subroutine ddaslx (ddasf, neq, t, y, yprime, tout, info, rtol, & atol, idid, rwork, lrw, iwork, liw) c Copyright (c) 2006, Math a la Carte, Inc. c>> 2009-10-31 ddaslx Krogh Insured that lenpd was defined. c>> 2008-11-13 ddaslx Hanson correct out of date comments about starting c>> 2008-08-27 ddaslx Hanson correct leading dimension for banded problems c>> 2008-08-26 ddaslx Hanson add argument of leading dimension to ddasf c>> 2008-04-18 ddaslx Hanson allowed INFO(11) =0,1,2 for new startup c>> 2006-05-19 ddaslx Hanson flag (-29 -> -31) & new const errs -29, -30 c>> 2006-04-24 ddaslx Krogh Added logical variable, nomat. c>> 2003-07-15 ddaslx Hanson Install Soderlind stepsize code. c>> 2002-06-26 ddaslx Krogh Insured iwork(lk) has current value. c>> 2001-12-29 ddaslx Krogh Many changes in usage and documentation. c>> 2001-12-12 ddaslx Krogh Changed code for reverse communication c>> 2001-11-23 ddaslx Krogh Changed many names per library conventions. c>> 2001-11-04 ddaslx Krogh Fixes for F77 and conversion to single c>> 2001-11-01 ddaslx Hanson Provide code to Math a la Carte. c--D replaces "?":?daslx, ?daswt, ?das1, ?dasnm, ?dastp, c--& ?dasin, ?daswt, ?dasf, ?dasdb, ?mess c***BEGIN PROLOGUE DDASLX c***PURPOSE This code solves a system of differential/algebraic c equations of the form G(T,Y,YPRIME) = 0. c***LIBRARY SLATEC (DDASLX) c***CATEGORY I1A2 c***TYPE DOUBLE PRECISION (SDASSL-S, DDASLX-D) c***KEYWORDS BACKWARD DIFFERENTIATION FORMULAS, DASSL, c DIFFERENTIAL/ALGEBRAIC, IMPLICIT DIFFERENTIAL SYSTEMS c***AUTHOR Petzold, Linda R., (LLNL) c Computing and Mathematics Research Division c Lawrence Livermore National Laboratory c L - 316, P.O. Box 808, c Livermore, CA. 94550 c Changes made by R. J. Hanson, Math a la Carte. c***DESCRIPTION c Subroutine DDASLX uses the backward differentiation formulas of c orders one through five to solve a system of the above form for Y and c YPRIME. Values for T and Y at the initial time must be given as c input. The subroutine solves the system from T to TOUT. It is easy c easy to continue the solution to get results at additional TOUT. c This is the interval mode of operation. Intermediate results can c also be obtained easily by using the intermediate-output capability. c The code automatically selects the order and step size. c c *Usage: c c EXTERNAL DDASF c INTEGER NEQ, INFO(N), IDID, LRW, LIW, IWORK(LIW), c DOUBLE PRECISION T, Y(NEQ), YPRIME(NEQ), TOUT, RTOL, ATOL, c * RWORK(LRW) c c CALL DDASLX (DDASF, NEQ, T, Y, YPRIME, TOUT, INFO, RTOL, ATOL, c IDID, RWORK, LRW, IWORK, LIW) c c ************************* Arguments ********************************** c c DDASF:EXT -- Depending on how INFO(5) is set, this routine may be c used to compute g, compute the iteration matrix, and/or to factor c and solve the linear systems as requested. It has the form c SUBROUTINE DDASF(T,Y,YPRIME,DELTA,PD,LDD,CJ,IRES,RWORK,IWORK) c The actions to be taken depend on the value of the IRES argument c to DDASF. IRES also provides a way for the user to signal certain c things to DDASLX. The values of IRES are c = 0 Initialize subroutine parameters or any storage requirements. c If this does not apply, then ignore this value and return c without any further action. Initial conditions for t, y and y' c can be assigned to T, Y(:) and YPRIME(:). Do not alter c DELTA, D or C when IRES = 0. c When reverse communication is used, IRES is in IWORK(3), DELTA c starts in RWORK(IWORK(4)), the matrix PD starts in c RWORK(IWORK(5)), and CJ is in RWORK(1). c In your code you can ignore any values for IRES which can not c occur for the value of INFO(5) which is set. c = 1 Evaluate the residual of the differential/algebraic system. c Place values in the array DELTA = g}(t, y, y'). c = 2 This case occurs for |INFO(5)| = 2, 4, 5, 6, 8, and 10. c Evaluate the partial derivatives (p used for partial c symbol), D(i,j) = (pg_i / py_j) + c (pg_i) / py'_j. c The value LDD is the row dimension of D(*,*). The c scalar c is an input argument to DDASF. For the given values of c T, Y, YPRIME, the routine must evaluate the non-zero partial c derivatives for each equation, and store these values in the c matrix D. For dense or banded matrices, stored in the work c space RWORK(:), the elements of D are set to zero before each c call to DDASF so only non-zero elements need to be defined. Do c not alter T, Y, YPRIME, or C. You must dimension D with a first c first dimension of LDD. The way you must store the elements c into D depends on the structure of the matrix, indicated by c INFO(5) as follows. c INFO(5)=2, 8, and 12: Full (dense) matrix. When you evaluate c the (non-zero) partial derivative of equation i with c respect to variable j, you must store it in D according c to D(i,j) = (p g_{i} / py_{j})+c (pg_{i}/py'_{j}). c INFO(5)=4, 10, and 14: Banded Jacobian with ML lower and MU c upper diagonal bands (refer to INFO(6) description of ML c and MU). Give D a first dimension of 2*ML+MU+1. When you c evaluate the (non-zero) partial derivative of equation i c with respect to variable j, you must store it in D c according to irow=i-j+ ML+MU+1, D(irow,j) = (pg_i /py_j) + c c (pg_i / py'_j). c INFO(5) = 5 and 6: The array D is an unused dummy argument. c Save the partials (pg_i / py_j) + c (p g_{i} / py'_j) in c the subroutine DDASF. This case requires that you factor c the matrix (if you ever do) at this time as the value c IRES=3 is not provided in this case. As for that case you c should set IRES = 0 to flag the fact that there were no c problems in obtaining the factorization. c = 3 This case occurs for |INFO(5)| > 6. Factor the matrix of c partials. Prepare to solve systems involving the matrix D = c pg_i / py_j) + c (pg_i / py'_j). The solution method can be a c convenient one of the user's choice. If the matrix is non- c singular it is important to return IRES = 0 as a signal. c Otherwise return IRES = 3, if the system is nearly singular. c = 4 Solve a linear algebraic system using the matrix of partials D, c i.e. solve Dw = r for w. The vector r is input in array c DELTA(:). The solution w overwrites DELTA(:). If for any c reason the system cannot be solved, return w = r as the c approximate solution. This may cause the integrator to take c corrective action such as reducing the step-size or the order of c the formulas used. This situation may occur when iteratively c solving a linear system, but requiring an excessive number of c iterations. c = 5 Compute the residual and partials for projecting the solution c Y(:) onto the constraints after a step has been computed and the c corrector equation has converged. If you are handling the c linear algebra (|INFO(5)| .ge. 5) then you should also compute c the projection and store it in DELTA(:). The DDASLX code c applies the projection, Y(:) = Y(:) - DELTA(:). (Note that the c flag is given to DDASF if INFO(5) .ge. 0, and else is provided c using reverse commiunication. If for example you are using c forward communication for derivatives and doing you own linear c algebra using reverse communication, you will need to either c deal with the linear algebra in DDASF, or call a routine from c there that will do the job.) c Ordinarily subroutine DDASF should not change the value of IRES. c The following values can be set for special cases. c 0 This must be set if you are factoring the iteration matrix, to c let DDASLX know that your matrix is not singular. c -1 Some kind of difficulty has been encountered. This causes c DDASLX to reduce the stepsize or order which may cure the c problem. c -2 Return immediately to the main program, for one reason or the c other it is time to quit. c <-2 This has the same effect as setting INFO(6) to the negative of c this number. It provides a way of turning on debug print at c any time. c c NEQ:IN This is the number of equations to be solved. NEQ .ge. 1. c c T:INOUT This is the current value of the independent variable. c Initialize to the value at the initial point. c c Y(.ge. NEQ):INOUT This array contains the solution components at T. c Initialize to the value at the initial point. c c YPRIME(.ge. NEQ):INOUT This array contains the derivatives of the c solution components at T. You can either provide intial values, c or use INFO(11) = 1 to have them computed. c c If =1, use the method given in the origial DDASL code of Petzold. c If =2, use the Newton method of Krogh and Hanson, c Solving Constrained Differential-Algebraic Systems Using c Projections, 2008. This is recommended but requires more storage c than the option with vaue =1. c c TOUT:IN This is the first point at which a solution is desired. The c value of TOUT relative to T defines the direction of the c integration, so TOUT = T is not allowed at the intial point. c c INFO(.ge. 16):IN The basic task of the code is to solve the system c from T to TOUT and return an answer at TOUT. INFO is an integer c array which is used to communicate exactly how you want this task c to be carried out. You must provide inputs for the first 16 c entries. The simplest use of the code is obtained by setting all c entries to 0, although you probably want to check INFO(5) at c least c c INFO(1) - This parameter enables the code to initialize itself. You c must set it to indicate the start of every new problem. c c **** Is this the first call for this problem ... c Yes - Set INFO(1) = 0 c No - Not applicable here. c See below for continuation calls. **** c c INFO(2) - How much accuracy you want of your solution is specified c by the error tolerances RTOL and ATOL. The simplest use is to c take them both to be scalars. To obtain more flexibility, they c can both be vectors. The code must be told your choice. c c **** Are both error tolerances RTOL, ATOL scalars ... c Yes - Set INFO(2) = 0 c and input scalars for both RTOL and ATOL c No - Set INFO(2) = 1 c and input arrays for both RTOL and ATOL **** c c INFO(3) - The code integrates from T in the direction of TOUT by c steps. If you wish, it will return the computed solution and c derivative at the next intermediate step (the intermediate-output c mode) or TOUT, whichever comes first. This is a good way to c proceed if you want to see the behavior of the solution. If you c must have solutions at a great many specific TOUT points, this c code will compute them efficiently. You can change this mode c at any time. c c **** Do you want the solution only at c TOUT (and not at the next intermediate step) ... c Yes - Set INFO(3) = 0 c No - Set INFO(3) = 1 **** c c INFO(4) - To handle solutions at a great many specific values TOUT c efficiently, this code may integrate past TOUT and interpolate to c obtain the result at TOUT. Sometimes it is not possible to c integrate beyond some point TSTOP because the equation changes c there or it is not defined past TSTOP. Then you must tell the code c not to go past. The code will never give values beyond TSTOP, but c you can change this mode at any time. However when you set c INFO(4) to 1, you must always set RWORK(1) to a value past the c current time. c c **** Can the integration be carried out without any c restrictions on the independent variable T ... c Yes - Set INFO(4)=0 c No - Set INFO(4)=1 c and define the stopping point TSTOP by c setting RWORK(1)=TSTOP **** c c INFO(5) - To solve differential/algebraic problems it is necessary c to use a matrix of partial derivatives for the system of c differential equations. This flag must be set to define your c mtarix. c = 1 (or 0) Matrix is full and dense and numerical derivatives are c generated internally. Although it is less trouble for you to have c the code compute derivatives by numerical differencing, the c solution will be more reliable if you provide the derivatives c using formulas. c = 2 Matrix is full and dense and the user is computing derivatives. c = 3 As for 1, but the matrix, (p stands for the partial operator) c D = pg/py+ c (pg}/py', is banded. Here c is a factor determined c by DDASLX. The differential equation is said to have c half-bandwidths ML (lower) and MU (upper) if D involves only some c of the unknowns y_{J} with i - ML .le. J .le. i + MU for all c i=1,...,NEQ. Thus, ML and MU are the widths of the lower and c upper parts of the band, with the main diagonal being excluded. c If the matrix is stored in the RWORK(:) array and you do not c indicate that the equation has a banded matrix of partial c derivatives, the code works with a full array of NEQ**2 elements, c stored in the conventional Fortran array style. Computations with c banded matrices typically require less time and storage than with c full matrices, if 2 * ML + MU < NEQ. If you tell the code that c the matrix of partial derivatives has a banded structure and you c want to provide subroutine DDASF to compute the partial c derivatives, then you must store the elements of the matrix in the c Linpack band-matrix specification, indicated in the description of c DDASF. c Provide the lower ML and upper MU bandwidths by setting c IWORK(1) = ML and IWORK(2) = MU. c = 4 As for 3, but derivatives are to be computed by the user. c = 5 The user is going to take care of all matters connected with the c factorization and solving of the systems. c = 6 As for 5, except the linear algebra is to be accomplished using c reverse communication. c = 7 - 10 As for 1 - 4, except user is doing linear algebra in DDASF. c = 11 - 14 As for 7 - 10, except using reverse communication. c = k<0 Same as -k, except the evaluation of g and all partials is done c using reverse communciation rather than by calling DDASF. c c INFO(6) - Set the level of debugging print. A value of 0 gives no c print. Otherwise this is a 7 digit number d_6d_5d_4d_3d_2d_1d_0 c defining what to print as follows. c d_0 Print on entry to DDASLX c = 0 No print c = 1 IDID, INFO(1), NEQ, T, TOUT c = 2 y, y' c = 3 Tolerances c = 4 INFO (all of it) c d_1 Print on exit from DDASLX. Print is as for d_0 c d_2 After a call to DDASF (or return from reverse communication c that would ordinarily call DDASF. c = 0 No print c = 1 Print whatever was just computed, except no matrices. c = 2 Print whatever was just computed, including matrices. c d_3 As for the case above, except print is for what is in the c locations about to be stored to. c d_4 Internal print inside subroutine DDASTP. c = 0 No print c = 1 y, y' and corrections. c = 2 tolerances c = 3 difference tables c = 4 integration coefficients c d_5 Determines how much of WORK and IWORK are printed, when c there is other print. c = 0 No print c = 1 Always print IWORK(1:16) c = 2 Always print WORK(1:9) c = 3 Always print both of the above. c d_6 For turning off, or limiting the amount of print. c = 0 No effect c = 1 No effect, but gives a way to specify a value of 0, 1 or 2 c when passing a negative value of IRES after starting. c > 1 Print data for just this many of the first variables, and c just this many of the first rows in matrices when c variables or matrices are printed. c c INFO(7) - You can specify a maximum (absolute value of) stepsize, so c that the code will avoid passing over very large regions. c c **** Do you want the code to decide on its own maximum stepsize? c Yes - Set INFO(7)=0 c No - Set INFO(7)=1 c and define HMAX by setting c RWORK(2)=HMAX **** c c INFO(8) - Differential/algebraic problems may occasionally suffer c from severe scaling difficulties on the first step. If you know a c great deal about the scaling of your problem, you can help to c alleviate this problem by specifying an initial stepsize HO. c c **** Do you want the code to define its own initial stepsize? c Yes - Set INFO(8)=0 c No - Set INFO(8)=1 c and define HO by setting c RWORK(3)=HO **** c c INFO(9) - If storage is a severe problem, you can save some c locations by restricting the maximum order MAXORD. the default c value is 5. for each order decrease below 5, the code requires NEQ c fewer locations, however it is likely to be slower. In any case, c you must have 1 .LE. MAXORD .LE. 5 c c **** Do you want the maximum order to default to 5? c Yes - Set INFO(9)=0 c No - Set INFO(9)=MAXORD .le. 5 c c INFO(10) - If you know that the solutions to your equations satisfy c some constraints you may want to use this option. It is expected c to be especially useful when one differentiates to reduce the c index of a problem. This option must not be used when INFO(5) = c 3 or 4. c c **** Do you want the code to solve the problem without c invoking constraints? c Yes - Set INFO(10)=0 c No - Set INFO(10)=k, where k is >0 if one wants c to solve for k constraints using forward c communication, and k < 0 to solve for -k c constraints using reverse communication. c Artificially require |k| .le. 1000. c c INFO(11) - DDASLX normally requires the initial T, Y, and YPRIME to c be consistent. That is, you must have G(T,Y,YPRIME) = 0 at the c initial time. If you do not know the initial derivative precisely, c you can let DDASLX try to compute it. c c **** Are the initial T, Y, YPRIME consistent? c Yes - Set INFO(11) = 0 c No - Set INFO(11) = 1, and set YPRIME to an initial c approximation to YPRIME. (If you have no idea what c YPRIME should be, set it to zero. Note that the c initial Y should be such that there must exist a c YPRIME so that G(T,Y,YPRIME) = 0.) c No - Set INFO(11) = 2, and set YPRIME to an initial c approximation to YPRIME. This option uses the c Krogh/Hanson starting algorithm found in the report c Solving Constrained Differential-Algebraic Systems c Using Projections, 2008. This is the preferred c option if the code solves for YPRIME. c c INFO(12) - DDASLX} normally allows up to 500 internal steps between c each output points. c Set INFO(12)=0 for the code to use up to 500 internal steps c between output points. c Set INFO(12)=k} and the code will use up to k internal steps c between output points. c INFO(13) - DDASLX} normally uses the smoothed step control algorithm c developed by Gustaf Soderlind. c Set INFO(13)=0 for the code to use the step control method c of Soderlind. c Set INFO(13)=1 and the code will use the original step c control logic of DASSL. c c INFO(14:16) -- Not used currently by the code, but must be set to 0. c c RTOL,ATOL:INOUT These quantities represent relative and absolute c error tolerances which you provide to indicate how accurately you c wish the solution to be computed. You may choose them to be both c scalars (INFO(2)=0) or else both vectors (INFO(2)=1). Caution: c In Fortran 77, a scalar is not the same as an array of length 1. c Some compilers may require to make the scalars into arrays of c length 1. All components must be non-negative. c The tolerances are used by the code in a local error test at each c step which requires roughly that c ABS(LOCAL ERROR) .LE. RTOL*ABS(Y)+ATOL c for each vector component. (More specifically, a root-mean-square c norm is used to measure the size of vectors, and the error test c uses the magnitude of the solution at the beginning of the step.) c c The true (global) error is the difference between the true c solution of the initial value problem and the computed c approximation. Practically all present day codes, including this c one, control the local error at each step and do not even attempt c to control the global error directly. Usually, but not always, c the true accuracy of the computed Y is comparable to the error c tolerances. This code will usually, but not always, deliver a more c accurate solution if you reduce the tolerances and integrate c again. By comparing two such solutions you can get a fairly c reliable idea of the true error in the solution at the bigger c tolerances. c c Setting ATOL=0. results in a pure relative error test on that c component. Setting RTOL=0. results in a pure absolute error test c on that component. A mixed test with non-zero RTOL and ATOL c corresponds roughly to a relative error test when the solution c component is much bigger than ATOL and to an absolute error test c when the solution component is smaller than the threshold ATOL. c c The code will not attempt to compute a solution at an accuracy c unreasonable for the machine being used. It will advise you if c you ask for too much accuracy and inform you as to the maximum c accuracy it believes possible. c c IDID:OUT This scalar quantity is an indicator reporting what the c code did. You must monitor this integer variable to decide what c action to take next. Values returned are as follows. c Task Completed or Ongoing c 1 A step was successfully taken in the intermediate-output c mode. The code has not yet reached TOUT. c 2 The integration to TSTOP was successfully completed to c T = TSTOP by stepping exactly to TSTOP. c 3 The integration to TOUT was successfully completed T = TOUT by c stepping past TOUT. Y(:)} is obtained by interpolation. c YPRIME(:) is obtained by interpolation. c 4 The integration has paused for reverse communication. Respond c based on the values of IWORK(3). c Task Interupted c -1 IRES set to -2 by the user. c -2 Accuracy requested exceeds machine precision. RTOL and ATOL c have been increased. c -3 There have been too many steps between output points. c Quit or Restart Integration c -4 No convergence due to IRES being set to -1. c -5 A weight for computing error norms is .le. 0. c -6 The error test has failed repeatedly. c -7 Repeated failure of the corrector to converge. c -8 The iteration matrix is singular. c -9 Repeated corrector convergence failures, with singular c matrices flagged. c -10 Could not solve for the initial y'. c Invalid input c -11 An INFO entry has a value not allowed for that option. c -12 The number of equations was set .le. 0. c -13 The maximum order does not have a value from 1 to 5. c -14 The size of RWORK is too small. c -15 The size of IWORK is too small. c -16 An entry of RTOL is < 0. c -17 An entry of ATOL is < 0. c -18 All entries of RTOL and ATOL are 0. c -19 The value of TOUT is > TSTOP. c -20 The maximum stepsize is set .le. 0. c -21 The current TOUT is behind T. c -22 The initial stepsize has been set to 0. c -23 TOUT is too close to the starting T. c -24 TSTOP is not consistent with the current T. c -25 An illegal bandwidth. c -26 The current T and TOUT are equal. c -27 Constraints used with band matrices. c -28 When solving constraints for a user defined matrix, IRES was c not changed from 5. c -29 Inconsistency in the constraints when starting. c -30 During later integration steps, an inconsistency and rank c deficiency were noted. Print amount constraint residual norm c and linear system residual norm. c -31 No appropriate action was taken after IDID was set < 0. c c RWORK:WORK A real work array of length LRW which provides the c code with needed storage space. The data in RWORK is necessary c for subsequent calls. You may find the following locations of c interest. c RWORK(3) The step size H to be attempted on the next step. c RWORK(4) The current value of the independent variable, i.e., c the farthest point integration has reached. This will c be different from T only when interpolation has been c performed (IDID=3). c RWORK(7) The stepsize used on the last successful step. c c LRW:IN The length of RWORK. Set it to the declared length of the c RWORK array. You must have LRW .GE. 45+(MAXORD+4)*NEQ c + NEQ**2 for the full (dense) JACOBIAN case c (INFO(5) =0,1,2,7,8,11,12), or c + (2*ML+MU+1)*NEQ for the banded user-defined JACOBIAN case c (INFO(5)=3, 9, 13), or c + (2*ML+MU+1)*NEQ+2*(NEQ/(ML+MU+1)+1) for the banded finite- c difference-generated JACOBIAN (INFO(5)=4, 10, 14), c + 0 For the user defined matrix case. c c IWORK:WORK An integer work array of length LIW which provides the c code with needed storage space. As for RWORK, you may find the c following locations of interest. c IWORK(4) The order of the method to be tried on the next step. c IWORK(5) The order of the method used on the last step. c IWORK(8) The number of steps taken so far. c IWORK(9) The number of calls to DDASF so far. c IWORK(10) The number of evaluations of the matrix of partial c derivatives needed so far. c IWORK(11) The total number of error test failures so far. c IWORK(12) Which contains the total number of convergence test c failures so far. (Includes singular iteration matrix c failures.) cc c LIW:IN The length of IWORK. Set it to the declared length of the c IWORK array. You must have LIW .GE. 30+NEQ c c OPTIONALLY REPLACEABLE NORM ROUTINE: c c DDASLX uses a weighted norm DDASNM to measure the size of vectors c such as the estimated error in each step. A FUNCTION subprogram c DOUBLE PRECISION FUNCTION DDASNM(NEQ,V,WT,RWORK,IWORK) c DIMENSION V(NEQ),WT(NEQ) c is used to define this norm. Here, V is the vector whose norm is c to be computed, and WT is a vector of weights. A DDASNM routine c has been included with DDASLX which computes the weighted c root-mean-square norm given by DDASNM = SQRT((1/NEQ) * c SUM(V(I)/WT(I))**2) this norm is suitable for most problems. In c some special cases, it may be more convenient and/or efficient to c define your own norm by writing a function subprogram to be called c instead of DDASNM. This should, however, be attempted only after c careful consideration. c c ********************* Local Variables ******************************* c c NOTE -- Parameter names are given followed by (ii name) where ii c is the integer value of the parameter, and name is the array the c parameter is used to reference. c c icnstr (10 info) Number of constraints. c idb (6 info) Digits define rules for debugging. From right to left C the digits define output for: entry to ddaslx, exit from ddaslx, C after actions in ddasf, before actions id ddasf, actions inside C ddastp, and what to do about printing iwork and rwork. The bigger C the digit, the more the print; see above for details. c ih0 (8 info) If set, the initial stepsize is stored in rwork(lh=3). c imat (5 info) Defines how the Jacobian is computed and processed. c If < 0, then computation of g and of all partials is done using c reverse communication. Absolute value is used as follows: c 1.Full matrix, derivatives computed using finite differences. c 2 Full matrix, user computes derivatives. c 3 Band matrix, ML=iwork(lml=1), MU=iwork(lmu=2) define the left c and right bandwidths. Derivative approximated with differences. c 4 As for 3, but with the derivatives computed with user supplied c derivatives. c 5 User defined matrix. User does all of the work! c 6 As for 5, but the work is done using reverse communication. c 7-10 As for 1 - 4, but user does linear algebra themselves inside c ddasf. c 11-14 As for 7-10, but reverse communication is used. c imaxh (7 info) If Set, the maximum stepsize has been set by the c user in rwork(lhmax=2) c inityp (11 info) If set computations must be done to get c consistent initial value for y'. c iord (9 info) If set, gives the maximum integration order to use. c iout (3 info) If set, gives output at the end of every step. c istop (4 info) If set, then the user has set rwork(tstop=1) to a c value that the integrator must not go past. c itol (2 info) If set, then atol() and rtol() are arrays, else they c are scalars. c ixstep (12 info) If set, then iwork(mxstep=21) has been set to c info(istop=4) giving the maximum number of steps between outputs. c lalpha (11 rwork) Start of the alphas used in computing integ. coeff. c lbeta (17 rwork) Start of the betas used in computing integ. coeff. c lcj (1 rwork) The constant which multiplies the partials with c respect to y' in the iteration matrix. c lcjold (6 rwork) The previous value of rwork(lcj=5). c lcnstr (10 iwork) The number of constraints. c lctf (17 iwork) Counts the total number of various convergence c failures. c ldelt (4, iwwork) Always equal to ldelta below. c ldelta (46 rwork) Place where residuals and corrections to y are c stored. c le (23 iwork) rwork(iwork(le=22) is where error estimates are stored. c letf (16 iwork) Contains the number of error test failures so far. c lgamma (23 rwork) Start of the gammas used in computing integ. coeff. c lh (4 rwork) Stepsize to be used on the next step. c lhmax (3 rwork) The largest stepsize permitted. c lhmin (10 rwork) The minimum stepsize permitted. c lhold (7 rwork) The previous stepsize. c lipvt (31 iwork) Used for permutations when solving linear systems. c lires (3 iwork) Value of IRES for reverse communication. c ljcalc (19 iwork) This is set to -1 when a new Jacobian is needed, c set to 0 when it is computed, and set to 1 when it has been used. c lk (7 iwork) Contains the current integration order. c lkold (8 iwork) Contains the previous integration order. c lmat (9 iwork) Type of matrix, from info(imat). c lml (1 iwork) Contains the left bandwith (if used). c lmu (2 iwork) Contains the right bandwidth (if used). c lmxord (6 iwork) Contains the maximum integration order allowed. c lnjac (8 rwork) A large number here means we need a new Jacobian. c Small number means that convergence is good. c lnje (15 iwork) Counts the number of Jacobian evaluations so far. c lnpd (18 iwork) This gives the space required to hold partial c derivatives. Includes partials for constraints. c lnre (14 iwork) Counts the number of evaluations of g. c lns (11 iwork) Counts the number of steps without a stepsize change. c lnst (13 iwork) Counts the number of steps taken so far. c lnstl (12 iwork) Number of steps when last returned to user (not c counting computations done in ddasf when not using reverse comm.) c lphase (20 iwork) This is set to 0 initially allowing rapid order c and stepsize increases. It is set to 1 if either the maximum c order is reached, or the order is decreased or there has been some c kind of failure. c lphi (25 iwork) Location of difference tables in rwork. c lpsi (29 rwork) Start of the psi's used in computing integ. coeff. c lround (9 rwork) The machine epsilon. c lsigma (35 rwork) Start of the sigmas used in error estimation. c ltn (5 rwork) The point up to which we have integrated. Local c variable tn in ddaslx contains this value. c ltstop (2 rwork) Value that the integrator is not to go past. c lwm (5 iwork) rwork(iwork(lwm=25)) is the start of matrix data. c lwt (24 iwork) rwork(iwork(lwt=23)) is the start of weights used c in computing error norms. c mxstep (22 iwork) Contains the maximum number of steps to take c between output points. c nomat Logical variable set = .true. if matrix storage not used here. c ntemp (26 iwork) Location relative to work(iwork(lwm=25)) of c extra matrix data required to compute finite difference Jacobians. c revloc (21 iwork) This is nonzero if reverse communication is in c the process of being used. The low order 3 bits define the c location to go to in the current routine, the next three for the c routine that this one calls, and the next 3 bits for the routine c called by that one, etc. c c --------------------------------------------------------------------- c c***REFERENCES A DESCRIPTION OF DASSL: A DIFFERENTIAL/ALGEBRAIC c SYSTEM SOLVER, L. R. PETZOLD, SAND82-8637, c SANDIA NATIONAL LABORATORIES, SEPTEMBER 1982. c***ROUTINES CALLED D1MACH, DDAS1, DDASNM, DDASTP, DDASIN, DDASWT, c XERMSG c***REVISION HISTORY (YYMMDD) c 830315 DATE WRITTEN c 880387 Code changes made. All common statements have been c replaced by a DATA statement, which defines pointers into c RWORK, and PARAMETER statements which define pointers c into IWORK. As well the documentation has gone through c grammatical changes. c 881005 The prologue has been changed to mixed case. c The subordinate routines had revision dates changed to c this date, although the documentation for these routines c is all upper case. No code changes. c 890511 Code changes made. The DATA statement in the declaration c section of DDASLX was replaced with a PARAMETER c statement. Also the statement S = 100.D0 was removed c from the top of the Newton iteration in DDASTP. c The subordinate routines had revision dates changed to c this date. c 890517 The revision date syntax was replaced with the revision c history syntax. Also the "DECK" comment was added to c the top of all subroutines. These changes are consistent c with new SLATEC guidelines. c The subordinate routines had revision dates changed to c this date. No code changes. c 891013 Code changes made. c Removed all occurrences of FLOAT or DBLE. All operations c are now performed with "mixed-mode" arithmetic. c Also, specific function names were replaced with generic c function names to be consistent with new SLATEC guidelines. c In particular: c Replaced DSQRT with SQRT everywhere. c Replaced DABS with ABS everywhere. c Replaced DMIN1 with MIN everywhere. c Replaced MIN0 with MIN everywhere. c Replaced DMAX1 with MAX everywhere. c Replaced MAX0 with MAX everywhere. c Replaced DSIGN with SIGN everywhere. c Also replaced REVISION DATE with REVISION HISTORY in all c subordinate routines. c 901004 Miscellaneous changes to prologue to complete conversion c to SLATEC 4.0 format. No code changes. (F.N.Fritsch) c 901009 Corrected GAMS classification code and converted subsidiary c routines to 4.0 format. No code changes. (F.N.Fritsch) c 901010 Converted XERRWV calls to XERMSG calls. (R.Clemens, AFWL) c 901019 Code changes made. c Merged SLATEC 4.0 changes with previous changes made c by C. Ulrich. Below is a history of the changes made by c C. Ulrich. (Changes in subsidiary routines are implied c by this history) c 891228 Bug was found and repaired inside the DDASLX c and DDAS1 routines. DDAS1 was incorrectly c returning the initial T with Y and YPRIME c computed at T+H. The routine now returns T+H c rather than the initial T. c Cosmetic changes made to DDASTP. c 900904 Three modifications were made to fix a bug (inside c DDASLX) re interpolation for continuation calls and c cases where TN is very close to TSTOP: c c 1) In testing for whether H is too large, just c compare H to (TSTOP - TN), rather than c (TSTOP - TN) * (1-4*UROUND), and set H to c TSTOP - TN. This will force DDASTP to step c exactly to TSTOP under certain situations c (i.e. when H returned from DDASTP would otherwise c take TN beyond TSTOP). c c 2) Inside the DDASTP loop, interpolate exactly to c TSTOP if TN is very close to TSTOP (rather than c interpolating to within roundoff of TSTOP). c c 3) Modified IDID description for IDID = 2 to say c that the solution is returned by stepping exactly c to TSTOP, rather than TOUT. (In some cases the c solution is actually obtained by extrapolating c over a distance near unit roundoff to TSTOP, c but this small distance is deemed acceptable in c these circumstances.) c 901026 Added explicit declarations for all variables and minor c cosmetic changes to prologue, removed unreferenced labels, c and improved XERMSG calls. (FNF) c 901030 Added ERROR MESSAGES section and reworked other sections to c be of more uniform format. (FNF) c 910624 Fixed minor bug related to HMAX (six lines after label c 525). (LRP) c 981110 Major revision on several codes in the package. Now c use one external user routine. Also support reverse c communication. Install JPL error processor in place of c XERMSG. c***END PROLOGUE DDASLX c c**End c c Declare arguments. c integer neq, ldd, info(16), idid, lrw, iwork(*), liw double precision t, y(*), yprime(*), tout, rtol(*), atol(*), & rwork(lrw) external ddasf c c Declare externals. c external d1mach, ddas1, ddasnm, ddastp, ddasin, ddaswt, dmess double precision d1mach, ddasnm c c Declare local variables. c integer i, infobd(2:16), itemp, leniw,lenpd, lenrw, & mband, msave, nzflg, ires double precision atoli, h, hmax, ho, r, rh, rtoli, tdist, & tn, tnext, tstop, ypnorm INTEGER ISMOOT, IFIRST logical done, nomat c c The following locations are reserved for future expansion: c iwork(26:30), work(41:45),; and info(13:16) c c POINTERS INTO IWORK integer lml, lmu, lires, ldelt, lwm, lmxord, lk, lkold, lmat, & lcnstr, lns, lnstl, lnst, lnre, lnje, letf, lctf, lnpd, & ljcalc, lphase, revloc, mxstep, le, lwt, lphi, ntemp, lipvt parameter (lml=1, lmu=2, lires=3, ldelt=4, lwm=5, lmxord=6, lk=7, & lkold=8, lmat=9, lcnstr=10, lns=11, lnstl=12, lnst=13, & lnre=14, lnje=15, letf=16, lctf=17, lnpd=18, ljcalc=19, & lphase=20, revloc=21, mxstep=22, le=23, lwt=24, lphi=25, & ntemp=26, lipvt=31) c c POINTERS INTO RWORK integer lcj, ltstop, lhmax, lh, ltn, lcjold, lhold, lnjac, & lround, lhmin, lalpha, lbeta, lgamma, lpsi, lsigma, ldelta parameter (lcj=1, ltstop=2, lhmax=3, lh=4, ltn=5, lcjold=6, & lhold=7, lnjac=8, lround=9, lhmin=10, lalpha=11, lbeta=17, & lgamma=23, lpsi=29, lsigma=35, ldelta=46) c c POINTERS INTO INFO integer itol, iout, istop, imat, idb, imaxh, ih0, iord, icnstr, & inityp, ixstep parameter (itol=2, iout=3, istop=4, imat=5, idb=6, imaxh=7, & ih0=8, iord=9, icnstr=10, inityp=11, ixstep= 12) parameter (ISMOOT=13, IFIRST=16) c c ********* Error message text *************** c[Last 2 letters of Param. name] [Text generating message.] cAA DDASLX$B cAB Evaluation terminated integration by returning IRES=-2.$N c Current T = $F, Stepsize H = $F.$E cAC Too much accuracy requested for machine precision.$N c RTOL(:) and ATOL(:) replaced by appropriate values.$N c Current T = $F. Last and next step BDF order =$I,$I.$E cAD Did not reach TOUT in MXSTEPS steps.$N c Current T = $F, TOUT = $1$F, MXSTEPS=$2$I.$E cAE Corrector could not converge because evaluation returned $C c IRES = -1.$N c Current T = $F, Stepsize H = $F.$E cAF After starting, an entry of WT(:) has become <= 0.0.$N c Current T = $F.$E cAG The error test failed repeatedly.$N c Current T = $F, Stepsize H = $F. $N c Last and next step BDF order =$I,$I.$E cAH The corrector failed to converge repeatedly.$N c Current T = $F, Stepsize H = $F. $N c Last and next step BDF order =$I,$I.$E cAI The iteration matrix is singular.$N c Current T = $F, Stepsize H = $F.$E cAJ Repeated corrector convergence failures with singular matrices $C c flagged.$N c Current T = $F, $C c Last and next step BDF order =$I,$I.$E cAK Could not initially solve G(t,y,y')=0 for y'. Consider using $C c larger tolerances.$N c Current T = $F, Stepsize H = $F. Last and next $C c step BDF order =$I,$I.$E cAL Setting INFO($4$I) = $I is not an allowed option.$E cAM Number of equations, NEQ, is <= 0. NEQ = $12$I.$E cAN MAXORD, maximum order of BDF used, must be >= 1 and <= 5. $C c MAXORD = $3$I.$E cAO Size of RWORK(:) is too small.$N c Now have LRW = $6$I, but need LRW = $I.$E cAP Size of IWORK(:) is too small.$N c Now have LIW = $8$I, but need LIW = $I.$E cAQ RTOL($4$I) is < 0.$E cAR ATOL($4$I) is < 0.$E cAS All entries of RTOL(:) and ATOL(:) = 0.0$E cAT Your TOUT of $2$F is > than your TSTOP of $1$F whidh is $C c not allowed$E cAU Value of maximum stepsize HMAX = $3$F is <= 0.0.$E cAV The current T = $F with H=$F is preceded by TOUT = $F.$E cAW Value of initial stepsize H0 is = 0.$E cAX With a starting T = $F, the TOUT = $1$F is too close.$E cAY With the current T=$F and H=$F, the TSTOP=$2$F is inconsistent.$E cAZ Lower bandwith is $10$I, upper bandwidth is $I and both should $C c be .ge. 0, and < neq which is $I.$E cBA The current T = $F and TOUT = $1$F are equal.$E cBB Code does not handle constraints when using a band matrix.$E cBC Flag not set for computing constraint correction as it must be $C c with a user defined matrix and constraints.$E cBD At the initial point constraints were not consistent.$E cBE At T = $F, with H = $F, the constraints appear inconsistent. $C c The norm of the residual with maximal perturbation of the $C c solution is $F and we did not reduce the norm below $F.$E cBF Last step terminated with IDID < 0. No appropriate action was $C c taken. IDID = $I. Repeated occurrences of illegal input. $C c Run terminated. Apparent infinite loop.$E c $ D cBG Adjustments to Y in order to get back on the constraints were:$N$B c $ cBH Residuals in the constraints were:$N$B integer LTXTAA,LTXTAB,LTXTAC,LTXTAD,LTXTAE,LTXTAF,LTXTAG,LTXTAH, * LTXTAI,LTXTAJ,LTXTAK,LTXTAL,LTXTAM,LTXTAN,LTXTAO,LTXTAP,LTXTAQ, * LTXTAR,LTXTAS,LTXTAT,LTXTAU,LTXTAV,LTXTAW,LTXTAX,LTXTAY,LTXTAZ, * LTXTBA,LTXTBB,LTXTBC,LTXTBD,LTXTBE,LTXTBF,LTXTBG,LTXTBH parameter (LTXTAA= 1,LTXTAB= 9,LTXTAC= 100,LTXTAD= 259, *LTXTAE= 341,LTXTAF= 444,LTXTAG= 515,LTXTAH= 624,LTXTAI= 744, *LTXTAJ= 813,LTXTAK= 941,LTXTAL=1095,LTXTAM=1146,LTXTAN=1196, *LTXTAO=1270,LTXTAP=1344,LTXTAQ=1417,LTXTAR=1437,LTXTAS=1457, *LTXTAT=1499,LTXTAU=1568,LTXTAV=1618,LTXTAW=1674,LTXTAX=1712, *LTXTAY=1767,LTXTAZ=1832,LTXTBA=1930,LTXTBB=1978,LTXTBC=2038, *LTXTBD=2146,LTXTBE=2201,LTXTBF=2378,LTXTBG= 1,LTXTBH= 1) character MTXTAA(11) * (231) character MTXTAB(1) * (66) character MTXTAC(1) * (38) C++ Save data by elements if ~.C. data MTXTAA(1)/'DDASLX$BEvaluation terminated integration by retur *ning IRES=-2.$NCurrent T = $F, Stepsize H = $F.$EToo much accuracy * requested for machine precision.$NRTOL(:) and ATOL(:) replaced by * appropriate values.$NCurrent T = $F. Last and ne'/ data MTXTAA(2)/'xt step BDF order =$I,$I.$EDid not reach TOUT in M *XSTEPS steps.$NCurrent T = $F, TOUT = $1$F, MXSTEPS=$2$I.$ECorrect *or could not converge because evaluation returned IRES = -1.$NCurr *ent T = $F, Stepsize H = $F.$EAfter starting, an '/ data MTXTAA(3)/'entry of WT(:) has become <= 0.0.$NCurrent T = $F. *$EThe error test failed repeatedly.$NCurrent T = $F, Stepsize H =$ * $F. $NLast and next step BDF order =$I,$I.$EThe corrector failed$ * to converge repeatedly.$NCurrent T = $F, Stepsiz'/ data MTXTAA(4)/'e H = $F. $NLast and next step BDF order =$I,$I.$E *The iteration matrix is singular.$NCurrent T = $F, Stepsize H = $F *.$ERepeated corrector convergence failures with singular matrices$ * flagged.$NCurrent T = $F, Last and next step BDF'/ data MTXTAA(5)/' order =$I,$I.$ECould not initially solve G(t,y,y' *')=0 for y''. Consider using larger tolerances.$NCurrent T = $F,$ * Stepsize H = $F. Last and next step BDF order =$I,$I.$ESetting IN *FO($4$I) = $I is not an allowed option.$ENumber of '/ data MTXTAA(6)/'equations, NEQ, is <= 0. NEQ = $12$I.$EMAXORD, ma *ximum order of BDF used, must be >= 1 and <= 5. MAXORD = $3$I.$ESi *ze of RWORK(:) is too small.$NNow have LRW = $6$I, but need LRW =$ * $I.$ESize of IWORK(:) is too small.$NNow have LI'/ data MTXTAA(7)/'W = $8$I, but need LIW = $I.$ERTOL($4$I) is < 0.$E *ATOL($4$I) is < 0.$EAll entries of RTOL(:) and ATOL(:) = 0.0$EYour * TOUT of $2$F is > than your TSTOP of $1$F whidh is not allowed$EV *alue of maximum stepsize HMAX = $3$F is <= 0.0.$E'/ data MTXTAA(8)/'The current T = $F with H=$F is preceded by TOUT = * $F.$EValue of initial stepsize H0 is = 0.$EWith a starting T = $F *, the TOUT = $1$F is too close.$EWith the current T=$F and H=$F, t *he TSTOP=$2$F is inconsistent.$ELower bandwith is'/ data MTXTAA(9)/' $10$I, upper bandwidth is $I and both should be . *ge. 0, and < neq which is $I.$EThe current T = $F and TOUT = $1$F * are equal.$ECode does not handle constraints when using a band ma *trix.$EFlag not set for computing constraint corr'/ data MTXTAA(10)/'ection as it must be with a user defined matrix a *nd constraints.$EAt the initial point constraints were not consist *ent.$EAt T = $F, with H = $F, the constraints appear inconsistent. * The norm of the residual with maximal perturbati'/ data MTXTAA(11)/'on of the solution is $F and we did not reduce th *e norm below $F.$ELast step terminated with IDID < 0. No appropr *iate action was taken. IDID = $I. Repeated occurrences of illegal * input. Run terminated. Apparent infinite loop.$E '/ C End of storing data by elements. data MTXTAB/'Adjustments to Y in order to get back on the constrai *nts were:$N$B'/ data MTXTAC/'Residuals in the constraints were:$N$B'/ c End of data generated by pmess. c Staging arrays for error printing of data: double precision edat(5) integer mloc(32), mact(5), mactv(6), idat(14), locate, 1 mecont, meemes, mefvec, meret, metdig, metext parameter (mecont=50, meemes=52, mefvec=61, meret=51, metdig=22, 1 metext=53) data mloc/ltxtaa, ltxtab, ltxtac, ltxtad, ltxtae, ltxtaf, & ltxtag, ltxtah, ltxtai, ltxtaj, ltxtak, ltxtal, ltxtam, & ltxtan, ltxtao, ltxtap, ltxtaq, ltxtar, ltxtas, ltxtat, & ltxtau, ltxtav, ltxtaw, ltxtax, ltxtay, ltxtaz, ltxtba, & ltxtbb, ltxtbc, ltxtbd, ltxtbe, ltxtbf / data mact / meemes, 0, 0, 0, meret / data mactv / metdig, 6, metext, mefvec, 0, mecont / c 2-4 5 6 7 8 9 10 11 12 13-16 data infobd / 3*1, 14, 3422444, 1, 1, 5, 1000, 1, 1000000, 4*0 / integer lerrab, lerrac, lerrad, lerrae, lerraf, lerrag, & lerrah, lerrai, lerraj, lerrak, lerral, lerram, lerran, $ lerrao, lerrap, lerraq, lerrar, lerras, lerrat, lerrau, $ lerrav, lerraw, lerrax, lerray, lerraz, lerrba, lerrbb, $ lerrbc, lerrbd, lerrbe, lerrbf, lerrbg c The following parameters define the severity for stopping and c printing of the associated error message . parameter(lerrab = 32) parameter(lerrac = 62) parameter(lerrad = 62) parameter(lerrae = 84) parameter(lerraf = 84) parameter(lerrag = 86) parameter(lerrah = 88) parameter(lerrai = 88) parameter(lerraj = 88) parameter(lerrak = 88) parameter(lerral = 88) parameter(lerram = 88) parameter(lerran = 88) parameter(lerrao = 88) parameter(lerrap = 88) parameter(lerraq = 88) parameter(lerrar = 88) parameter(lerras = 88) parameter(lerrat = 88) parameter(lerrau = 88) parameter(lerrav = 88) parameter(lerraw = 88) parameter(lerrax = 88) parameter(lerray = 88) parameter(lerraz = 88) parameter(lerrba = 88) parameter(lerrbb = 88) parameter(lerrbc = 88) parameter(lerrbd = 88) parameter(lerrbe = 88) parameter(lerrbf = 99) integer maperr(31) data maperr / lerrab, lerrac, lerrad, lerrae, lerraf, lerrag, & lerrah, lerrai, lerraj, lerrak, lerral, lerram, lerran, $ lerrao, lerrap, lerraq, lerrar, lerras, lerrat, lerrau, $ lerrav, lerraw, lerrax, lerray, lerraz, lerrba, lerrbb, $ lerrbc, lerrbd, lerrbe, lerrbf / SAVE c c***FIRST EXECUTABLE STATEMENT DDASLX if (info(1) .eq. 0) then c Initialization c ---------------------------------------------------------------------- c THIS BLOCK IS EXECUTED FOR THE INITIAL CALL ONLY. c IT CONTAINS CHECKING OF INPUTS AND INITIALIZATIONS. c ---------------------------------------------------------------------- C First time flag use in ddastp for step smoothing logic. INFO(IFIRST)=0 c c FIRST CHECK INFO ARRAY TO MAKE SURE ALL ELEMENTS OF INFO c ARE valid. idid = -11 do 10 i=2, 16 if ((info(i) .gt. infobd(i)) .or. (info(i) .lt. 0)) then if ((i .ne. 5) .or. (info(i) .lt. -infobd(i))) then if (i .eq. iord) idid = -13 go to 500 end if end if 10 continue c Reverse communication controller is cleared and set iwork(ldelt) iwork(revloc) = 0 iwork(ldelt) = ldelta iwork(lkold) = 0 iwork(lk) = 1 c if (neq .le. 0) then idid = -12 go to 510 end if c c CHECK LENGTH IWORK leniw = 30 + neq if (liw .lt. leniw) then idid = -15 go to 510 end if c c Reserve space for constraints (if any) iwork(lcnstr) = info(icnstr) c May be changed later if Jacobian is banded. ldd = neq + iwork(lcnstr) c c May reset maximum number of steps between output points. iwork(mxstep) = info(ixstep) if (iwork(mxstep) .le. 0) iwork(mxstep) = 510 c c Set maximum order if (info(iord) .eq. 0) then iwork(lmxord) = 5 else iwork(lmxord) = info(iord) end if c c Compute lenpd,lenrw.check ml and mu. iwork(lmat) = info(imat) i = abs(iwork(lmat)) nomat = (i .eq. 5) .or. (i .eq. 6) lenrw = (iwork(lmxord)+4+iwork(lcnstr))*neq + 45 if (i .gt. 4) then lenpd = 0 if (i .le. 6) go to 70 i = i - 6 if (i .gt. 4) i = i - 4 end if c User is not taking care of linear algebra. if (i .lt. 3) then c The dense case lenpd = neq*(neq+iwork(lcnstr)) lenrw = lenrw + lenpd else c The banded case if (iwork(lcnstr) .ne. 0) then idid = -27 go to 510 end if if ((iwork(lml) .lt. 0) .or. (iwork(lml) .ge. neq) .or. & (iwork(lmu) .lt. 0) .or. (iwork(lmu) .ge. neq)) then idid = -25 go to 510 end if ldd = 2*iwork(lml)+iwork(lmu)+1 lenpd = ldd*neq if (i .eq. 3) then c Derivatives computed with finite differences. mband = iwork(lml) + iwork(lmu) + 1 msave = (neq/mband) + 1 lenrw = lenrw + lenpd+2*msave else lenrw = lenrw + lenpd end if end if c c CHECK LENGTH OF RWORK. 70 continue iwork(lnpd) = lenpd if (lrw .lt. lenrw) then idid = -14 go to 510 end if c Check consistent logic for storage of derivatives c and performing the linear algebra. c c CHECK TO SEE THAT TOUT IS DIFFERENT FROM T if (tout .eq. t) then idid = -26 go to 510 end if c c CHECK HMAX if (info(imaxh) .ne. 0) then hmax = rwork(lhmax) if (hmax .le. 0.0d0) then idid = -20 go to 510 end if end if c c INITIALIZE COUNTERS iwork(lnst) = 0 iwork(lnre) = 0 iwork(lnje) = 0 c iwork(lnstl) = 0 idid = 1 info(1) = 1 c Flag that we are not completely initialized. rwork(lround) = 0.d0 c A call to the evaluator for setting up or initializing. c If reverse communication is used this is either c a benign call to the dummy routine or else to the user's code, ires = 0 call ddasf (t, y, yprime, rwork, rwork, LDD, & rwork(lcj), ires, rwork, iwork) go to 110 end if c c ---------------------------------------------------------------------- c THIS BLOCK IS FOR CONTINUATION CALLS c ONLY. HERE WE CHECK INFO(1), AND IF THE c LAST STEP WAS INTERRUPTED WE CHECK WHETHER c APPROPRIATE ACTION WAS TAKEN. c ---------------------------------------------------------------------- c if (info(1) .ne. 1) then i = 1 if (info(1) .ne. -1) then idid = -11 go to 500 end if c c IF WE ARE HERE, THE LAST STEP WAS INTERRUPTED c BY AN ERROR CONDITION FROM DDASTP, AND c APPROPRIATE ACTION WAS NOT TAKEN. THIS c IS A FATAL ERROR. idat(1) = idid idid = -31 go to 520 end if if (iwork(revloc) .gt. 0) then c Using reverse communication locate = mod(iwork(revloc), 8) iwork(revloc) = iwork(revloc) / 8 go to (180, 380), locate end if iwork(lnstl) = iwork(lnst) c c ---------------------------------------------------------------------- c THIS BLOCK IS EXECUTED ON ALL CALLS. c THE ERROR TOLERANCE PARAMETERS ARE c CHECKED, AND THE WORK ARRAY POINTERS c ARE SET. c ---------------------------------------------------------------------- c 110 continue if (info(idb) .ne. 0) call ddasdb(0, neq, t, y, yprime, info, & rwork, iwork, idid, atol, rtol) c CHECK RTOL,ATOL nzflg = 0 itemp = 1 if (info(itol) .eq. 1) itemp = neq do 120 i=1, itemp rtoli = rtol(i) atoli = atol(i) if (rtoli.gt.0.0d0 .or. atoli.gt.0.0d0) nzflg = 1 if (rtoli .lt. 0.0d0) then idid = -16 go to 510 end if if (atoli .lt. 0.0d0) then idid = -17 go to 510 end if 120 continue if (nzflg .eq. 0) then idid = -18 go to 510 end if c c SET UP RWORK STORAGE.IWORK STORAGE IS FIXED c IN DATA STATEMENT. iwork(le) = ldelta + neq + iwork(lcnstr) iwork(lwt) = iwork(le) + neq iwork(lphi) = iwork(lwt) + neq if (nomat) then iwork(lwm) = lrw else iwork(lwm) = iwork(lphi) + (iwork(lmxord)+1)*neq end if iwork(ntemp) = 1 + iwork(lnpd) if (rwork(lround) .ne. 0.d0) go to 220 c c ---------------------------------------------------------------------- c THIS BLOCK IS EXECUTED ON THE INITIAL CALL c ONLY. SET THE INITIAL STEP SIZE, AND c THE ERROR WEIGHT VECTOR, AND PHI. c COMPUTE INITIAL YPRIME, IF NECESSARY. c ---------------------------------------------------------------------- c tn = t idid = 1 c c SET ERROR WEIGHT VECTOR WT. Pass info() to routine for c info(itol), info(ismoot) call ddaswt (neq, info, rtol, atol, y, rwork(iwork(lwt)), & rwork,iwork) do 130 i=1, neq if (rwork(iwork(lwt)+i-1) .le. 0.0d0) then idid = -5 go to 510 end if 130 continue c c COMPUTE UNIT ROUNDOFF AND HMIN rwork(lround) = d1mach(4) rwork(lhmin) = 16.0d0 * rwork(lround) * max(abs(t), abs(tout)) c c CHECK INITIAL INTERVAL TO SEE THAT IT IS LONG ENOUGH tdist = abs(tout-t) if (tdist .lt. rwork(lhmin)) then idid = -23 go to 510 end if c c CHECK HO, IF THIS WAS INPUT if (info(ih0) .ne. 0) then ho = rwork(lh) h = ho if ((tout-t)*h .lt. 0.0d0) then idid = -21 go to 510 end if if (h .eq. 0.0d0) then idid = -22 go to 510 end if else c c COMPUTE INITIAL STEPSIZE, TO BE USED BY EITHER c DDASTP OR DDAS1, DEPENDING ON INFO(inityp) ho = 0.001d0*tdist ypnorm = ddasnm(neq,yprime,rwork(iwork(lwt)),rwork,iwork) if (ypnorm .gt. 0.5d0/ho) ho = 0.5d0/ypnorm ho = sign(ho,tout-t) end if c ADJUST HO IF NECESSARY TO MEET HMAX BOUND if (info(imaxh) .eq. 0) go to 160 rh = abs(ho)/rwork(lhmax) if (rh .gt. 1.0d0) ho = ho/rh c COMPUTE TSTOP, IF APPLICABLE 160 if (info(istop) .ne. 0) then tstop = rwork(ltstop) if ((tstop-t)*ho .lt. 0.0d0) then idid = -24 go to 510 end if if ((t+ho-tstop)*ho .gt. 0.0d0) ho = tstop - t if ((tstop-tout)*ho .lt. 0.0d0) then idid = -19 go to 510 end if end if c c COMPUTE INITIAL DERIVATIVE, UPDATING TN AND Y, IF APPLICABLE rwork(ltn) = tn c Set initial error estimates to 0 (temporary??) do 170 i = iwork(le), iwork(lwt) - 1 rwork(i) = 0.d0 170 continue c If info(inityp)==0, user has provided initial values of y c and yprime. if (info(inityp) .eq. 0) go to 190 c REVERSE ENTRY 1: 180 continue c If info(inityp)==1, user has provided initial values of y c and a defined value of yprime. Use Petzold's starting method. c The preferred starting method uses info(inityp)==2. If(info(inityp) .eq. 1) THEN call ddas1 (rwork(ltn), y, yprime, neq, ldd, ddasf, info, ho, & rwork(iwork(lwt)), idid, rwork(iwork(lphi)), rwork(ldelta), & rwork(iwork(le)), rwork(iwork(lwm)), iwork, rwork) ELSE If(info(inityp) .eq. 2) THEN c Use algorithm of Krogh/Hanson based on Newton method for the c starting values of YPRIME. write(*,*) "This should check that F(t,y,y') is small" END IF tn = rwork(ltn) c See if reverse communication needed: if (iwork(revloc) .ne. 0) then iwork(revloc) = 8 * iwork(revloc) + 1 go to 490 end if if (idid .lt. 0) go to 510 c c LOAD H WITH HO. STORE H IN RWORK(LH) 190 h = ho rwork(lh) = h c Flag that we are starting for ddastp. iwork(revloc) = -1 c c LOAD Y AND H*YPRIME INTO PHI(*,1) AND PHI(*,2) itemp = iwork(lphi) + neq do 200 i=1, neq rwork(iwork(lphi)+i-1) = y(i) rwork(itemp+i-1) = h*yprime(i) 200 continue c go to 320 c c ------------------------------------------------------ c THIS BLOCK IS FOR CONTINUATION CALLS ONLY. ITS c PURPOSE IS TO CHECK STOP CONDITIONS BEFORE c TAKING A STEP. c ADJUST H IF NECESSARY TO MEET HMAX BOUND c ------------------------------------------------------ c 220 continue done = .false. tn = rwork(ltn) h = rwork(lh) if (info(imaxh) .eq. 0) go to 230 rh = abs(h)/rwork(lhmax) if (rh .gt. 1.0d0) h = h/rh 230 continue if (t .eq. tout) then idid = -26 go to 510 end if if ((t-tout)*h .gt. 0.0d0) then idid = -21 go to 510 end if if (info(istop) .eq. 1) go to 260 if (info(iout) .eq. 1) go to 240 if ((tn-tout)*h .lt. 0.0d0) go to 310 call ddasin (tn, tout, y, yprime, neq, iwork(lkold), & rwork(iwork(lphi)), rwork(lpsi)) t = tout idid = 3 done = .true. go to 480 240 if ((tn-t)*h .le. 0.0d0) go to 310 if ((tn-tout)*h .gt. 0.0d0) go to 250 call ddasin (tn, tn, y, yprime, neq, iwork(lkold), & rwork(iwork(lphi)),rwork(lpsi)) t = tn idid = 1 done = .true. go to 480 250 continue call ddasin (tn, tout, y, yprime, neq, iwork(lkold), & rwork(iwork(lphi)), rwork(lpsi)) t = tout idid = 3 done = .true. go to 480 260 if (info(iout) .eq. 1) go to 270 tstop = rwork(ltstop) if ((tn-tstop)*h .gt. 0.0d0) then idid = -24 go to 510 end if if ((tstop-tout)*h .lt. 0.0d0) then idid = -19 go to 510 end if if ((tn-tout)*h .lt. 0.0d0) go to 290 call ddasin (tn, tout, y, yprime, neq, iwork(lkold), & rwork(iwork(lphi)), rwork(lpsi)) t = tout idid = 3 done = .true. go to 480 270 tstop = rwork(ltstop) if ((tn-tstop)*h .gt. 0.0d0) then idid = -24 go to 510 end if if ((tstop-tout)*h .lt. 0.0d0) then idid = -19 go to 510 end if if ((tn-t)*h .le. 0.0d0) go to 290 if ((tn-tout)*h .gt. 0.0d0) go to 280 call ddasin (tn, tn, y, yprime, neq, iwork(lkold), & rwork(iwork(lphi)),rwork(lpsi)) t = tn idid = 1 done = .true. go to 310 280 continue call ddasin (tn, tout, y, yprime, neq, iwork(lkold), & rwork(iwork(lphi)), rwork(lpsi)) t = tout idid = 3 done = .true. go to 310 290 continue c CHECK WHETHER WE ARE WITHIN ROUNDOFF OF TSTOP if (abs(tn-tstop) .gt. 100.0d0 * rwork(lround) * (abs(tn)+abs(h))) & go to 300 call ddasin (tn, tstop, y, yprime, neq, iwork(lkold), & rwork(iwork(lphi)), rwork(lpsi)) idid = 2 t = tstop done = .true. go to 310 300 tnext = tn + h if ((tnext-tstop)*h .le. 0.0d0) go to 310 h = tstop - tn rwork(lh) = h c 310 if (done) go to 480 c c ------------------------------------------------------ c THE NEXT BLOCK CONTAINS THE CALL TO THE c ONE-STEP INTEGRATOR DDASTP. c THIS IS A LOOPING POINT FOR THE INTEGRATION STEPS. c CHECK FOR TOO MANY STEPS. c UPDATE WT. c CHECK FOR TOO MUCH ACCURACY REQUESTED. c COMPUTE MINIMUM STEPSIZE. c ------------------------------------------------------ c 320 continue c c CHECK FOR TOO MANY STEPS if ((iwork(lnst)-iwork(lnstl)) .lt. iwork(mxstep)) go to 330 idid = -3 go to 390 c c UPDATE WT. Pass info() to routine for c info(itol), info(ismoot) 330 call ddaswt (neq, info, rtol, atol, rwork(iwork(lphi)), & rwork(iwork(lwt)),rwork, iwork) do 340 i=1, neq if (rwork(i+iwork(lwt)-1) .gt. 0.0d0) go to 340 idid = -5 go to 390 340 continue c c TEST FOR TOO MUCH ACCURACY REQUESTED. r = ddasnm(neq,rwork(iwork(lphi)),rwork(iwork(lwt)),rwork,iwork) * % 100.0d0 * rwork(lround) if (r .gt. 1.0d0) then c MULTIPLY RTOL AND ATOL BY R AND RETURN if (info(itol) .eq. 0) then rtol(1) = r*rtol(1) atol(1) = r*atol(1) else do 360 i=1, neq rtol(i) = r*rtol(i) atol(i) = r*atol(i) 360 continue end if idid = -2 go to 510 end if c c COMPUTE MINIMUM STEPSIZE rwork(lhmin) = 4.0d0 * rwork(lround) * max(abs(tn),abs(tout)) c c TEST H VS. HMAX if (info(imaxh) .ne. 0) then rh = abs(h)/rwork(lhmax) if (rh .gt. 1.0d0) h = h/rh end if c REVERSE ENTRY 2: 380 continue call ddastp (rwork(ltn), y, yprime, neq, ldd, ddasf, info, h, & rwork(iwork(lwt)), idid, rwork(iwork(lphi)), rwork(ldelta), & rwork(iwork(le)), rwork(iwork(lwm)), iwork, rwork, & rwork(lalpha), rwork(lbeta), rwork(lgamma), rwork(lpsi), & rwork(lsigma), iwork(lk)) tn = rwork(ltn) c See if reverse communication needed: if (iwork(revloc) .ne. 0) then iwork(revloc) = 8 * iwork(revloc) + 2 go to 490 end if 390 if (idid .lt. 0) go to 510 c c ------------------------------------------------------- c THIS BLOCK HANDLES THE CASE OF A SUCCESSFUL RETURN c FROM DDASTP (IDID=1). TEST FOR STOP CONDITIONS. c ------------------------------------------------------- c if (info(istop) .ne. 0) go to 420 if (info(iout) .ne. 0) go to 410 if ((tn-tout)*h .lt. 0.0d0) go to 320 400 call ddasin (tn, tout, y, yprime, neq, iwork(lkold), & rwork(iwork(lphi)), rwork(lpsi)) idid = 3 t = tout go to 480 410 if ((tn-tout)*h .ge. 0.0d0) go to 400 t = tn idid = 1 go to 480 420 if (info(iout) .ne. 0) go to 450 if ((tn-tout)*h .ge. 0.0d0) go to 400 if (abs(tn-tstop) .le. 100.0d0 * rwork(lround) * (abs(tn)+abs(h))) & go to 440 tnext = tn + h if ((tnext-tstop)*h .le. 0.0d0) go to 320 h = tstop - tn go to 320 440 call ddasin (tn, tstop, y, yprime, neq, iwork(lkold), & rwork(iwork(lphi)), rwork(lpsi)) idid = 2 t = tstop go to 480 450 if ((tn-tout)*h .ge. 0.0d0) go to 400 if (abs(tn-tstop) .le. 100.0d0 * rwork(lround) * (abs(tn)+abs(h))) & go to 440 t = tn idid = 1 go to 480 c c ------------------------------------------------------- c ALL SUCCESSFUL RETURNS FROM DDASLX ARE MADE FROM c THIS BLOCK. c ------------------------------------------------------- c 480 continue rwork(lh) = h if (info(idb) .ne. 0) call ddasdb(1, neq, t, y, yprime, & info, rwork, iwork, idid, atol, rtol) return c c ---------------------------------------------------------------------- c THIS BLOCK HANDLES ALL REVERSE COMMUNICATION RETURNS c ---------------------------------------------------------------------- 490 continue t = rwork(ltn) idid = 4 return c One central place to call DMESS to print error messages.: 500 continue idat(5) = i idat(6) = info(i) 510 continue idat(1) = iwork(lkold) idat(2) = iwork(lk) idat(3) = iwork(mxstep) idat(4) = iwork(lmxord) idat(7) = lrw idat(8) = lenrw idat(9) = liw idat(10) = leniw idat(11) = iwork(1) idat(12) = iwork(2) idat(13) = neq idat(14) = idid c edat(1) = tn edat(2) = h if (IDID .le. -29) then if (IDID .eq. -29) then mact(5) = mecont else if (IDID .eq. -30) then edat(3) = rwork(ldelta) edat(4) = rwork(ldelta+1) end if else edat(3) = tout edat(4) = hmax edat(5) = tstop end if c 520 continue mact(2) = maperr(-idid) mact(3) = -idid mact(4) = mloc(1 - idid) call dmess(mact, mtxtaa, idat, edat) info(1) = -1 if (IDID .eq. -29) then mact(5) = meret mactv(5) = neq mactv(6) = mecont call dmess(mactv, mtxtab, idat, rwork(ldelta)) mactv(5) = info(10) mactv(6) = meret call dmess(mactv, mtxtac, idat, rwork(ldelta+neq)) end if go to 480 c ----------END OF SUBROUTINE DDASLX------------------------------------ end .