---- Autoplot Help ----

EN[D] or QU[IT] or ST[OP] Terminate session.

---------------------------------------- Photon Help ----

E[nd]....terminates the plotting session, closes all open files and returns to the operating system.

See also : STOP

-------------------- End of S Photon Help ----

[End of Segment] closes the current GEOMETRY segment and goes one level up.

----------- Advanced PIL command --- -

ENDCASE is the final element of the PIL CASE construct. See the HELP entry on CASE for further information.

See DISPLAY

------------- Advanced PIL command --- -

ENDDO denotes the end of a PIL DO-loop. See the HELP entry on DO for further information.

------------- Advanced PIL command --- -

ENDIF is the final element of the PIL IF construct. See the HELP entry on IF for further information.

----- PIL real array; default= 1.E-3; -

When greater than zero, ENDIT influences the termination of iterations in the built-in linear-equation solver by requiring that, before termination can occur, either

- the absolute sum of corrections is less than ENDIT times the absolute sum of corrections on the first iteration, or
- for SELREF=T, the (sum of errors)/ RESREF is less than ENDIT* RESFAC, or
- for SELREF=F, the (sum of errors)/RESREF is less than 1.0.

unless LITER iterations have been performed.

When less than zero (but not = GRND1), ENDIT represents the negative of the over-relaxation factor to be used for the variable in question, in the built-in linear-equation solver, and has no other effect.

See PHENC entry: over-relaxation.

When ENDIT = GRND1, the conjugate-gradient solver is called instead of the built-in (Stone-type) solver for the PHI in question, whereas the setting CSG3=CNGR causes that solver to be called for all variables.

The conjugate-gradient solver ends its iterations by reference to RESREF, RESFAC and LITER in the same way as the Stone solver, with 0.001 replacing the ENDIT multiplier.

Whenever ENDIT is negative, the multipliers referred to in the first paragraph are taken as 1.0E-3.

Information on how the linear equation solver is performing can be elicited by adding the line

SPEDAT(SET,SELREF,PRNSOL,L,T)

to Group 19 of Q1. For each variable on each sweep the solver will write a line to RESULT stating how many iterations were performed, and giving an indication of why iteration stopped. A typical extract of the printout is shown here:

Done LU-CR solver for P1 isweep 73 IZ 20 iter 12 2 8.168808E-08 1.364782E+00
6.493849E-04

Done STONE solver for U1 isweep 73 IZ 20 iter 1 2 1.028983E-12 1.673497E-01
1.673497E-04

Done STONE solver for V1 isweep 73 IZ 20 iter 1 2 1.391361E-12 2.119821E-01
2.119821E-04

Done STONE solver for W1 isweep 73 IZ 20 iter 1 2 2.556446E-12 3.189903E-01
3.189903E-04

The next number, in this case 2, is an exit code. The exit codes for the various solvers are as follows:

1. Stone Solver

- 0 - all LITER iterations were performed
- 1 - solver terminated because the sum of corrections (second REAL number) is less than ENDIT* the sum of corrections on the first iteration (third REAL number)
- 2 - solver terminated because the sum of normalised residuals (first REAL number) is below ENDIT*RESFAC
- 3 -solver terminated because the sum of residuals is greater than 10* the sum of residuals on the previous iteration. This indicates divergence rather than convergence.

2. Conjugate-Gradient (LU-CR) Solver

- 0 - all LITER iterations were performed
- 1 - solver terminated because the sum of corrections (second REAL number) is less than ENDIT* the sum of corrections on the first iteration (third REAL number)
- 2 - solver terminated because the sum of normalised residuals (first REAL number) is below ENDIT*RESFAC.
- 3 - solver terminated because sum of corrections greater than sum of corrections on previous iteration and less than 0.1* sum of corrections on first iteration.

3. Sequential GCV Solver

- 0 - all LITER iterations were performed
- 1 - not used
- 2 - solver terminated because the sum of normalised residuals (first REAL number) is below ENDIT*RES1
- 2 - solver terminated because the sum of normalised residuals (first REAL number) is below ENDIT*RESFAC
- 3 - for LSG1=F (default), solver terminated because sum of residuals (second REAL number) is greater than the sum of residuals on previous iteration and is less than 0.1* the sum of residuals on the first iteration.
- 3 - for LSG1=T and P1, solver terminated because sum of normalised residuals (first REAL number) less than 0.1*sum of normalised residuals on first iteration (third REAL number).
- 4 - for LSG1=T and other variables, solver terminated because sum of normalised residuals less than 0.01*sum of normalised residuals on the first iteration.
- 5 - solver terminated because the denominator is zero

4. Parallel Solver

- 0 - all LITER iterations were performed
- 1 - solver terminated because the sum of corrections (second REAL number) is less than ENDIT* the sum of corrections on the first iteration (third REAL number)
- 2 - solver terminated because the sum of normalised residuals (first REAL number) is below ENDIT*RESFAC
- 3 - solver terminated because sum of corrections greater than sum of corrections on previous iteration and less than 0.1* sum of corrections on first iteration.
- 4 - solver terminated because denominator is zero

5. Parallel GCV Solver (Par-GCV)

- 0 - all LITER iterations were performed
- 3 - solver terminated because sum of corrections greater than sum of corrections on previous iteration and less than 0.1* sum of corrections on first iteration.
- 4 - Not used
- 5 - solver terminated because denominator is zero
- 6 - solver terminated because sum of residuals / (ENDIT*sum of residuals on entry) is less than 1
- 7 - solver terminated because sum of corrections / (ENDIT*sum of residuals on entry) is less than 1

----------- Advanced PIL command --- -

ENDLOOP marks the end of the a PIL loop. See the HELP entry on LOOP for further information, or inspect case b542 for an example.

----------- Advanced PIL command --- -

ENDMAIN marks the end of the 'main' PIL program. It causes the satellite to stop executing the Q1 file. All subroutines must follow the ENDMAIN statement in the Q1 file.

------------ Advanced PIL command --- -

ENDSUB denotes the end of a PIL subroutine. See the HELP entry on SUBROUTINE for further information.

wbs