 pivot or the off-diagonal entry of a 2×2 pivot. PIVTOL is not altered by MA27.
UNUSED_CNTL1,
UNUSED_CNTL2,
};

enum eINFO {
IFLAG = 0, // is an error flag. A value of zero indicates that the subroutine has performed successfully.
IERROR, // provides supplementary information when there is an error.
NRLTOT, // gives the total amount of REAL words required for a successful completion of MA27B/BD without the need for data compression provided no numerical pivoting is performed. The actual amount required may be higher because of numerical pivoting, but probably not by more than 3%.
NIRTOT, // gives the total amount of INTEGER words required for a successful completion of MA27B/BD without the need for data compression provided no numerical pivoting is performed. The actual amount required may be higher because of numerical pivoting, but probably not by more than 3%.
NRLNEC, // gives the amount of REAL words required for successful completion of MA27B/BD allowing data compression (see NCMPBR returned in INFO(12)), again provided no numerical pivoting is performed. Numerical pivoting may cause a higher value to be required, but probably not by more than 3%. If storage was conserved by equivalencing IW(1) with IRN(1), NRLNEC and NIRNEC cannot be calculated exactly but instead an upper bound will be returned. Experience has shown that this can overestimate the exact values by 50% although the tightness of the bound is very problem dependent. For example, a tight bound will generally be obtained if there are many more entries in the factors than in the input matrix.
NIRNEC, // gives the amount of INTEGER words required for successful completion of MA27B/BD allowing data compression (see NCMPBR returned in INFO(12)), again provided no numerical pivoting is performed. Numerical pivoting may cause a higher value to be required, but probably not by more than 3%. If storage was conserved by equivalencing IW(1) with IRN(1), NRLNEC and NIRNEC cannot be calculated exactly but instead an upper bound will be returned. Experience has shown that this can overestimate the exact values by 50% although the tightness of the bound is very problem dependent. For example, a tight bound will generally be obtained if there are many more entries in the factors than in the input matrix.
NRLADU, // gives the number of REAL words required to hold the matrix factors if no numerical pivoting is performed by MA27B/BD. Numerical pivoting may change this slightly.
NIRADU, // gives the number of INTEGER words required to hold the matrix factors if no numerical pivoting is performed by MA27B/BD. Numerical pivoting may change this slightly.
NRLBDU, // give the amount of REAL words actually used to hold the factorization.
NIRBDU, // give the amount of INTEGER words actually used to hold the factorization.
NCMPA, // holds the number of compresses of the internal data structure performed by MA27A/AD. If this is high (say > 10), the performance of MA27A/AD may be improved by increasing the length of array IW.
NCMPBR, // holds the number of compresses of the real data structure required by the factorization. If either of these is high (say > 10), then the speed of the factorization may be increased by allocating more space to the arrays A as appropriate.
NCMPBI, // holds the number of compresses of the integer data structure required by the factorization. If either of these is high (say > 10), then the speed of the factorization may be increased by allocating more space to the arrays IW as appropriate.
NTWO, // gives the number of 2×2 pivots used during the factorization.
NEIG, // gives the number of negative eigenvalues of A.
IFLAG = 0, // An error flag. A value of zero indicates that the subroutine has performed successfully.
IERROR, // Provides supplementary information when there is an error.
NRLTOT, // Gives the total amount of REAL words required for a successful completion of MA27B/BD without the need for data compression provided no numerical pivoting is performed. The actual amount required may be higher because of numerical pivoting, but probably not by more than 3%.
NIRTOT, // Gives the total amount of INTEGER words required for a successful completion of MA27B/BD without the need for data compression provided no numerical pivoting is performed. The actual amount required may be higher because of numerical pivoting, but probably not by more than 3%.
NRLNEC, // Gives the amount of REAL words required for successful completion of MA27B/BD allowing data compression (see NCMPBR returned in INFO(12)), again provided no numerical pivoting is performed. Numerical pivoting may cause a higher value to be required, but probably not by more than 3%. If storage was conserved by equivalencing IW(1) with IRN(1), NRLNEC and NIRNEC cannot be calculated exactly but instead an upper bound will be returned. Experience has shown that this can overestimate the exact values by 50% although the tightness of the bound is very problem dependent. For example, a tight bound will generally be obtained if there are many more entries in the factors than in the input matrix.
NIRNEC, // Gives the amount of INTEGER words required for successful completion of MA27B/BD allowing data compression (see NCMPBR returned in INFO(12)), again provided no numerical pivoting is performed. Numerical pivoting may cause a higher value to be required, but probably not by more than 3%. If storage was conserved by equivalencing IW(1) with IRN(1), NRLNEC and NIRNEC cannot be calculated exactly but instead an upper bound will be returned. Experience has shown that this can overestimate the exact values by 50% although the tightness of the bound is very problem dependent. For example, a tight bound will generally be obtained if there are many more entries in the factors than in the input matrix.
NRLADU, // Gives the number of REAL words required to hold the matrix factors if no numerical pivoting is performed by MA27B/BD. Numerical pivoting may change this slightly.
NIRADU, // Gives the number of INTEGER words required to hold the matrix factors if no numerical pivoting is performed by MA27B/BD. Numerical pivoting may change this slightly.
NRLBDU, // Gives the amount of REAL words actually used to hold the factorization.
NIRBDU, // Gives the amount of INTEGER words actually used to hold the factorization.
NCMPA, // Holds the number of compresses of the internal data structure performed by MA27A/AD. If this is high (say > 10), the performance of MA27A/AD may be improved by increasing the length of array IW.
NCMPBR, // Holds the number of compresses of the real data structure required by the factorization. If either of these is high (say > 10), then the speed of the factorization may be increased by allocating more space to the arrays A as appropriate.
NCMPBI, // Holds the number of compresses of the integer data structure required by the factorization. If either of these is high (say > 10), then the speed of the factorization may be increased by allocating more space to the arrays IW as appropriate.
NTWO, // Gives the number of 2×2 pivots used during the factorization.
NEIG, // Gives the number of negative eigenvalues of A.
UNUSED_INFO1,
UNUSED_INFO2,
UNUSED_INFO3,
Expand Down

0 comments on commit e048dc2

Please sign in to comment.