1 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~! |
---|
2 | ! Rosenbrock_TLM - Implementation of the Tangent Linear Model ! |
---|
3 | ! for several Rosenbrock methods: ! |
---|
4 | ! * Ros2 ! |
---|
5 | ! * Ros3 ! |
---|
6 | ! * Ros4 ! |
---|
7 | ! * Rodas3 ! |
---|
8 | ! * Rodas4 ! |
---|
9 | ! By default the code employs the KPP sparse linear algebra routines ! |
---|
10 | ! Compile with -DFULL_ALGEBRA to use full linear algebra (LAPACK) ! |
---|
11 | ! ! |
---|
12 | ! (C) Adrian Sandu, August 2004 ! |
---|
13 | ! Virginia Polytechnic Institute and State University ! |
---|
14 | ! Contact: sandu@cs.vt.edu ! |
---|
15 | ! Revised by Philipp Miehe and Adrian Sandu, May 2006 ! |
---|
16 | ! This implementation is part of KPP - the Kinetic PreProcessor ! |
---|
17 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~! |
---|
18 | |
---|
19 | MODULE KPP_ROOT_Integrator |
---|
20 | |
---|
21 | USE KPP_ROOT_Precision |
---|
22 | USE KPP_ROOT_Parameters |
---|
23 | USE KPP_ROOT_Global |
---|
24 | USE KPP_ROOT_LinearAlgebra |
---|
25 | USE KPP_ROOT_Rates |
---|
26 | USE KPP_ROOT_Function |
---|
27 | USE KPP_ROOT_Jacobian |
---|
28 | USE KPP_ROOT_Hessian |
---|
29 | USE KPP_ROOT_Util |
---|
30 | |
---|
31 | IMPLICIT NONE |
---|
32 | PUBLIC |
---|
33 | SAVE |
---|
34 | |
---|
35 | !~~~> Statistics on the work performed by the Rosenbrock method |
---|
36 | INTEGER, PARAMETER :: Nfun=1, Njac=2, Nstp=3, Nacc=4, & |
---|
37 | Nrej=5, Ndec=6, Nsol=7, Nsng=8, & |
---|
38 | Nhes=9, Ntexit=1, Nhexit=2, Nhnew = 3 |
---|
39 | |
---|
40 | |
---|
41 | CONTAINS ! Functions in the module KPP_ROOT_Integrator |
---|
42 | |
---|
43 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
44 | SUBROUTINE INTEGRATE_TLM( NTLM, Y, Y_tlm, TIN, TOUT, ATOL_tlm, RTOL_tlm,& |
---|
45 | ICNTRL_U, RCNTRL_U, ISTATUS_U, RSTATUS_U ) |
---|
46 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
47 | IMPLICIT NONE |
---|
48 | |
---|
49 | !~~~> Y - Concentrations |
---|
50 | KPP_REAL :: Y(NVAR) |
---|
51 | !~~~> NTLM - No. of sensitivity coefficients |
---|
52 | INTEGER NTLM |
---|
53 | !~~~> Y_tlm - Sensitivities of concentrations |
---|
54 | ! Note: Y_tlm (1:NVAR,j) contains sensitivities of |
---|
55 | ! Y(1:NVAR) w.r.t. the j-th parameter, j=1...NTLM |
---|
56 | KPP_REAL :: Y_tlm(NVAR,NTLM) |
---|
57 | KPP_REAL, INTENT(IN) :: TIN ! TIN - Start Time |
---|
58 | KPP_REAL, INTENT(IN) :: TOUT ! TOUT - End Time |
---|
59 | !~~~> Optional input parameters and statistics |
---|
60 | INTEGER, INTENT(IN), OPTIONAL :: ICNTRL_U(20) |
---|
61 | KPP_REAL, INTENT(IN), OPTIONAL :: RCNTRL_U(20) |
---|
62 | INTEGER, INTENT(OUT), OPTIONAL :: ISTATUS_U(20) |
---|
63 | KPP_REAL, INTENT(OUT), OPTIONAL :: RSTATUS_U(20) |
---|
64 | KPP_REAL, INTENT(IN), OPTIONAL :: RTOL_tlm(NVAR,NTLM),ATOL_tlm(NVAR,NTLM) |
---|
65 | |
---|
66 | INTEGER, SAVE :: IERR |
---|
67 | KPP_REAL :: RCNTRL(20), RSTATUS(20) |
---|
68 | INTEGER :: ICNTRL(20), ISTATUS(20) |
---|
69 | INTEGER, SAVE :: Ntotal = 0 |
---|
70 | |
---|
71 | ICNTRL(1:20) = 0 |
---|
72 | RCNTRL(1:20) = 0.0_dp |
---|
73 | ISTATUS(1:20) = 0 |
---|
74 | RSTATUS(1:20) = 0.0_dp |
---|
75 | |
---|
76 | ICNTRL(1) = 0 ! non-autonomous |
---|
77 | ICNTRL(2) = 1 ! vector tolerances |
---|
78 | ICNTRL(3) = 5 ! choice of the method |
---|
79 | ICNTRL(12) = 1 ! 0 - fwd trunc error only, 1 - tlm trunc error |
---|
80 | RCNTRL(3) = STEPMIN ! starting step |
---|
81 | |
---|
82 | ! if optional parameters are given, and if they are >=0, then they overwrite default settings |
---|
83 | IF (PRESENT(ICNTRL_U)) THEN |
---|
84 | WHERE(ICNTRL_U(:) >= 0) ICNTRL(:) = ICNTRL_U(:) |
---|
85 | ENDIF |
---|
86 | IF (PRESENT(RCNTRL_U)) THEN |
---|
87 | WHERE(RCNTRL_U(:) >= 0) RCNTRL(:) = RCNTRL_U(:) |
---|
88 | ENDIF |
---|
89 | |
---|
90 | CALL RosenbrockTLM(NVAR, VAR, NTLM, Y_tlm, & |
---|
91 | TIN,TOUT,ATOL,RTOL,ATOL_tlm,RTOL_tlm, & |
---|
92 | RCNTRL,ICNTRL,RSTATUS,ISTATUS,IERR) |
---|
93 | |
---|
94 | !~~~> Debug option: show number of steps |
---|
95 | ! Ntotal = Ntotal + ISTATUS(Nstp) |
---|
96 | ! PRINT*,'NSTEPS=',ISTATUS(Nstp),' (',Ntotal,')',' O3=', VAR(ind_O3) |
---|
97 | |
---|
98 | IF (IERR < 0) THEN |
---|
99 | print *,'Rosenbrock: Unsucessful step at T=', & |
---|
100 | TIN,' (IERR=',IERR,')' |
---|
101 | END IF |
---|
102 | |
---|
103 | STEPMIN = RSTATUS(Nhexit) |
---|
104 | ! if optional parameters are given for output they return information |
---|
105 | IF (PRESENT(ISTATUS_U)) ISTATUS_U(:) = ISTATUS(:) |
---|
106 | IF (PRESENT(RSTATUS_U)) RSTATUS_U(:) = RSTATUS(:) |
---|
107 | |
---|
108 | END SUBROUTINE INTEGRATE_TLM |
---|
109 | |
---|
110 | |
---|
111 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
112 | SUBROUTINE RosenbrockTLM(N,Y,NTLM,Y_tlm, & |
---|
113 | Tstart,Tend,AbsTol,RelTol,AbsTol_tlm,RelTol_tlm, & |
---|
114 | RCNTRL,ICNTRL,RSTATUS,ISTATUS,IERR) |
---|
115 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
116 | ! |
---|
117 | ! TLM = Tangent Linear Model of a Rosenbrock Method |
---|
118 | ! |
---|
119 | ! Solves the system y'=F(t,y) using a Rosenbrock method defined by: |
---|
120 | ! |
---|
121 | ! G = 1/(H*gamma(1)) - Jac(t0,Y0) |
---|
122 | ! T_i = t0 + Alpha(i)*H |
---|
123 | ! Y_i = Y0 + \sum_{j=1}^{i-1} A(i,j)*K_j |
---|
124 | ! G * K_i = Fun( T_i, Y_i ) + \sum_{j=1}^S C(i,j)/H * K_j + |
---|
125 | ! gamma(i)*dF/dT(t0, Y0) |
---|
126 | ! Y1 = Y0 + \sum_{j=1}^S M(j)*K_j |
---|
127 | ! |
---|
128 | ! For details on Rosenbrock methods and their implementation consult: |
---|
129 | ! E. Hairer and G. Wanner |
---|
130 | ! "Solving ODEs II. Stiff and differential-algebraic problems". |
---|
131 | ! Springer series in computational mathematics, Springer-Verlag, 1996. |
---|
132 | ! The codes contained in the book inspired this implementation. |
---|
133 | ! |
---|
134 | ! (C) Adrian Sandu, August 2004 |
---|
135 | ! Virginia Polytechnic Institute and State University |
---|
136 | ! Contact: sandu@cs.vt.edu |
---|
137 | ! Revised by Philipp Miehe and Adrian Sandu, May 2006 |
---|
138 | ! This implementation is part of KPP - the Kinetic PreProcessor |
---|
139 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
140 | ! |
---|
141 | !~~~> INPUT ARGUMENTS: |
---|
142 | ! |
---|
143 | !- Y(N) -> vector of initial conditions (at T=Tstart) |
---|
144 | ! NTLM -> dimension of linearized system, |
---|
145 | ! i.e. the number of sensitivity coefficients |
---|
146 | !- Y_tlm(N*NTLM) -> vector of initial sensitivity conditions (at T=Tstart) |
---|
147 | !- [Tstart,Tend] -> time range of integration |
---|
148 | ! (if Tstart>Tend the integration is performed backwards in time) |
---|
149 | !- RelTol, AbsTol -> user precribed accuracy |
---|
150 | !- SUBROUTINE Fun( T, Y, Ydot ) -> ODE function, |
---|
151 | ! returns Ydot = Y' = F(T,Y) |
---|
152 | !- SUBROUTINE Jac( T, Y, Jcb ) -> Jacobian of the ODE function, |
---|
153 | ! returns Jcb = dF/dY |
---|
154 | !- ICNTRL(1:20) -> integer inputs parameters |
---|
155 | !- RCNTRL(1:20) -> real inputs parameters |
---|
156 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
157 | ! |
---|
158 | !~~~> OUTPUT ARGUMENTS: |
---|
159 | ! |
---|
160 | !- Y(N) -> vector of final states (at T->Tend) |
---|
161 | !- Y_tlm(N*NTLM)-> vector of final sensitivities (at T=Tend) |
---|
162 | !- ISTATUS(1:20) -> integer output parameters |
---|
163 | !- RSTATUS(:20) -> real output parameters |
---|
164 | !- IERR -> job status upon return |
---|
165 | ! - succes (positive value) or failure (negative value) - |
---|
166 | ! = 1 : Success |
---|
167 | ! = -1 : Improper value for maximal no of steps |
---|
168 | ! = -2 : Selected Rosenbrock method not implemented |
---|
169 | ! = -3 : Hmin/Hmax/Hstart must be positive |
---|
170 | ! = -4 : FacMin/FacMax/FacRej must be positive |
---|
171 | ! = -5 : Improper tolerance values |
---|
172 | ! = -6 : No of steps exceeds maximum bound |
---|
173 | ! = -7 : Step size too small |
---|
174 | ! = -8 : Matrix is repeatedly singular |
---|
175 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
176 | ! |
---|
177 | !~~~> INPUT PARAMETERS: |
---|
178 | ! |
---|
179 | ! Note: For input parameters equal to zero the default values of the |
---|
180 | ! corresponding variables are used. |
---|
181 | ! |
---|
182 | ! ICNTRL(1) = 1: F = F(y) Independent of T (AUTONOMOUS) |
---|
183 | ! = 0: F = F(t,y) Depends on T (NON-AUTONOMOUS) |
---|
184 | ! |
---|
185 | ! ICNTRL(2) = 0: AbsTol, RelTol are N-dimensional vectors |
---|
186 | ! = 1: AbsTol, RelTol are scalars |
---|
187 | ! |
---|
188 | ! ICNTRL(3) -> selection of a particular Rosenbrock method |
---|
189 | ! = 0 : default method is Rodas3 |
---|
190 | ! = 1 : method is Ros2 |
---|
191 | ! = 2 : method is Ros3 |
---|
192 | ! = 3 : method is Ros4 |
---|
193 | ! = 4 : method is Rodas3 |
---|
194 | ! = 5 : method is Rodas4 |
---|
195 | ! |
---|
196 | ! ICNTRL(4) -> maximum number of integration steps |
---|
197 | ! For ICNTRL(4)=0) the default value of 100000 is used |
---|
198 | ! |
---|
199 | ! ICNTRL(12) -> switch for TLM truncation error control |
---|
200 | ! ICNTRL(12) = 0: TLM error is not used |
---|
201 | ! ICNTRL(12) = 1: TLM error is computed and used |
---|
202 | ! |
---|
203 | ! RCNTRL(1) -> Hmin, lower bound for the integration step size |
---|
204 | ! It is strongly recommended to keep Hmin = ZERO |
---|
205 | ! RCNTRL(2) -> Hmax, upper bound for the integration step size |
---|
206 | ! RCNTRL(3) -> Hstart, starting value for the integration step size |
---|
207 | ! |
---|
208 | ! RCNTRL(4) -> FacMin, lower bound on step decrease factor (default=0.2) |
---|
209 | ! RCNTRL(5) -> FacMin,upper bound on step increase factor (default=6) |
---|
210 | ! RCNTRL(6) -> FacRej, step decrease factor after multiple rejections |
---|
211 | ! (default=0.1) |
---|
212 | ! RCNTRL(7) -> FacSafe, by which the new step is slightly smaller |
---|
213 | ! than the predicted value (default=0.9) |
---|
214 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
215 | ! |
---|
216 | !~~~> OUTPUT PARAMETERS: |
---|
217 | ! |
---|
218 | ! Note: each call to Rosenbrock adds the corrent no. of fcn calls |
---|
219 | ! to previous value of ISTATUS(1), and similar for the other params. |
---|
220 | ! Set ISTATUS(1:10) = 0 before call to avoid this accumulation. |
---|
221 | ! |
---|
222 | ! ISTATUS(1) = No. of function calls |
---|
223 | ! ISTATUS(2) = No. of Jacobian calls |
---|
224 | ! ISTATUS(3) = No. of steps |
---|
225 | ! ISTATUS(4) = No. of accepted steps |
---|
226 | ! ISTATUS(5) = No. of rejected steps (except at the beginning) |
---|
227 | ! ISTATUS(6) = No. of LU decompositions |
---|
228 | ! ISTATUS(7) = No. of forward/backward substitutions |
---|
229 | ! ISTATUS(8) = No. of singular matrix decompositions |
---|
230 | ! ISTATUS(9) = No. of Hessian calls |
---|
231 | ! |
---|
232 | ! RSTATUS(1) -> Texit, the time corresponding to the |
---|
233 | ! computed Y upon return |
---|
234 | ! RSTATUS(2) -> Hexit, last accepted step before exit |
---|
235 | ! For multiple restarts, use Hexit as Hstart in the following run |
---|
236 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
237 | |
---|
238 | IMPLICIT NONE |
---|
239 | |
---|
240 | !~~~> Arguments |
---|
241 | INTEGER, INTENT(IN) :: N, NTLM |
---|
242 | KPP_REAL, INTENT(INOUT) :: Y(N) |
---|
243 | KPP_REAL, INTENT(INOUT) :: Y_tlm(N,NTLM) |
---|
244 | KPP_REAL, INTENT(IN) :: Tstart, Tend |
---|
245 | KPP_REAL, INTENT(IN) :: AbsTol(N),RelTol(N) |
---|
246 | KPP_REAL, INTENT(IN) :: AbsTol_tlm(N,NTLM),RelTol_tlm(N,NTLM) |
---|
247 | INTEGER, INTENT(IN) :: ICNTRL(20) |
---|
248 | KPP_REAL, INTENT(IN) :: RCNTRL(20) |
---|
249 | INTEGER, INTENT(INOUT) :: ISTATUS(20) |
---|
250 | KPP_REAL, INTENT(INOUT) :: RSTATUS(20) |
---|
251 | INTEGER, INTENT(OUT) :: IERR |
---|
252 | !~~~> Parameters of the Rosenbrock method, up to 6 stages |
---|
253 | INTEGER :: ros_S, rosMethod |
---|
254 | INTEGER, PARAMETER :: RS2=1, RS3=2, RS4=3, RD3=4, RD4=5 |
---|
255 | KPP_REAL :: ros_A(15), ros_C(15), ros_M(6), ros_E(6), & |
---|
256 | ros_Alpha(6), ros_Gamma(6), ros_ELO |
---|
257 | LOGICAL :: ros_NewF(6) |
---|
258 | CHARACTER(LEN=12) :: ros_Name |
---|
259 | !~~~> Local variables |
---|
260 | KPP_REAL :: Roundoff, FacMin, FacMax, FacRej, FacSafe |
---|
261 | KPP_REAL :: Hmin, Hmax, Hstart, Hexit |
---|
262 | KPP_REAL :: Texit |
---|
263 | INTEGER :: i, UplimTol, Max_no_steps |
---|
264 | LOGICAL :: Autonomous, VectorTol, TLMtruncErr |
---|
265 | !~~~> Parameters |
---|
266 | KPP_REAL, PARAMETER :: ZERO = 0.0d0, ONE = 1.0d0 |
---|
267 | KPP_REAL, PARAMETER :: DeltaMin = 1.0d-5 |
---|
268 | |
---|
269 | !~~~> Initialize the statistics |
---|
270 | IERR = 0 |
---|
271 | ISTATUS(1:20) = 0 |
---|
272 | RSTATUS(1:20) = ZERO |
---|
273 | |
---|
274 | !~~~> Autonomous or time dependent ODE. Default is time dependent. |
---|
275 | Autonomous = .NOT.(ICNTRL(1) == 0) |
---|
276 | |
---|
277 | !~~~> For Scalar tolerances (ICNTRL(2).NE.0) the code uses AbsTol(1) and RelTol(1) |
---|
278 | ! For Vector tolerances (ICNTRL(2) == 0) the code uses AbsTol(1:N) and RelTol(1:N) |
---|
279 | IF (ICNTRL(2) == 0) THEN |
---|
280 | VectorTol = .TRUE. |
---|
281 | UplimTol = N |
---|
282 | ELSE |
---|
283 | VectorTol = .FALSE. |
---|
284 | UplimTol = 1 |
---|
285 | END IF |
---|
286 | |
---|
287 | !~~~> Initialize the particular Rosenbrock method selected |
---|
288 | SELECT CASE (ICNTRL(3)) |
---|
289 | CASE (1) |
---|
290 | CALL Ros2 |
---|
291 | CASE (2) |
---|
292 | CALL Ros3 |
---|
293 | CASE (3) |
---|
294 | CALL Ros4 |
---|
295 | CASE (0,4) |
---|
296 | CALL Rodas3 |
---|
297 | CASE (5) |
---|
298 | CALL Rodas4 |
---|
299 | CASE DEFAULT |
---|
300 | PRINT * , 'Unknown Rosenbrock method: ICNTRL(3)=',ICNTRL(3) |
---|
301 | CALL ros_ErrorMsg(-2,Tstart,ZERO,IERR) |
---|
302 | RETURN |
---|
303 | END SELECT |
---|
304 | |
---|
305 | !~~~> The maximum number of steps admitted |
---|
306 | IF (ICNTRL(4) == 0) THEN |
---|
307 | Max_no_steps = 200000 |
---|
308 | ELSEIF (Max_no_steps > 0) THEN |
---|
309 | Max_no_steps=ICNTRL(4) |
---|
310 | ELSE |
---|
311 | PRINT * ,'User-selected max no. of steps: ICNTRL(4)=',ICNTRL(4) |
---|
312 | CALL ros_ErrorMsg(-1,Tstart,ZERO,IERR) |
---|
313 | RETURN |
---|
314 | END IF |
---|
315 | !~~~> TLM truncation error control selection |
---|
316 | IF (ICNTRL(12) == 0) THEN |
---|
317 | TLMtruncErr = .FALSE. |
---|
318 | ELSE |
---|
319 | TLMtruncErr = .TRUE. |
---|
320 | END IF |
---|
321 | |
---|
322 | !~~~> Unit roundoff (1+Roundoff>1) |
---|
323 | Roundoff = WLAMCH('E') |
---|
324 | |
---|
325 | !~~~> Lower bound on the step size: (positive value) |
---|
326 | IF (RCNTRL(1) == ZERO) THEN |
---|
327 | Hmin = ZERO |
---|
328 | ELSEIF (RCNTRL(1) > ZERO) THEN |
---|
329 | Hmin = RCNTRL(1) |
---|
330 | ELSE |
---|
331 | PRINT * , 'User-selected Hmin: RCNTRL(1)=', RCNTRL(1) |
---|
332 | CALL ros_ErrorMsg(-3,Tstart,ZERO,IERR) |
---|
333 | RETURN |
---|
334 | END IF |
---|
335 | !~~~> Upper bound on the step size: (positive value) |
---|
336 | IF (RCNTRL(2) == ZERO) THEN |
---|
337 | Hmax = ABS(Tend-Tstart) |
---|
338 | ELSEIF (RCNTRL(2) > ZERO) THEN |
---|
339 | Hmax = MIN(ABS(RCNTRL(2)),ABS(Tend-Tstart)) |
---|
340 | ELSE |
---|
341 | PRINT * , 'User-selected Hmax: RCNTRL(2)=', RCNTRL(2) |
---|
342 | CALL ros_ErrorMsg(-3,Tstart,ZERO,IERR) |
---|
343 | RETURN |
---|
344 | END IF |
---|
345 | !~~~> Starting step size: (positive value) |
---|
346 | IF (RCNTRL(3) == ZERO) THEN |
---|
347 | Hstart = MAX(Hmin,DeltaMin) |
---|
348 | ELSEIF (RCNTRL(3) > ZERO) THEN |
---|
349 | Hstart = MIN(ABS(RCNTRL(3)),ABS(Tend-Tstart)) |
---|
350 | ELSE |
---|
351 | PRINT * , 'User-selected Hstart: RCNTRL(3)=', RCNTRL(3) |
---|
352 | CALL ros_ErrorMsg(-3,Tstart,ZERO,IERR) |
---|
353 | RETURN |
---|
354 | END IF |
---|
355 | !~~~> Step size can be changed s.t. FacMin < Hnew/Hexit < FacMax |
---|
356 | IF (RCNTRL(4) == ZERO) THEN |
---|
357 | FacMin = 0.2d0 |
---|
358 | ELSEIF (RCNTRL(4) > ZERO) THEN |
---|
359 | FacMin = RCNTRL(4) |
---|
360 | ELSE |
---|
361 | PRINT * , 'User-selected FacMin: RCNTRL(4)=', RCNTRL(4) |
---|
362 | CALL ros_ErrorMsg(-4,Tstart,ZERO,IERR) |
---|
363 | RETURN |
---|
364 | END IF |
---|
365 | IF (RCNTRL(5) == ZERO) THEN |
---|
366 | FacMax = 6.0d0 |
---|
367 | ELSEIF (RCNTRL(5) > ZERO) THEN |
---|
368 | FacMax = RCNTRL(5) |
---|
369 | ELSE |
---|
370 | PRINT * , 'User-selected FacMax: RCNTRL(5)=', RCNTRL(5) |
---|
371 | CALL ros_ErrorMsg(-4,Tstart,ZERO,IERR) |
---|
372 | RETURN |
---|
373 | END IF |
---|
374 | !~~~> FacRej: Factor to decrease step after 2 succesive rejections |
---|
375 | IF (RCNTRL(6) == ZERO) THEN |
---|
376 | FacRej = 0.1d0 |
---|
377 | ELSEIF (RCNTRL(6) > ZERO) THEN |
---|
378 | FacRej = RCNTRL(6) |
---|
379 | ELSE |
---|
380 | PRINT * , 'User-selected FacRej: RCNTRL(6)=', RCNTRL(6) |
---|
381 | CALL ros_ErrorMsg(-4,Tstart,ZERO,IERR) |
---|
382 | RETURN |
---|
383 | END IF |
---|
384 | !~~~> FacSafe: Safety Factor in the computation of new step size |
---|
385 | IF (RCNTRL(7) == ZERO) THEN |
---|
386 | FacSafe = 0.9d0 |
---|
387 | ELSEIF (RCNTRL(7) > ZERO) THEN |
---|
388 | FacSafe = RCNTRL(7) |
---|
389 | ELSE |
---|
390 | PRINT * , 'User-selected FacSafe: RCNTRL(7)=', RCNTRL(7) |
---|
391 | CALL ros_ErrorMsg(-4,Tstart,ZERO,IERR) |
---|
392 | RETURN |
---|
393 | END IF |
---|
394 | !~~~> Check if tolerances are reasonable |
---|
395 | DO i=1,UplimTol |
---|
396 | IF ( (AbsTol(i) <= ZERO) .OR. (RelTol(i) <= 10.d0*Roundoff) & |
---|
397 | .OR. (RelTol(i) >= 1.0d0) ) THEN |
---|
398 | PRINT * , ' AbsTol(',i,') = ',AbsTol(i) |
---|
399 | PRINT * , ' RelTol(',i,') = ',RelTol(i) |
---|
400 | CALL ros_ErrorMsg(-5,Tstart,ZERO,IERR) |
---|
401 | RETURN |
---|
402 | END IF |
---|
403 | END DO |
---|
404 | |
---|
405 | |
---|
406 | !~~~> CALL Rosenbrock method |
---|
407 | CALL ros_TLM_Int(Y, NTLM, Y_tlm, & |
---|
408 | Tstart, Tend, Texit, & |
---|
409 | ! Error indicator |
---|
410 | IERR) |
---|
411 | |
---|
412 | |
---|
413 | CONTAINS ! Procedures internal to RosenbrockTLM |
---|
414 | |
---|
415 | |
---|
416 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
417 | SUBROUTINE ros_ErrorMsg(Code,T,H,IERR) |
---|
418 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
419 | ! Handles all error messages |
---|
420 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
421 | |
---|
422 | KPP_REAL, INTENT(IN) :: T, H |
---|
423 | INTEGER, INTENT(IN) :: Code |
---|
424 | INTEGER, INTENT(OUT) :: IERR |
---|
425 | |
---|
426 | IERR = Code |
---|
427 | PRINT * , & |
---|
428 | 'Forced exit from Rosenbrock due to the following error:' |
---|
429 | |
---|
430 | SELECT CASE (Code) |
---|
431 | CASE (-1) |
---|
432 | PRINT * , '--> Improper value for maximal no of steps' |
---|
433 | CASE (-2) |
---|
434 | PRINT * , '--> Selected Rosenbrock method not implemented' |
---|
435 | CASE (-3) |
---|
436 | PRINT * , '--> Hmin/Hmax/Hstart must be positive' |
---|
437 | CASE (-4) |
---|
438 | PRINT * , '--> FacMin/FacMax/FacRej must be positive' |
---|
439 | CASE (-5) |
---|
440 | PRINT * , '--> Improper tolerance values' |
---|
441 | CASE (-6) |
---|
442 | PRINT * , '--> No of steps exceeds maximum bound' |
---|
443 | CASE (-7) |
---|
444 | PRINT * , '--> Step size too small: T + 10*H = T', & |
---|
445 | ' or H < Roundoff' |
---|
446 | CASE (-8) |
---|
447 | PRINT * , '--> Matrix is repeatedly singular' |
---|
448 | CASE DEFAULT |
---|
449 | PRINT *, 'Unknown Error code: ', Code |
---|
450 | END SELECT |
---|
451 | |
---|
452 | PRINT *, "T=", T, "and H=", H |
---|
453 | |
---|
454 | END SUBROUTINE ros_ErrorMsg |
---|
455 | |
---|
456 | |
---|
457 | |
---|
458 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
459 | SUBROUTINE ros_TLM_Int (Y, NTLM, Y_tlm, & |
---|
460 | Tstart, Tend, T, & |
---|
461 | !~~~> Error indicator |
---|
462 | IERR ) |
---|
463 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
464 | ! Template for the implementation of a generic Rosenbrock method |
---|
465 | ! defined by ros_S (no of stages) |
---|
466 | ! and its coefficients ros_{A,C,M,E,Alpha,Gamma} |
---|
467 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
468 | |
---|
469 | IMPLICIT NONE |
---|
470 | |
---|
471 | !~~~> Input: the initial condition at Tstart; Output: the solution at T |
---|
472 | KPP_REAL, INTENT(INOUT) :: Y(N) |
---|
473 | !~~~> Input: Number of sensitivity coefficients |
---|
474 | INTEGER, INTENT(IN) :: NTLM |
---|
475 | !~~~> Input: the initial sensitivites at Tstart; Output: the sensitivities at T |
---|
476 | KPP_REAL, INTENT(INOUT) :: Y_tlm(N,NTLM) |
---|
477 | !~~~> Input: integration interval |
---|
478 | KPP_REAL, INTENT(IN) :: Tstart,Tend |
---|
479 | !~~~> Output: time at which the solution is returned (T=Tend if success) |
---|
480 | KPP_REAL, INTENT(OUT) :: T |
---|
481 | !~~~> Output: Error indicator |
---|
482 | INTEGER, INTENT(OUT) :: IERR |
---|
483 | ! ~~~~ Local variables |
---|
484 | KPP_REAL :: Ynew(N), Fcn0(N), Fcn(N) |
---|
485 | KPP_REAL :: K(N*ros_S) |
---|
486 | KPP_REAL :: Ynew_tlm(N,NTLM), Fcn0_tlm(N,NTLM), Fcn_tlm(N,NTLM) |
---|
487 | KPP_REAL :: K_tlm(N*ros_S,NTLM) |
---|
488 | KPP_REAL :: Hes0(NHESS), Tmp(N) |
---|
489 | KPP_REAL :: dFdT(N), dJdT(LU_NONZERO) |
---|
490 | KPP_REAL :: Jac0(LU_NONZERO), Jac(LU_NONZERO), Ghimj(LU_NONZERO) |
---|
491 | KPP_REAL :: H, Hnew, HC, HG, Fac, Tau |
---|
492 | KPP_REAL :: Err, Err0, Err1, Yerr(N), Yerr_tlm(N,NTLM) |
---|
493 | INTEGER :: Pivot(N), Direction, ioffset, j, istage, itlm |
---|
494 | LOGICAL :: RejectLastH, RejectMoreH, Singular |
---|
495 | !~~~> Local parameters |
---|
496 | KPP_REAL, PARAMETER :: DeltaMin = 1.0d-5 |
---|
497 | !~~~> Locally called functions |
---|
498 | ! KPP_REAL WLAMCH |
---|
499 | ! EXTERNAL WLAMCH |
---|
500 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
501 | |
---|
502 | |
---|
503 | !~~~> Initial preparations |
---|
504 | T = Tstart |
---|
505 | RSTATUS(Nhexit) = ZERO |
---|
506 | H = MIN( MAX(ABS(Hmin),ABS(Hstart)) , ABS(Hmax) ) |
---|
507 | IF (ABS(H) <= 10.D0*Roundoff) H = DeltaMin |
---|
508 | |
---|
509 | IF (Tend >= Tstart) THEN |
---|
510 | Direction = +1 |
---|
511 | ELSE |
---|
512 | Direction = -1 |
---|
513 | END IF |
---|
514 | H = Direction*H |
---|
515 | |
---|
516 | RejectLastH=.FALSE. |
---|
517 | RejectMoreH=.FALSE. |
---|
518 | |
---|
519 | !~~~> Time loop begins below |
---|
520 | |
---|
521 | TimeLoop: DO WHILE ( (Direction > 0).AND.((T-Tend)+Roundoff <= ZERO) & |
---|
522 | .OR. (Direction < 0).AND.((Tend-T)+Roundoff <= ZERO) ) |
---|
523 | |
---|
524 | IF ( ISTATUS(Nstp) > Max_no_steps ) THEN ! Too many steps |
---|
525 | CALL ros_ErrorMsg(-6,T,H,IERR) |
---|
526 | RETURN |
---|
527 | END IF |
---|
528 | IF ( ((T+0.1d0*H) == T).OR.(H <= Roundoff) ) THEN ! Step size too small |
---|
529 | CALL ros_ErrorMsg(-7,T,H,IERR) |
---|
530 | RETURN |
---|
531 | END IF |
---|
532 | |
---|
533 | !~~~> Limit H if necessary to avoid going beyond Tend |
---|
534 | Hexit = H |
---|
535 | H = MIN(H,ABS(Tend-T)) |
---|
536 | |
---|
537 | !~~~> Compute the function at current time |
---|
538 | CALL FunTemplate(T,Y,Fcn0) |
---|
539 | ISTATUS(Nfun) = ISTATUS(Nfun) + 1 |
---|
540 | |
---|
541 | !~~~> Compute the Jacobian at current time |
---|
542 | CALL JacTemplate(T,Y,Jac0) |
---|
543 | ISTATUS(Njac) = ISTATUS(Njac) + 1 |
---|
544 | |
---|
545 | !~~~> Compute the Hessian at current time |
---|
546 | CALL HessTemplate(T,Y,Hes0) |
---|
547 | ISTATUS(Nhes) = ISTATUS(Nhes) + 1 |
---|
548 | |
---|
549 | !~~~> Compute the TLM function at current time |
---|
550 | DO itlm = 1, NTLM |
---|
551 | CALL Jac_SP_Vec ( Jac0, Y_tlm(1,itlm), Fcn0_tlm(1,itlm) ) |
---|
552 | END DO |
---|
553 | |
---|
554 | !~~~> Compute the function and Jacobian derivatives with respect to T |
---|
555 | IF (.NOT.Autonomous) THEN |
---|
556 | CALL ros_FunTimeDerivative ( T, Roundoff, Y, Fcn0, dFdT ) |
---|
557 | CALL ros_JacTimeDerivative ( T, Roundoff, Y, Jac0, dJdT ) |
---|
558 | END IF |
---|
559 | |
---|
560 | !~~~> Repeat step calculation until current step accepted |
---|
561 | UntilAccepted: DO |
---|
562 | |
---|
563 | CALL ros_PrepareMatrix(H,Direction,ros_Gamma(1),& |
---|
564 | Jac0,Ghimj,Pivot,Singular) |
---|
565 | IF (Singular) THEN ! More than 5 consecutive failed decompositions |
---|
566 | CALL ros_ErrorMsg(-8,T,H,IERR) |
---|
567 | RETURN |
---|
568 | END IF |
---|
569 | |
---|
570 | !~~~> Compute the stages |
---|
571 | Stage: DO istage = 1, ros_S |
---|
572 | |
---|
573 | ! Current istage offset. Current istage vector is K(ioffset+1:ioffset+N) |
---|
574 | ioffset = N*(istage-1) |
---|
575 | |
---|
576 | ! Initialize stage solution |
---|
577 | CALL WCOPY(N,Y,1,Ynew,1) |
---|
578 | CALL WCOPY(N*NTLM,Y_tlm,1,Ynew_tlm,1) |
---|
579 | |
---|
580 | ! For the 1st istage the function has been computed previously |
---|
581 | IF ( istage == 1 ) THEN |
---|
582 | CALL WCOPY(N,Fcn0,1,Fcn,1) |
---|
583 | CALL WCOPY(N*NTLM,Fcn0_tlm,1,Fcn_tlm,1) |
---|
584 | ! istage>1 and a new function evaluation is needed at the current istage |
---|
585 | ELSEIF ( ros_NewF(istage) ) THEN |
---|
586 | DO j = 1, istage-1 |
---|
587 | CALL WAXPY(N,ros_A((istage-1)*(istage-2)/2+j), & |
---|
588 | K(N*(j-1)+1),1,Ynew,1) |
---|
589 | DO itlm=1,NTLM |
---|
590 | CALL WAXPY(N,ros_A((istage-1)*(istage-2)/2+j), & |
---|
591 | K_tlm(N*(j-1)+1,itlm),1,Ynew_tlm(1,itlm),1) |
---|
592 | END DO |
---|
593 | END DO |
---|
594 | Tau = T + ros_Alpha(istage)*Direction*H |
---|
595 | CALL FunTemplate(Tau,Ynew,Fcn) |
---|
596 | ISTATUS(Nfun) = ISTATUS(Nfun) + 1 |
---|
597 | CALL JacTemplate(Tau,Ynew,Jac) |
---|
598 | ISTATUS(Njac) = ISTATUS(Njac) + 1 |
---|
599 | DO itlm=1,NTLM |
---|
600 | CALL Jac_SP_Vec ( Jac, Ynew_tlm(1,itlm), Fcn_tlm(1,itlm) ) |
---|
601 | END DO |
---|
602 | END IF ! if istage == 1 elseif ros_NewF(istage) |
---|
603 | CALL WCOPY(N,Fcn,1,K(ioffset+1),1) |
---|
604 | DO itlm=1,NTLM |
---|
605 | CALL WCOPY(N,Fcn_tlm(1,itlm),1,K_tlm(ioffset+1,itlm),1) |
---|
606 | END DO |
---|
607 | DO j = 1, istage-1 |
---|
608 | HC = ros_C((istage-1)*(istage-2)/2+j)/(Direction*H) |
---|
609 | CALL WAXPY(N,HC,K(N*(j-1)+1),1,K(ioffset+1),1) |
---|
610 | DO itlm=1,NTLM |
---|
611 | CALL WAXPY(N,HC,K_tlm(N*(j-1)+1,itlm),1,K_tlm(ioffset+1,itlm),1) |
---|
612 | END DO |
---|
613 | END DO |
---|
614 | IF ((.NOT. Autonomous).AND.(ros_Gamma(istage).NE.ZERO)) THEN |
---|
615 | HG = Direction*H*ros_Gamma(istage) |
---|
616 | CALL WAXPY(N,HG,dFdT,1,K(ioffset+1),1) |
---|
617 | DO itlm=1,NTLM |
---|
618 | CALL Jac_SP_Vec ( dJdT, Ynew_tlm(1,itlm), Tmp ) |
---|
619 | CALL WAXPY(N,HG,Tmp,1,K_tlm(ioffset+1,itlm),1) |
---|
620 | END DO |
---|
621 | END IF |
---|
622 | CALL ros_Solve(Ghimj, Pivot, K(ioffset+1)) |
---|
623 | DO itlm=1,NTLM |
---|
624 | CALL Hess_Vec ( Hes0, K(ioffset+1), Y_tlm(1,itlm), Tmp ) |
---|
625 | CALL WAXPY(N,ONE,Tmp,1,K_tlm(ioffset+1,itlm),1) |
---|
626 | CALL ros_Solve(Ghimj, Pivot, K_tlm(ioffset+1,itlm)) |
---|
627 | END DO |
---|
628 | |
---|
629 | END DO Stage |
---|
630 | |
---|
631 | |
---|
632 | !~~~> Compute the new solution |
---|
633 | CALL WCOPY(N,Y,1,Ynew,1) |
---|
634 | DO j=1,ros_S |
---|
635 | CALL WAXPY(N,ros_M(j),K(N*(j-1)+1),1,Ynew,1) |
---|
636 | END DO |
---|
637 | DO itlm=1,NTLM |
---|
638 | CALL WCOPY(N,Y_tlm(1,itlm),1,Ynew_tlm(1,itlm),1) |
---|
639 | DO j=1,ros_S |
---|
640 | CALL WAXPY(N,ros_M(j),K_tlm(N*(j-1)+1,itlm),1,Ynew_tlm(1,itlm),1) |
---|
641 | END DO |
---|
642 | END DO |
---|
643 | |
---|
644 | !~~~> Compute the error estimation |
---|
645 | CALL Set2zero(N,Yerr) |
---|
646 | DO j=1,ros_S |
---|
647 | CALL WAXPY(N,ros_E(j),K(N*(j-1)+1),1,Yerr,1) |
---|
648 | END DO |
---|
649 | Err = ros_ErrorNorm ( Y, Ynew, Yerr, AbsTol, RelTol, VectorTol ) |
---|
650 | IF (TLMtruncErr) THEN |
---|
651 | Err1 = 0.0d0 |
---|
652 | CALL Set2zero(N*NTLM,Yerr_tlm) |
---|
653 | DO itlm=1,NTLM |
---|
654 | DO j=1,ros_S |
---|
655 | CALL WAXPY(N,ros_E(j),K_tlm(N*(j-1)+1,itlm),1,Yerr_tlm(1,itlm),1) |
---|
656 | END DO |
---|
657 | END DO |
---|
658 | Err = ros_ErrorNorm_tlm(Y_tlm,Ynew_tlm,Yerr_tlm,AbsTol_tlm,RelTol_tlm,Err,VectorTol) |
---|
659 | END IF |
---|
660 | |
---|
661 | !~~~> New step size is bounded by FacMin <= Hnew/H <= FacMax |
---|
662 | Fac = MIN(FacMax,MAX(FacMin,FacSafe/Err**(ONE/ros_ELO))) |
---|
663 | Hnew = H*Fac |
---|
664 | |
---|
665 | !~~~> Check the error magnitude and adjust step size |
---|
666 | ISTATUS(Nstp) = ISTATUS(Nstp) + 1 |
---|
667 | IF ( (Err <= ONE).OR.(H <= Hmin) ) THEN !~~~> Accept step |
---|
668 | ISTATUS(Nacc) = ISTATUS(Nacc) + 1 |
---|
669 | CALL WCOPY(N,Ynew,1,Y,1) |
---|
670 | CALL WCOPY(N*NTLM,Ynew_tlm,1,Y_tlm,1) |
---|
671 | T = T + Direction*H |
---|
672 | Hnew = MAX(Hmin,MIN(Hnew,Hmax)) |
---|
673 | IF (RejectLastH) THEN ! No step size increase after a rejected step |
---|
674 | Hnew = MIN(Hnew,H) |
---|
675 | END IF |
---|
676 | RSTATUS(Nhexit) = H |
---|
677 | RSTATUS(Nhnew) = Hnew |
---|
678 | RSTATUS(Ntexit) = T |
---|
679 | RejectLastH = .FALSE. |
---|
680 | RejectMoreH = .FALSE. |
---|
681 | H = Hnew |
---|
682 | EXIT UntilAccepted ! EXIT THE LOOP: WHILE STEP NOT ACCEPTED |
---|
683 | ELSE !~~~> Reject step |
---|
684 | IF (RejectMoreH) THEN |
---|
685 | Hnew = H*FacRej |
---|
686 | END IF |
---|
687 | RejectMoreH = RejectLastH |
---|
688 | RejectLastH = .TRUE. |
---|
689 | H = Hnew |
---|
690 | IF (ISTATUS(Nacc) >= 1) THEN |
---|
691 | ISTATUS(Nrej) = ISTATUS(Nrej) + 1 |
---|
692 | END IF |
---|
693 | END IF ! Err <= 1 |
---|
694 | |
---|
695 | END DO UntilAccepted |
---|
696 | |
---|
697 | END DO TimeLoop |
---|
698 | |
---|
699 | !~~~> Succesful exit |
---|
700 | IERR = 1 !~~~> The integration was successful |
---|
701 | |
---|
702 | END SUBROUTINE ros_TLM_Int |
---|
703 | |
---|
704 | |
---|
705 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
706 | KPP_REAL FUNCTION ros_ErrorNorm ( Y, Ynew, Yerr, & |
---|
707 | AbsTol, RelTol, VectorTol ) |
---|
708 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
709 | !~~~> Computes the "scaled norm" of the error vector Yerr |
---|
710 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
711 | IMPLICIT NONE |
---|
712 | |
---|
713 | ! Input arguments |
---|
714 | KPP_REAL, INTENT(IN) :: Y(N), Ynew(N), & |
---|
715 | Yerr(N), AbsTol(N), RelTol(N) |
---|
716 | LOGICAL, INTENT(IN) :: VectorTol |
---|
717 | ! Local variables |
---|
718 | KPP_REAL :: Err, Scale, Ymax |
---|
719 | INTEGER :: i |
---|
720 | KPP_REAL, PARAMETER :: ZERO = 0.0d0 |
---|
721 | |
---|
722 | Err = ZERO |
---|
723 | DO i=1,N |
---|
724 | Ymax = MAX(ABS(Y(i)),ABS(Ynew(i))) |
---|
725 | IF (VectorTol) THEN |
---|
726 | Scale = AbsTol(i)+RelTol(i)*Ymax |
---|
727 | ELSE |
---|
728 | Scale = AbsTol(1)+RelTol(1)*Ymax |
---|
729 | END IF |
---|
730 | Err = Err+(Yerr(i)/Scale)**2 |
---|
731 | END DO |
---|
732 | Err = SQRT(Err/N) |
---|
733 | |
---|
734 | ros_ErrorNorm = MAX(Err,1.0d-10) |
---|
735 | |
---|
736 | END FUNCTION ros_ErrorNorm |
---|
737 | |
---|
738 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
739 | KPP_REAL FUNCTION ros_ErrorNorm_tlm ( Y_tlm, Ynew_tlm, Yerr_tlm, & |
---|
740 | AbsTol_tlm, RelTol_tlm, Fwd_Err, VectorTol ) |
---|
741 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
742 | !~~~> Computes the "scaled norm" of the error vector Yerr_tlm |
---|
743 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
744 | IMPLICIT NONE |
---|
745 | |
---|
746 | ! Input arguments |
---|
747 | KPP_REAL, INTENT(IN) :: Y_tlm(N,NTLM), Ynew_tlm(N,NTLM), & |
---|
748 | Yerr_tlm(N,NTLM), AbsTol_tlm(N,NTLM), RelTol_tlm(N,NTLM), Fwd_Err |
---|
749 | LOGICAL, INTENT(IN) :: VectorTol |
---|
750 | ! Local variables |
---|
751 | KPP_REAL :: TMP, Err |
---|
752 | INTEGER :: itlm |
---|
753 | |
---|
754 | Err = FWD_Err |
---|
755 | DO itlm = 1,NTLM |
---|
756 | TMP = ros_ErrorNorm(Y_tlm(1,itlm), Ynew_tlm(1,itlm),Yerr_tlm(1,itlm), & |
---|
757 | AbsTol_tlm(1,itlm), RelTol_tlm(1,itlm), VectorTol) |
---|
758 | Err = MAX(Err, TMP) |
---|
759 | END DO |
---|
760 | |
---|
761 | ros_ErrorNorm_tlm = MAX(Err,1.0d-10) |
---|
762 | |
---|
763 | END FUNCTION ros_ErrorNorm_tlm |
---|
764 | |
---|
765 | |
---|
766 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
767 | SUBROUTINE ros_FunTimeDerivative ( T, Roundoff, Y, & |
---|
768 | Fcn0, dFdT ) |
---|
769 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
770 | !~~~> The time partial derivative of the function by finite differences |
---|
771 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
772 | IMPLICIT NONE |
---|
773 | |
---|
774 | !~~~> Input arguments |
---|
775 | KPP_REAL, INTENT(IN) :: T, Roundoff, Y(N), Fcn0(N) |
---|
776 | !~~~> Output arguments |
---|
777 | KPP_REAL, INTENT(OUT) :: dFdT(N) |
---|
778 | !~~~> Local variables |
---|
779 | KPP_REAL :: Delta |
---|
780 | KPP_REAL, PARAMETER :: DeltaMin = 1.0d-6 |
---|
781 | |
---|
782 | Delta = SQRT(Roundoff)*MAX(DeltaMin,ABS(T)) |
---|
783 | CALL FunTemplate(T+Delta,Y,dFdT) |
---|
784 | ISTATUS(Nfun) = ISTATUS(Nfun) + 1 |
---|
785 | CALL WAXPY(N,(-ONE),Fcn0,1,dFdT,1) |
---|
786 | CALL WSCAL(N,(ONE/Delta),dFdT,1) |
---|
787 | |
---|
788 | END SUBROUTINE ros_FunTimeDerivative |
---|
789 | |
---|
790 | |
---|
791 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
792 | SUBROUTINE ros_JacTimeDerivative ( T, Roundoff, Y, & |
---|
793 | Jac0, dJdT ) |
---|
794 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
795 | !~~~> The time partial derivative of the Jacobian by finite differences |
---|
796 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
797 | IMPLICIT NONE |
---|
798 | |
---|
799 | !~~~> Input arguments |
---|
800 | KPP_REAL, INTENT(IN) :: T, Roundoff, Y(N), Jac0(LU_NONZERO) |
---|
801 | !~~~> Output arguments |
---|
802 | KPP_REAL, INTENT(OUT) :: dJdT(LU_NONZERO) |
---|
803 | !~~~> Local variables |
---|
804 | KPP_REAL Delta |
---|
805 | KPP_REAL, PARAMETER :: ONE = 1.0d0, DeltaMin = 1.0d-6 |
---|
806 | |
---|
807 | Delta = SQRT(Roundoff)*MAX(DeltaMin,ABS(T)) |
---|
808 | CALL JacTemplate(T+Delta,Y,dJdT) |
---|
809 | ISTATUS(Njac) = ISTATUS(Njac) + 1 |
---|
810 | CALL WAXPY(LU_NONZERO,(-ONE),Jac0,1,dJdT,1) |
---|
811 | CALL WSCAL(LU_NONZERO,(ONE/Delta),dJdT,1) |
---|
812 | |
---|
813 | END SUBROUTINE ros_JacTimeDerivative |
---|
814 | |
---|
815 | |
---|
816 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
817 | SUBROUTINE ros_PrepareMatrix ( H, Direction, gam, & |
---|
818 | Jac0, Ghimj, Pivot, Singular ) |
---|
819 | ! --- --- --- --- --- --- --- --- --- --- --- --- --- |
---|
820 | ! Prepares the LHS matrix for stage calculations |
---|
821 | ! 1. Construct Ghimj = 1/(H*ham) - Jac0 |
---|
822 | ! "(Gamma H) Inverse Minus Jacobian" |
---|
823 | ! 2. Repeat LU decomposition of Ghimj until successful. |
---|
824 | ! -half the step size if LU decomposition fails and retry |
---|
825 | ! -exit after 5 consecutive fails |
---|
826 | ! --- --- --- --- --- --- --- --- --- --- --- --- --- |
---|
827 | IMPLICIT NONE |
---|
828 | |
---|
829 | !~~~> Input arguments |
---|
830 | KPP_REAL, INTENT(IN) :: gam, Jac0(LU_NONZERO) |
---|
831 | INTEGER, INTENT(IN) :: Direction |
---|
832 | !~~~> Output arguments |
---|
833 | KPP_REAL, INTENT(OUT) :: Ghimj(LU_NONZERO) |
---|
834 | LOGICAL, INTENT(OUT) :: Singular |
---|
835 | INTEGER, INTENT(OUT) :: Pivot(N) |
---|
836 | !~~~> Inout arguments |
---|
837 | KPP_REAL, INTENT(INOUT) :: H ! step size is decreased when LU fails |
---|
838 | !~~~> Local variables |
---|
839 | INTEGER :: i, ISING, Nconsecutive |
---|
840 | KPP_REAL :: ghinv |
---|
841 | KPP_REAL, PARAMETER :: ONE = 1.0d0, HALF = 0.5d0 |
---|
842 | |
---|
843 | Nconsecutive = 0 |
---|
844 | Singular = .TRUE. |
---|
845 | |
---|
846 | DO WHILE (Singular) |
---|
847 | |
---|
848 | !~~~> Construct Ghimj = 1/(H*ham) - Jac0 |
---|
849 | CALL WCOPY(LU_NONZERO,Jac0,1,Ghimj,1) |
---|
850 | CALL WSCAL(LU_NONZERO,(-ONE),Ghimj,1) |
---|
851 | ghinv = ONE/(Direction*H*gam) |
---|
852 | DO i=1,N |
---|
853 | Ghimj(LU_DIAG(i)) = Ghimj(LU_DIAG(i))+ghinv |
---|
854 | END DO |
---|
855 | !~~~> Compute LU decomposition |
---|
856 | CALL ros_Decomp( Ghimj, Pivot, ISING ) |
---|
857 | IF (ISING == 0) THEN |
---|
858 | !~~~> If successful done |
---|
859 | Singular = .FALSE. |
---|
860 | ELSE ! ISING .ne. 0 |
---|
861 | !~~~> If unsuccessful half the step size; if 5 consecutive fails then return |
---|
862 | ISTATUS(Nsng) = ISTATUS(Nsng) + 1 |
---|
863 | Nconsecutive = Nconsecutive+1 |
---|
864 | Singular = .TRUE. |
---|
865 | PRINT*,'Warning: LU Decomposition returned ISING = ',ISING |
---|
866 | IF (Nconsecutive <= 5) THEN ! Less than 5 consecutive failed decompositions |
---|
867 | H = H*HALF |
---|
868 | ELSE ! More than 5 consecutive failed decompositions |
---|
869 | RETURN |
---|
870 | END IF ! Nconsecutive |
---|
871 | END IF ! ISING |
---|
872 | |
---|
873 | END DO ! WHILE Singular |
---|
874 | |
---|
875 | END SUBROUTINE ros_PrepareMatrix |
---|
876 | |
---|
877 | |
---|
878 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
879 | SUBROUTINE ros_Decomp( A, Pivot, ISING ) |
---|
880 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
881 | ! Template for the LU decomposition |
---|
882 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
883 | IMPLICIT NONE |
---|
884 | !~~~> Inout variables |
---|
885 | KPP_REAL, INTENT(INOUT) :: A(LU_NONZERO) |
---|
886 | !~~~> Output variables |
---|
887 | INTEGER, INTENT(OUT) :: Pivot(N), ISING |
---|
888 | |
---|
889 | CALL KppDecomp ( A, ISING ) |
---|
890 | !~~~> Note: for a full matrix use Lapack: |
---|
891 | ! CALL DGETRF( N, N, A, N, Pivot, ISING ) |
---|
892 | Pivot(1) = 1 |
---|
893 | |
---|
894 | ISTATUS(Ndec) = ISTATUS(Ndec) + 1 |
---|
895 | |
---|
896 | END SUBROUTINE ros_Decomp |
---|
897 | |
---|
898 | |
---|
899 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
900 | SUBROUTINE ros_Solve( A, Pivot, b ) |
---|
901 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
902 | ! Template for the forward/backward substitution (using pre-computed LU decomposition) |
---|
903 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
904 | IMPLICIT NONE |
---|
905 | !~~~> Input variables |
---|
906 | #ifdef FULL_ALGEBRA |
---|
907 | KPP_REAL, INTENT(IN) :: A(N,N) |
---|
908 | INTEGER :: ISING |
---|
909 | #else |
---|
910 | KPP_REAL, INTENT(IN) :: A(LU_NONZERO) |
---|
911 | #endif |
---|
912 | INTEGER, INTENT(IN) :: Pivot(N) |
---|
913 | !~~~> InOut variables |
---|
914 | KPP_REAL, INTENT(INOUT) :: b(N) |
---|
915 | |
---|
916 | #ifdef FULL_ALGEBRA |
---|
917 | CALL DGETRS( 'N', N , 1, A, N, Pivot, b, N, ISING ) |
---|
918 | #else |
---|
919 | CALL KppSolve( A, b ) |
---|
920 | #endif |
---|
921 | |
---|
922 | ISTATUS(Nsol) = ISTATUS(Nsol) + 1 |
---|
923 | |
---|
924 | END SUBROUTINE ros_Solve |
---|
925 | |
---|
926 | |
---|
927 | |
---|
928 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
929 | SUBROUTINE Ros2 |
---|
930 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
931 | ! --- AN L-STABLE METHOD, 2 stages, order 2 |
---|
932 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
933 | |
---|
934 | IMPLICIT NONE |
---|
935 | DOUBLE PRECISION g |
---|
936 | |
---|
937 | g = 1.0d0 + 1.0d0/SQRT(2.0d0) |
---|
938 | |
---|
939 | rosMethod = RS2 |
---|
940 | !~~~> Name of the method |
---|
941 | ros_Name = 'ROS-2' |
---|
942 | !~~~> Number of stages |
---|
943 | ros_S = 2 |
---|
944 | |
---|
945 | !~~~> The coefficient matrices A and C are strictly lower triangular. |
---|
946 | ! The lower triangular (subdiagonal) elements are stored in row-wise order: |
---|
947 | ! A(2,1) = ros_A(1), A(3,1)=ros_A(2), A(3,2)=ros_A(3), etc. |
---|
948 | ! The general mapping formula is: |
---|
949 | ! A(i,j) = ros_A( (i-1)*(i-2)/2 + j ) |
---|
950 | ! C(i,j) = ros_C( (i-1)*(i-2)/2 + j ) |
---|
951 | |
---|
952 | ros_A(1) = (1.d0)/g |
---|
953 | ros_C(1) = (-2.d0)/g |
---|
954 | !~~~> Does the stage i require a new function evaluation (ros_NewF(i)=TRUE) |
---|
955 | ! or does it re-use the function evaluation from stage i-1 (ros_NewF(i)=FALSE) |
---|
956 | ros_NewF(1) = .TRUE. |
---|
957 | ros_NewF(2) = .TRUE. |
---|
958 | !~~~> M_i = Coefficients for new step solution |
---|
959 | ros_M(1)= (3.d0)/(2.d0*g) |
---|
960 | ros_M(2)= (1.d0)/(2.d0*g) |
---|
961 | ! E_i = Coefficients for error estimator |
---|
962 | ros_E(1) = 1.d0/(2.d0*g) |
---|
963 | ros_E(2) = 1.d0/(2.d0*g) |
---|
964 | !~~~> ros_ELO = estimator of local order - the minimum between the |
---|
965 | ! main and the embedded scheme orders plus one |
---|
966 | ros_ELO = 2.0d0 |
---|
967 | !~~~> Y_stage_i ~ Y( T + H*Alpha_i ) |
---|
968 | ros_Alpha(1) = 0.0d0 |
---|
969 | ros_Alpha(2) = 1.0d0 |
---|
970 | !~~~> Gamma_i = \sum_j gamma_{i,j} |
---|
971 | ros_Gamma(1) = g |
---|
972 | ros_Gamma(2) =-g |
---|
973 | |
---|
974 | END SUBROUTINE Ros2 |
---|
975 | |
---|
976 | |
---|
977 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
978 | SUBROUTINE Ros3 |
---|
979 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
980 | ! --- AN L-STABLE METHOD, 3 stages, order 3, 2 function evaluations |
---|
981 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
982 | |
---|
983 | IMPLICIT NONE |
---|
984 | |
---|
985 | rosMethod = RS3 |
---|
986 | !~~~> Name of the method |
---|
987 | ros_Name = 'ROS-3' |
---|
988 | !~~~> Number of stages |
---|
989 | ros_S = 3 |
---|
990 | |
---|
991 | !~~~> The coefficient matrices A and C are strictly lower triangular. |
---|
992 | ! The lower triangular (subdiagonal) elements are stored in row-wise order: |
---|
993 | ! A(2,1) = ros_A(1), A(3,1)=ros_A(2), A(3,2)=ros_A(3), etc. |
---|
994 | ! The general mapping formula is: |
---|
995 | ! A(i,j) = ros_A( (i-1)*(i-2)/2 + j ) |
---|
996 | ! C(i,j) = ros_C( (i-1)*(i-2)/2 + j ) |
---|
997 | |
---|
998 | ros_A(1)= 1.d0 |
---|
999 | ros_A(2)= 1.d0 |
---|
1000 | ros_A(3)= 0.d0 |
---|
1001 | |
---|
1002 | ros_C(1) = -0.10156171083877702091975600115545d+01 |
---|
1003 | ros_C(2) = 0.40759956452537699824805835358067d+01 |
---|
1004 | ros_C(3) = 0.92076794298330791242156818474003d+01 |
---|
1005 | !~~~> Does the stage i require a new function evaluation (ros_NewF(i)=TRUE) |
---|
1006 | ! or does it re-use the function evaluation from stage i-1 (ros_NewF(i)=FALSE) |
---|
1007 | ros_NewF(1) = .TRUE. |
---|
1008 | ros_NewF(2) = .TRUE. |
---|
1009 | ros_NewF(3) = .FALSE. |
---|
1010 | !~~~> M_i = Coefficients for new step solution |
---|
1011 | ros_M(1) = 0.1d+01 |
---|
1012 | ros_M(2) = 0.61697947043828245592553615689730d+01 |
---|
1013 | ros_M(3) = -0.42772256543218573326238373806514d+00 |
---|
1014 | ! E_i = Coefficients for error estimator |
---|
1015 | ros_E(1) = 0.5d+00 |
---|
1016 | ros_E(2) = -0.29079558716805469821718236208017d+01 |
---|
1017 | ros_E(3) = 0.22354069897811569627360909276199d+00 |
---|
1018 | !~~~> ros_ELO = estimator of local order - the minimum between the |
---|
1019 | ! main and the embedded scheme orders plus 1 |
---|
1020 | ros_ELO = 3.0d0 |
---|
1021 | !~~~> Y_stage_i ~ Y( T + H*Alpha_i ) |
---|
1022 | ros_Alpha(1)= 0.0d+00 |
---|
1023 | ros_Alpha(2)= 0.43586652150845899941601945119356d+00 |
---|
1024 | ros_Alpha(3)= 0.43586652150845899941601945119356d+00 |
---|
1025 | !~~~> Gamma_i = \sum_j gamma_{i,j} |
---|
1026 | ros_Gamma(1)= 0.43586652150845899941601945119356d+00 |
---|
1027 | ros_Gamma(2)= 0.24291996454816804366592249683314d+00 |
---|
1028 | ros_Gamma(3)= 0.21851380027664058511513169485832d+01 |
---|
1029 | |
---|
1030 | END SUBROUTINE Ros3 |
---|
1031 | |
---|
1032 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1033 | |
---|
1034 | |
---|
1035 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1036 | SUBROUTINE Ros4 |
---|
1037 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1038 | ! L-STABLE ROSENBROCK METHOD OF ORDER 4, WITH 4 STAGES |
---|
1039 | ! L-STABLE EMBEDDED ROSENBROCK METHOD OF ORDER 3 |
---|
1040 | ! |
---|
1041 | ! E. HAIRER AND G. WANNER, SOLVING ORDINARY DIFFERENTIAL |
---|
1042 | ! EQUATIONS II. STIFF AND DIFFERENTIAL-ALGEBRAIC PROBLEMS. |
---|
1043 | ! SPRINGER SERIES IN COMPUTATIONAL MATHEMATICS, |
---|
1044 | ! SPRINGER-VERLAG (1990) |
---|
1045 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1046 | |
---|
1047 | IMPLICIT NONE |
---|
1048 | |
---|
1049 | rosMethod = RS4 |
---|
1050 | !~~~> Name of the method |
---|
1051 | ros_Name = 'ROS-4' |
---|
1052 | !~~~> Number of stages |
---|
1053 | ros_S = 4 |
---|
1054 | |
---|
1055 | !~~~> The coefficient matrices A and C are strictly lower triangular. |
---|
1056 | ! The lower triangular (subdiagonal) elements are stored in row-wise order: |
---|
1057 | ! A(2,1) = ros_A(1), A(3,1)=ros_A(2), A(3,2)=ros_A(3), etc. |
---|
1058 | ! The general mapping formula is: |
---|
1059 | ! A(i,j) = ros_A( (i-1)*(i-2)/2 + j ) |
---|
1060 | ! C(i,j) = ros_C( (i-1)*(i-2)/2 + j ) |
---|
1061 | |
---|
1062 | ros_A(1) = 0.2000000000000000d+01 |
---|
1063 | ros_A(2) = 0.1867943637803922d+01 |
---|
1064 | ros_A(3) = 0.2344449711399156d+00 |
---|
1065 | ros_A(4) = ros_A(2) |
---|
1066 | ros_A(5) = ros_A(3) |
---|
1067 | ros_A(6) = 0.0D0 |
---|
1068 | |
---|
1069 | ros_C(1) =-0.7137615036412310d+01 |
---|
1070 | ros_C(2) = 0.2580708087951457d+01 |
---|
1071 | ros_C(3) = 0.6515950076447975d+00 |
---|
1072 | ros_C(4) =-0.2137148994382534d+01 |
---|
1073 | ros_C(5) =-0.3214669691237626d+00 |
---|
1074 | ros_C(6) =-0.6949742501781779d+00 |
---|
1075 | !~~~> Does the stage i require a new function evaluation (ros_NewF(i)=TRUE) |
---|
1076 | ! or does it re-use the function evaluation from stage i-1 (ros_NewF(i)=FALSE) |
---|
1077 | ros_NewF(1) = .TRUE. |
---|
1078 | ros_NewF(2) = .TRUE. |
---|
1079 | ros_NewF(3) = .TRUE. |
---|
1080 | ros_NewF(4) = .FALSE. |
---|
1081 | !~~~> M_i = Coefficients for new step solution |
---|
1082 | ros_M(1) = 0.2255570073418735d+01 |
---|
1083 | ros_M(2) = 0.2870493262186792d+00 |
---|
1084 | ros_M(3) = 0.4353179431840180d+00 |
---|
1085 | ros_M(4) = 0.1093502252409163d+01 |
---|
1086 | !~~~> E_i = Coefficients for error estimator |
---|
1087 | ros_E(1) =-0.2815431932141155d+00 |
---|
1088 | ros_E(2) =-0.7276199124938920d-01 |
---|
1089 | ros_E(3) =-0.1082196201495311d+00 |
---|
1090 | ros_E(4) =-0.1093502252409163d+01 |
---|
1091 | !~~~> ros_ELO = estimator of local order - the minimum between the |
---|
1092 | ! main and the embedded scheme orders plus 1 |
---|
1093 | ros_ELO = 4.0d0 |
---|
1094 | !~~~> Y_stage_i ~ Y( T + H*Alpha_i ) |
---|
1095 | ros_Alpha(1) = 0.D0 |
---|
1096 | ros_Alpha(2) = 0.1145640000000000d+01 |
---|
1097 | ros_Alpha(3) = 0.6552168638155900d+00 |
---|
1098 | ros_Alpha(4) = ros_Alpha(3) |
---|
1099 | !~~~> Gamma_i = \sum_j gamma_{i,j} |
---|
1100 | ros_Gamma(1) = 0.5728200000000000d+00 |
---|
1101 | ros_Gamma(2) =-0.1769193891319233d+01 |
---|
1102 | ros_Gamma(3) = 0.7592633437920482d+00 |
---|
1103 | ros_Gamma(4) =-0.1049021087100450d+00 |
---|
1104 | |
---|
1105 | END SUBROUTINE Ros4 |
---|
1106 | |
---|
1107 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1108 | SUBROUTINE Rodas3 |
---|
1109 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1110 | ! --- A STIFFLY-STABLE METHOD, 4 stages, order 3 |
---|
1111 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1112 | |
---|
1113 | IMPLICIT NONE |
---|
1114 | |
---|
1115 | rosMethod = RD3 |
---|
1116 | !~~~> Name of the method |
---|
1117 | ros_Name = 'RODAS-3' |
---|
1118 | !~~~> Number of stages |
---|
1119 | ros_S = 4 |
---|
1120 | |
---|
1121 | !~~~> The coefficient matrices A and C are strictly lower triangular. |
---|
1122 | ! The lower triangular (subdiagonal) elements are stored in row-wise order: |
---|
1123 | ! A(2,1) = ros_A(1), A(3,1)=ros_A(2), A(3,2)=ros_A(3), etc. |
---|
1124 | ! The general mapping formula is: |
---|
1125 | ! A(i,j) = ros_A( (i-1)*(i-2)/2 + j ) |
---|
1126 | ! C(i,j) = ros_C( (i-1)*(i-2)/2 + j ) |
---|
1127 | |
---|
1128 | ros_A(1) = 0.0d+00 |
---|
1129 | ros_A(2) = 2.0d+00 |
---|
1130 | ros_A(3) = 0.0d+00 |
---|
1131 | ros_A(4) = 2.0d+00 |
---|
1132 | ros_A(5) = 0.0d+00 |
---|
1133 | ros_A(6) = 1.0d+00 |
---|
1134 | |
---|
1135 | ros_C(1) = 4.0d+00 |
---|
1136 | ros_C(2) = 1.0d+00 |
---|
1137 | ros_C(3) =-1.0d+00 |
---|
1138 | ros_C(4) = 1.0d+00 |
---|
1139 | ros_C(5) =-1.0d+00 |
---|
1140 | ros_C(6) =-(8.0d+00/3.0d+00) |
---|
1141 | |
---|
1142 | !~~~> Does the stage i require a new function evaluation (ros_NewF(i)=TRUE) |
---|
1143 | ! or does it re-use the function evaluation from stage i-1 (ros_NewF(i)=FALSE) |
---|
1144 | ros_NewF(1) = .TRUE. |
---|
1145 | ros_NewF(2) = .FALSE. |
---|
1146 | ros_NewF(3) = .TRUE. |
---|
1147 | ros_NewF(4) = .TRUE. |
---|
1148 | !~~~> M_i = Coefficients for new step solution |
---|
1149 | ros_M(1) = 2.0d+00 |
---|
1150 | ros_M(2) = 0.0d+00 |
---|
1151 | ros_M(3) = 1.0d+00 |
---|
1152 | ros_M(4) = 1.0d+00 |
---|
1153 | !~~~> E_i = Coefficients for error estimator |
---|
1154 | ros_E(1) = 0.0d+00 |
---|
1155 | ros_E(2) = 0.0d+00 |
---|
1156 | ros_E(3) = 0.0d+00 |
---|
1157 | ros_E(4) = 1.0d+00 |
---|
1158 | !~~~> ros_ELO = estimator of local order - the minimum between the |
---|
1159 | ! main and the embedded scheme orders plus 1 |
---|
1160 | ros_ELO = 3.0d+00 |
---|
1161 | !~~~> Y_stage_i ~ Y( T + H*Alpha_i ) |
---|
1162 | ros_Alpha(1) = 0.0d+00 |
---|
1163 | ros_Alpha(2) = 0.0d+00 |
---|
1164 | ros_Alpha(3) = 1.0d+00 |
---|
1165 | ros_Alpha(4) = 1.0d+00 |
---|
1166 | !~~~> Gamma_i = \sum_j gamma_{i,j} |
---|
1167 | ros_Gamma(1) = 0.5d+00 |
---|
1168 | ros_Gamma(2) = 1.5d+00 |
---|
1169 | ros_Gamma(3) = 0.0d+00 |
---|
1170 | ros_Gamma(4) = 0.0d+00 |
---|
1171 | |
---|
1172 | END SUBROUTINE Rodas3 |
---|
1173 | |
---|
1174 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1175 | SUBROUTINE Rodas4 |
---|
1176 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1177 | ! STIFFLY-STABLE ROSENBROCK METHOD OF ORDER 4, WITH 6 STAGES |
---|
1178 | ! |
---|
1179 | ! E. HAIRER AND G. WANNER, SOLVING ORDINARY DIFFERENTIAL |
---|
1180 | ! EQUATIONS II. STIFF AND DIFFERENTIAL-ALGEBRAIC PROBLEMS. |
---|
1181 | ! SPRINGER SERIES IN COMPUTATIONAL MATHEMATICS, |
---|
1182 | ! SPRINGER-VERLAG (1996) |
---|
1183 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1184 | |
---|
1185 | IMPLICIT NONE |
---|
1186 | |
---|
1187 | rosMethod = RD4 |
---|
1188 | !~~~> Name of the method |
---|
1189 | ros_Name = 'RODAS-4' |
---|
1190 | !~~~> Number of stages |
---|
1191 | ros_S = 6 |
---|
1192 | |
---|
1193 | !~~~> Y_stage_i ~ Y( T + H*Alpha_i ) |
---|
1194 | ros_Alpha(1) = 0.000d0 |
---|
1195 | ros_Alpha(2) = 0.386d0 |
---|
1196 | ros_Alpha(3) = 0.210d0 |
---|
1197 | ros_Alpha(4) = 0.630d0 |
---|
1198 | ros_Alpha(5) = 1.000d0 |
---|
1199 | ros_Alpha(6) = 1.000d0 |
---|
1200 | |
---|
1201 | !~~~> Gamma_i = \sum_j gamma_{i,j} |
---|
1202 | ros_Gamma(1) = 0.2500000000000000d+00 |
---|
1203 | ros_Gamma(2) =-0.1043000000000000d+00 |
---|
1204 | ros_Gamma(3) = 0.1035000000000000d+00 |
---|
1205 | ros_Gamma(4) =-0.3620000000000023d-01 |
---|
1206 | ros_Gamma(5) = 0.0d0 |
---|
1207 | ros_Gamma(6) = 0.0d0 |
---|
1208 | |
---|
1209 | !~~~> The coefficient matrices A and C are strictly lower triangular. |
---|
1210 | ! The lower triangular (subdiagonal) elements are stored in row-wise order: |
---|
1211 | ! A(2,1) = ros_A(1), A(3,1)=ros_A(2), A(3,2)=ros_A(3), etc. |
---|
1212 | ! The general mapping formula is: A(i,j) = ros_A( (i-1)*(i-2)/2 + j ) |
---|
1213 | ! C(i,j) = ros_C( (i-1)*(i-2)/2 + j ) |
---|
1214 | |
---|
1215 | ros_A(1) = 0.1544000000000000d+01 |
---|
1216 | ros_A(2) = 0.9466785280815826d+00 |
---|
1217 | ros_A(3) = 0.2557011698983284d+00 |
---|
1218 | ros_A(4) = 0.3314825187068521d+01 |
---|
1219 | ros_A(5) = 0.2896124015972201d+01 |
---|
1220 | ros_A(6) = 0.9986419139977817d+00 |
---|
1221 | ros_A(7) = 0.1221224509226641d+01 |
---|
1222 | ros_A(8) = 0.6019134481288629d+01 |
---|
1223 | ros_A(9) = 0.1253708332932087d+02 |
---|
1224 | ros_A(10) =-0.6878860361058950d+00 |
---|
1225 | ros_A(11) = ros_A(7) |
---|
1226 | ros_A(12) = ros_A(8) |
---|
1227 | ros_A(13) = ros_A(9) |
---|
1228 | ros_A(14) = ros_A(10) |
---|
1229 | ros_A(15) = 1.0d+00 |
---|
1230 | |
---|
1231 | ros_C(1) =-0.5668800000000000d+01 |
---|
1232 | ros_C(2) =-0.2430093356833875d+01 |
---|
1233 | ros_C(3) =-0.2063599157091915d+00 |
---|
1234 | ros_C(4) =-0.1073529058151375d+00 |
---|
1235 | ros_C(5) =-0.9594562251023355d+01 |
---|
1236 | ros_C(6) =-0.2047028614809616d+02 |
---|
1237 | ros_C(7) = 0.7496443313967647d+01 |
---|
1238 | ros_C(8) =-0.1024680431464352d+02 |
---|
1239 | ros_C(9) =-0.3399990352819905d+02 |
---|
1240 | ros_C(10) = 0.1170890893206160d+02 |
---|
1241 | ros_C(11) = 0.8083246795921522d+01 |
---|
1242 | ros_C(12) =-0.7981132988064893d+01 |
---|
1243 | ros_C(13) =-0.3152159432874371d+02 |
---|
1244 | ros_C(14) = 0.1631930543123136d+02 |
---|
1245 | ros_C(15) =-0.6058818238834054d+01 |
---|
1246 | |
---|
1247 | !~~~> M_i = Coefficients for new step solution |
---|
1248 | ros_M(1) = ros_A(7) |
---|
1249 | ros_M(2) = ros_A(8) |
---|
1250 | ros_M(3) = ros_A(9) |
---|
1251 | ros_M(4) = ros_A(10) |
---|
1252 | ros_M(5) = 1.0d+00 |
---|
1253 | ros_M(6) = 1.0d+00 |
---|
1254 | |
---|
1255 | !~~~> E_i = Coefficients for error estimator |
---|
1256 | ros_E(1) = 0.0d+00 |
---|
1257 | ros_E(2) = 0.0d+00 |
---|
1258 | ros_E(3) = 0.0d+00 |
---|
1259 | ros_E(4) = 0.0d+00 |
---|
1260 | ros_E(5) = 0.0d+00 |
---|
1261 | ros_E(6) = 1.0d+00 |
---|
1262 | |
---|
1263 | !~~~> Does the stage i require a new function evaluation (ros_NewF(i)=TRUE) |
---|
1264 | ! or does it re-use the function evaluation from stage i-1 (ros_NewF(i)=FALSE) |
---|
1265 | ros_NewF(1) = .TRUE. |
---|
1266 | ros_NewF(2) = .TRUE. |
---|
1267 | ros_NewF(3) = .TRUE. |
---|
1268 | ros_NewF(4) = .TRUE. |
---|
1269 | ros_NewF(5) = .TRUE. |
---|
1270 | ros_NewF(6) = .TRUE. |
---|
1271 | |
---|
1272 | !~~~> ros_ELO = estimator of local order - the minimum between the |
---|
1273 | ! main and the embedded scheme orders plus 1 |
---|
1274 | ros_ELO = 4.0d0 |
---|
1275 | |
---|
1276 | END SUBROUTINE Rodas4 |
---|
1277 | |
---|
1278 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1279 | END SUBROUTINE RosenbrockTLM |
---|
1280 | ! and all its internal procedures |
---|
1281 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1282 | |
---|
1283 | |
---|
1284 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1285 | SUBROUTINE FunTemplate( T, Y, Ydot ) |
---|
1286 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1287 | ! Template for the ODE function call. |
---|
1288 | ! Updates the rate coefficients (and possibly the fixed species) at each call |
---|
1289 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1290 | |
---|
1291 | IMPLICIT NONE |
---|
1292 | !~~~> Input variables |
---|
1293 | KPP_REAL :: T, Y(NVAR) |
---|
1294 | !~~~> Output variables |
---|
1295 | KPP_REAL :: Ydot(NVAR) |
---|
1296 | !~~~> Local variables |
---|
1297 | KPP_REAL :: Told |
---|
1298 | |
---|
1299 | Told = TIME |
---|
1300 | TIME = T |
---|
1301 | CALL Update_SUN() |
---|
1302 | CALL Update_RCONST() |
---|
1303 | CALL Fun( Y, FIX, RCONST, Ydot ) |
---|
1304 | TIME = Told |
---|
1305 | |
---|
1306 | END SUBROUTINE FunTemplate |
---|
1307 | |
---|
1308 | |
---|
1309 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1310 | SUBROUTINE JacTemplate( T, Y, Jcb ) |
---|
1311 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1312 | ! Template for the ODE Jacobian call. |
---|
1313 | ! Updates the rate coefficients (and possibly the fixed species) at each call |
---|
1314 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1315 | IMPLICIT NONE |
---|
1316 | |
---|
1317 | !~~~> Input variables |
---|
1318 | KPP_REAL :: T, Y(NVAR) |
---|
1319 | !~~~> Output variables |
---|
1320 | KPP_REAL :: Jcb(LU_NONZERO) |
---|
1321 | !~~~> Local variables |
---|
1322 | KPP_REAL :: Told |
---|
1323 | |
---|
1324 | Told = TIME |
---|
1325 | TIME = T |
---|
1326 | CALL Update_SUN() |
---|
1327 | CALL Update_RCONST() |
---|
1328 | CALL Jac_SP( Y, FIX, RCONST, Jcb ) |
---|
1329 | TIME = Told |
---|
1330 | |
---|
1331 | END SUBROUTINE JacTemplate |
---|
1332 | |
---|
1333 | |
---|
1334 | |
---|
1335 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1336 | SUBROUTINE HessTemplate( T, Y, Hes ) |
---|
1337 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1338 | ! Template for the ODE Hessian call. |
---|
1339 | ! Updates the rate coefficients (and possibly the fixed species) at each call |
---|
1340 | !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
---|
1341 | IMPLICIT NONE |
---|
1342 | |
---|
1343 | !~~~> Input variables |
---|
1344 | KPP_REAL :: T, Y(NVAR) |
---|
1345 | !~~~> Output variables |
---|
1346 | KPP_REAL :: Hes(NHESS) |
---|
1347 | !~~~> Local variables |
---|
1348 | KPP_REAL :: Told |
---|
1349 | |
---|
1350 | Told = TIME |
---|
1351 | TIME = T |
---|
1352 | CALL Update_SUN() |
---|
1353 | CALL Update_RCONST() |
---|
1354 | CALL Hessian( Y, FIX, RCONST, Hes ) |
---|
1355 | TIME = Told |
---|
1356 | |
---|
1357 | END SUBROUTINE HessTemplate |
---|
1358 | |
---|
1359 | END MODULE KPP_ROOT_Integrator |
---|
1360 | |
---|
1361 | |
---|
1362 | |
---|
1363 | |
---|