strange error

From Michael_Looby@sandwich.pfizer.com Fri Oct 4 03:14:19 1996 Subject: strange error I have encountered a strange error while checking the nmtran pred code listed below. When I run this code (without the estimation step) on a VMS Alpha, I get the following NONMEM error: PROGRAM TERMINATED BY OBJ, ERROR IN ELS WITH INDIVIDUAL 1 (IN INDIVIDUAL RECORD ORDERING) VAR-COV OF DATA FROM INDIVIDUAL RECORD ESTIMATED TO BE SINGULAR MESSAGE ISSUED UPON EVALUATION OF OBJ. FUNCTION *** Line 1 *** which calculates A3 from the negative sum of A1 and A2 seems to be causing the problem. When I replace LINE 1 with LINE 2 (which gives the result of the calculation that should be performed in LINE 1) the code runs without error Can anyone explain this? Best wishes Mick Looby ------------------------------------------------- $PROBLEM TEST $INPUT ID TIME CON=DV ND $DATA TEST.DAT $PRED A1 = THETA(1) A2 = THETA(2) LA1 = THETA(3) LA2 = THETA(4) LA3 = THETA(5) ; *** LINE 1 *** A3 = -(A1 + A2) ; *** LINE 2 *** ;A3 = -9.95 TAU = 8 TI = TIME - (ND - 1) * TAU EX1 = EXP(-LA1 * TI) EX2 = EXP(-LA2 * TI) EX3 = EXP(-LA3 * TI) EX11 = 1 - EXP(-LA1 * TAU * ND) EX22 = 1 - EXP(-LA2 * TAU * ND) EX33 = 1 - EXP(-LA3 * TAU * ND) EX111 = 1 - EXP(-LA1 * TAU) EX222 = 1 - EXP(-LA2 * TAU) EX333 = 1 - EXP(-LA3 * TAU) AA1 = A1 * EX1 * EX11/EX111 AA2 = A2 * EX2 * EX22/EX222 AA3 = A3 * EX3 * EX33/EX333 F = AA1 + AA2 + AA3 Y = F * EXP(ERR(1)) $THETA 6.75 3.2 10.7 0.136 9.17 $OMEGA .2 ;$ESTIM ------------------------------------ Data to be simulated: 1 0 0 1 1 1 0 1 1 4 0 1 1 8 0 2 1 9 0 2