[14] | 1 | /**CFile*********************************************************************** |
---|
| 2 | |
---|
| 3 | FileName [markFPSolve.c] |
---|
| 4 | |
---|
| 5 | PackageName [mark] |
---|
| 6 | |
---|
| 7 | Synopsis [This file contains functions that implement the fixed point |
---|
| 8 | method. For more details please refer |
---|
| 9 | |
---|
| 10 | G. D. Hachtel, E. Macii, A. Pardo and F. Somenzi, "Markovian Analysis |
---|
| 11 | of Large Finite State Machines", IEEE Trans. on CAD, December 1996. ] |
---|
| 12 | |
---|
| 13 | Description [This file contains functions that implement the fixed point |
---|
| 14 | method. For more details please refer |
---|
| 15 | |
---|
| 16 | G. D. Hachtel, E. Macii, A. Pardo and F. Somenzi, "Markovian Analysis |
---|
| 17 | of Large Finite State Machines", IEEE Trans. on CAD, December 1996. ] |
---|
| 18 | |
---|
| 19 | Author [Balakrishna Kumthekar] |
---|
| 20 | |
---|
| 21 | Copyright [This file was created at the University of Colorado at |
---|
| 22 | Boulder. The University of Colorado at Boulder makes no warranty |
---|
| 23 | about the suitability of this software for any purpose. It is |
---|
| 24 | presented on an AS IS basis.] |
---|
| 25 | |
---|
| 26 | ******************************************************************************/ |
---|
| 27 | |
---|
| 28 | #include "markInt.h" |
---|
| 29 | |
---|
| 30 | /*---------------------------------------------------------------------------*/ |
---|
| 31 | /* Constant declarations */ |
---|
| 32 | /*---------------------------------------------------------------------------*/ |
---|
| 33 | |
---|
| 34 | |
---|
| 35 | /*---------------------------------------------------------------------------*/ |
---|
| 36 | /* Stucture declarations */ |
---|
| 37 | /*---------------------------------------------------------------------------*/ |
---|
| 38 | |
---|
| 39 | |
---|
| 40 | /*---------------------------------------------------------------------------*/ |
---|
| 41 | /* Type declarations */ |
---|
| 42 | /*---------------------------------------------------------------------------*/ |
---|
| 43 | |
---|
| 44 | |
---|
| 45 | /*---------------------------------------------------------------------------*/ |
---|
| 46 | /* Variable declarations */ |
---|
| 47 | /*---------------------------------------------------------------------------*/ |
---|
| 48 | |
---|
| 49 | |
---|
| 50 | /*---------------------------------------------------------------------------*/ |
---|
| 51 | /* Macro declarations */ |
---|
| 52 | /*---------------------------------------------------------------------------*/ |
---|
| 53 | |
---|
| 54 | |
---|
| 55 | /**AutomaticStart*************************************************************/ |
---|
| 56 | |
---|
| 57 | /*---------------------------------------------------------------------------*/ |
---|
| 58 | /* Static function prototypes */ |
---|
| 59 | /*---------------------------------------------------------------------------*/ |
---|
| 60 | |
---|
| 61 | |
---|
| 62 | /**AutomaticEnd***************************************************************/ |
---|
| 63 | |
---|
| 64 | |
---|
| 65 | /*---------------------------------------------------------------------------*/ |
---|
| 66 | /* Definition of exported functions */ |
---|
| 67 | /*---------------------------------------------------------------------------*/ |
---|
| 68 | |
---|
| 69 | |
---|
| 70 | /*---------------------------------------------------------------------------*/ |
---|
| 71 | /* Definition of internal functions */ |
---|
| 72 | /*---------------------------------------------------------------------------*/ |
---|
| 73 | |
---|
| 74 | /**Function******************************************************************** |
---|
| 75 | |
---|
| 76 | Synopsis [Computes steady state probabilities vis fixed point method.] |
---|
| 77 | |
---|
| 78 | Description [Computes steady state probabilities via fixed point method. |
---|
| 79 | The function returns an array of two ADDs. The ADD with index 0, represents |
---|
| 80 | the steady state probabilities and the second one the one-step transition |
---|
| 81 | probability matrix.] |
---|
| 82 | |
---|
| 83 | SideEffects [None] |
---|
| 84 | |
---|
| 85 | SeeAlso [] |
---|
| 86 | |
---|
| 87 | ******************************************************************************/ |
---|
| 88 | bdd_node ** |
---|
| 89 | MarkAddFPSolve( |
---|
| 90 | CK *ck) |
---|
| 91 | { |
---|
| 92 | bdd_manager *manager; |
---|
| 93 | bdd_node **xAddVars, **yAddVars; |
---|
| 94 | bdd_node *Scaling; |
---|
| 95 | bdd_node *InitG, *G, *NewG; |
---|
| 96 | bdd_node *p, *q, *newTr; |
---|
| 97 | bdd_node *xCube, *ddTemp, *guess; |
---|
| 98 | bdd_node *zero; |
---|
| 99 | bdd_node **result; |
---|
| 100 | bdd_node *probMatrix; |
---|
| 101 | int nVars, iter = 0; |
---|
| 102 | int converged; |
---|
| 103 | double max; |
---|
| 104 | |
---|
| 105 | manager = ck->manager; |
---|
| 106 | nVars = ck->nVars; |
---|
| 107 | xAddVars = ck->xAddVars; |
---|
| 108 | yAddVars = ck->yAddVars; |
---|
| 109 | |
---|
| 110 | result = ALLOC(bdd_node *, 2); |
---|
| 111 | if (result == NULL) |
---|
| 112 | return NULL; |
---|
| 113 | zero = bdd_read_zero(manager); |
---|
| 114 | |
---|
| 115 | /* Build an ADD for one step transition probability matrix */ |
---|
| 116 | bdd_ref(probMatrix = MarkAddBuildCoeff(manager,ck->coeff, ck->piAddVars, |
---|
| 117 | ck->inputProb, ck->scale, nVars, |
---|
| 118 | ck->nPi)); |
---|
| 119 | result[1] = probMatrix; |
---|
| 120 | |
---|
| 121 | /* create initial guess and print it; |
---|
| 122 | * equiprobability to all the states and probability 1 to one of the |
---|
| 123 | * reset states are currently supported |
---|
| 124 | */ |
---|
| 125 | switch(ck->start) { |
---|
| 126 | case Start_EquiProb_c: |
---|
| 127 | (void)printf("Initial guess: equiprob\n"); |
---|
| 128 | bdd_ref(ck->init_guess = ck->term_scc); |
---|
| 129 | p = bdd_add_const(manager, |
---|
| 130 | (double) (1/(double)(ck->term_SCC_states))); |
---|
| 131 | bdd_ref(p); |
---|
| 132 | bdd_ref(q = bdd_add_ite(manager,ck->init_guess,p,zero)); |
---|
| 133 | bdd_recursive_deref(manager,p); |
---|
| 134 | InitG = bdd_add_swap_variables(manager,q,xAddVars,yAddVars,nVars); |
---|
| 135 | bdd_ref(InitG); |
---|
| 136 | bdd_recursive_deref(manager,q); |
---|
| 137 | bdd_recursive_deref(manager,ck->init_guess); |
---|
| 138 | break; |
---|
| 139 | case Start_Reset_c: |
---|
| 140 | default: |
---|
| 141 | /* Pick one of the states in the TSCC as the initial guess */ |
---|
| 142 | bdd_ref(ddTemp = bdd_add_bdd_threshold(manager,ck->term_scc,1)); |
---|
| 143 | bdd_ref(guess = bdd_bdd_pick_one_minterm(manager, ddTemp, |
---|
| 144 | ck->xVars, nVars)); |
---|
| 145 | bdd_recursive_deref(manager,ddTemp); |
---|
| 146 | bdd_ref(ck->init_guess = bdd_bdd_to_add(manager,guess)); |
---|
| 147 | bdd_recursive_deref(manager,guess); |
---|
| 148 | bdd_ref(InitG = bdd_add_swap_variables(manager,ck->init_guess, |
---|
| 149 | xAddVars, yAddVars,nVars)); |
---|
| 150 | bdd_recursive_deref(manager,ck->init_guess); |
---|
| 151 | break; |
---|
| 152 | } |
---|
| 153 | |
---|
| 154 | /* put prob. transition matrix in appropriate form (transpose)*/ |
---|
| 155 | newTr = bdd_add_swap_variables(manager,probMatrix,xAddVars,yAddVars,nVars); |
---|
| 156 | bdd_ref(newTr); |
---|
| 157 | |
---|
| 158 | /* calculate the x-cube for abstraction */ |
---|
| 159 | bdd_ref(xCube = bdd_add_compute_cube(manager,xAddVars,NULL,nVars)); |
---|
| 160 | |
---|
| 161 | do { |
---|
| 162 | iter++; |
---|
| 163 | G = bdd_add_matrix_multiply(manager,newTr,InitG,yAddVars,nVars); |
---|
| 164 | bdd_ref(G); |
---|
| 165 | bdd_ref(Scaling = bdd_add_exist_abstract(manager,G,xCube)); |
---|
| 166 | |
---|
| 167 | bdd_ref(NewG = bdd_add_apply(manager,bdd_add_divide,G,Scaling)); |
---|
| 168 | bdd_recursive_deref(manager,G); |
---|
| 169 | G = NewG; |
---|
| 170 | bdd_recursive_deref(manager,Scaling); |
---|
| 171 | q = bdd_add_swap_variables(manager,G,xAddVars,yAddVars,nVars); |
---|
| 172 | bdd_ref(q); |
---|
| 173 | max = bdd_add_value(bdd_add_find_max(manager,InitG)); |
---|
| 174 | converged = bdd_equal_sup_norm(manager,q,InitG,ck->reltol*max,0); |
---|
| 175 | bdd_recursive_deref( manager,InitG); |
---|
| 176 | if (converged) { |
---|
| 177 | (void) fprintf(vis_stdout,"Iteration = %d\n",iter); |
---|
| 178 | bdd_recursive_deref( manager,newTr); |
---|
| 179 | bdd_recursive_deref( manager,q); |
---|
| 180 | bdd_recursive_deref(manager,xCube); |
---|
| 181 | result[0] = G; |
---|
| 182 | return result; |
---|
| 183 | } |
---|
| 184 | bdd_recursive_deref( manager,G); |
---|
| 185 | InitG = q; |
---|
| 186 | } while (1); |
---|
| 187 | |
---|
| 188 | } /* end of addFPSolve */ |
---|
| 189 | |
---|
| 190 | |
---|
| 191 | /**Function******************************************************************** |
---|
| 192 | |
---|
| 193 | Synopsis [Builds the one-step transition probability matrix given primary |
---|
| 194 | input probabilities and 0-1 ADD transition relation.] |
---|
| 195 | |
---|
| 196 | Description [Builds the one-step transition probability matrix given primary |
---|
| 197 | input probabilities and 0-1 ADD transition relation.] |
---|
| 198 | |
---|
| 199 | SideEffects [None] |
---|
| 200 | |
---|
| 201 | SeeAlso [] |
---|
| 202 | |
---|
| 203 | ******************************************************************************/ |
---|
| 204 | bdd_node * |
---|
| 205 | MarkAddBuildCoeff( |
---|
| 206 | bdd_manager *manager, |
---|
| 207 | bdd_node *func, |
---|
| 208 | bdd_node **piAddVars, |
---|
| 209 | st_table *inputProb, |
---|
| 210 | double scale, |
---|
| 211 | int nVars, |
---|
| 212 | int nPi) |
---|
| 213 | { |
---|
| 214 | /* Given func, the coefficient matrix is built */ |
---|
| 215 | /* this function is used first to build the collapsed coeff matrix */ |
---|
| 216 | /* and then the matrix for every TSCC */ |
---|
| 217 | |
---|
| 218 | bdd_node *Correction; |
---|
| 219 | bdd_node *q; |
---|
| 220 | bdd_node *matrix; |
---|
| 221 | bdd_node *piAddCube; |
---|
| 222 | |
---|
| 223 | bdd_ref(piAddCube = bdd_add_compute_cube(manager,piAddVars,NULL,nPi)); |
---|
| 224 | |
---|
| 225 | /* Create the transition matrix either with equiprobable |
---|
| 226 | or specific input probs. */ |
---|
| 227 | |
---|
| 228 | if (inputProb != NULL) { |
---|
| 229 | bdd_ref(matrix = Mark_addInProb(manager,func,piAddCube,inputProb)); |
---|
| 230 | } |
---|
| 231 | else { |
---|
| 232 | /* create correction ADD and print it */ |
---|
| 233 | bdd_ref(Correction = bdd_add_const(manager, |
---|
| 234 | (scale/(double)(1 << nPi)))); |
---|
| 235 | |
---|
| 236 | bdd_ref(q = bdd_add_exist_abstract(manager,func,piAddCube)); |
---|
| 237 | |
---|
| 238 | /* apply correction to the transition relation matrix and print it */ |
---|
| 239 | bdd_ref(matrix = bdd_add_apply(manager, bdd_add_times,q, Correction)); |
---|
| 240 | bdd_recursive_deref( manager,Correction); |
---|
| 241 | bdd_recursive_deref(manager,q); |
---|
| 242 | } |
---|
| 243 | bdd_recursive_deref(manager,piAddCube); |
---|
| 244 | bdd_deref(matrix); |
---|
| 245 | return(matrix); |
---|
| 246 | } |
---|
| 247 | |
---|
| 248 | |
---|