Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.
Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Help understanding detailed solver log and time-step size selection
Posted 2013年3月15日 GMT-4 23:17 Version 4.3a 2 Replies
Please login with a confirmed email address before reporting spam
Hello Everyone,
Could someone please help me understand what is happening in this excerpt from one of my solver logs? I'm trying to understand what exactly the non-linear solver is 'thinking' so that I can understand why its failing to converge.
I'm using a fully coupled solver with automatic time-stepping. From my reading I've gathered that the automatic time-stepping attempts to estimate a timestep such that the error will be less than the prescribed tolerance. When it fails it should be taking shorter time-steps. However, in my log it appears to me that the solver is barely decreasing it's step size (0.019 to 0.0069) despite failing to converge numerous times. Why is this? I also notice that the damping does not always decrease despite not finding a solution.
I am missing a good deal of information about the detailed solver log. Can anyone direct me to a good source? Why is the stepsize not displayed in seconds? Is the stepsize the time-step before or after applying the damping? In automatic time-steping mode the Jacobian is supposed to be updated on each iteration. Why is the number of Jacobian evaluations less than half the number of solution evaluations?
Step Time Stepsize Res Jac Sol Order Tfail NLfail
81 6.6048e+006 65536 out 238 118 267 2 0 5
Time t = 6.6048e+006.
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.018 0.0093350 0.019 241 119 270
2 0.019 0.0164827 0.019 242 120 272
3 0.018 0.2523938 0.019 243 121 274
4 0.013 0.6283830 0.017 244 122 276
5 0.32 0.0013936 0.32 245 123 278
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0072 0.2885796 0.0097 248 124 281
2 0.01 0.2970095 0.012 249 125 283
3 0.015 0.0299090 0.015 251 126 286
4 0.15 0.0003656 0.15 252 127 288
5 0.69 0.0000205 0.69 253 128 290
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0077 0.1348999 0.0089 256 129 293
2 0.013 0.0343265 0.013 257 130 295
3 0.1 0.0005208 0.1 258 131 297
4 0.28 0.0000161 0.28 261 132 301
5 0.34 0.0000000 0.34 274 133 315
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.029 0.0020000 0.029 277 134 318
2 0.021 0.0060284 0.021 278 135 320
3 0.0026 0.0018224 0.0026 280 136 323
4 0.0036 0.0014934 0.0036 281 137 325
5 0.0047 0.0029224 0.0048 282 138 327
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0074 0.0020000 0.0074 285 139 330
2 0.0088 0.0018893 0.0088 286 140 332
3 0.0087 0.0161047 0.0089 287 141 334
4 0.0035 0.0015245 0.0035 289 142 337
5 0.0063 0.0006903 0.0064 290 143 339
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0066 0.0024182 0.0066 293 144 342
2 0.0084 0.0029417 0.0084 294 145 344
3 0.008 0.0244975 0.0083 295 146 346
4 0.037 0.0010122 0.037 296 147 348
5 0.34 0.0000000 0.34 305 148 358
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0066 0.0030088 0.0066 308 149 361
2 0.0086 0.0035414 0.0086 309 150 363
3 0.0076 0.0318030 0.008 310 151 365
4 0.079 0.0003457 0.079 311 152 367
5 0.2 0.0000335 0.2 313 153 370
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0067 0.0031310 0.0067 316 154 373
2 0.0088 0.0036122 0.0088 317 155 375
3 0.0077 0.0320992 0.0081 318 156 377
4 0.076 0.0003791 0.076 319 157 379
5 0.17 0.0000648 0.17 321 158 382
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0068 0.0031033 0.0068 324 159 385
2 0.0091 0.0034830 0.0091 325 160 387
3 0.008 0.0313162 0.0084 326 161 389
4 0.11 0.0001941 0.11 327 162 391
5 0.24 0.0000691 0.24 328 163 393
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0069 0.0030643 0.0069 331 164 396
2 0.0093 0.0033734 0.0093 332 165 398
3 0.0082 0.0306668 0.0086 333 166 400
4 0.16 0.0000882 0.16 334 167 402
5 0.37 0.0000291 0.37 335 168 404
Step Time Stepsize Res Jac Sol Order Tfail NLfail
81 6.6048e+006 65536 out 335 168 404 2 0 15
Time-Dependent Solver 1 in Solver 1: Solution time: 756 s. (12 minutes, 36 seconds)
Thanks for your help guys,
Andrew
Could someone please help me understand what is happening in this excerpt from one of my solver logs? I'm trying to understand what exactly the non-linear solver is 'thinking' so that I can understand why its failing to converge.
I'm using a fully coupled solver with automatic time-stepping. From my reading I've gathered that the automatic time-stepping attempts to estimate a timestep such that the error will be less than the prescribed tolerance. When it fails it should be taking shorter time-steps. However, in my log it appears to me that the solver is barely decreasing it's step size (0.019 to 0.0069) despite failing to converge numerous times. Why is this? I also notice that the damping does not always decrease despite not finding a solution.
I am missing a good deal of information about the detailed solver log. Can anyone direct me to a good source? Why is the stepsize not displayed in seconds? Is the stepsize the time-step before or after applying the damping? In automatic time-steping mode the Jacobian is supposed to be updated on each iteration. Why is the number of Jacobian evaluations less than half the number of solution evaluations?
Step Time Stepsize Res Jac Sol Order Tfail NLfail
81 6.6048e+006 65536 out 238 118 267 2 0 5
Time t = 6.6048e+006.
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.018 0.0093350 0.019 241 119 270
2 0.019 0.0164827 0.019 242 120 272
3 0.018 0.2523938 0.019 243 121 274
4 0.013 0.6283830 0.017 244 122 276
5 0.32 0.0013936 0.32 245 123 278
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0072 0.2885796 0.0097 248 124 281
2 0.01 0.2970095 0.012 249 125 283
3 0.015 0.0299090 0.015 251 126 286
4 0.15 0.0003656 0.15 252 127 288
5 0.69 0.0000205 0.69 253 128 290
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0077 0.1348999 0.0089 256 129 293
2 0.013 0.0343265 0.013 257 130 295
3 0.1 0.0005208 0.1 258 131 297
4 0.28 0.0000161 0.28 261 132 301
5 0.34 0.0000000 0.34 274 133 315
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.029 0.0020000 0.029 277 134 318
2 0.021 0.0060284 0.021 278 135 320
3 0.0026 0.0018224 0.0026 280 136 323
4 0.0036 0.0014934 0.0036 281 137 325
5 0.0047 0.0029224 0.0048 282 138 327
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0074 0.0020000 0.0074 285 139 330
2 0.0088 0.0018893 0.0088 286 140 332
3 0.0087 0.0161047 0.0089 287 141 334
4 0.0035 0.0015245 0.0035 289 142 337
5 0.0063 0.0006903 0.0064 290 143 339
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0066 0.0024182 0.0066 293 144 342
2 0.0084 0.0029417 0.0084 294 145 344
3 0.008 0.0244975 0.0083 295 146 346
4 0.037 0.0010122 0.037 296 147 348
5 0.34 0.0000000 0.34 305 148 358
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0066 0.0030088 0.0066 308 149 361
2 0.0086 0.0035414 0.0086 309 150 363
3 0.0076 0.0318030 0.008 310 151 365
4 0.079 0.0003457 0.079 311 152 367
5 0.2 0.0000335 0.2 313 153 370
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0067 0.0031310 0.0067 316 154 373
2 0.0088 0.0036122 0.0088 317 155 375
3 0.0077 0.0320992 0.0081 318 156 377
4 0.076 0.0003791 0.076 319 157 379
5 0.17 0.0000648 0.17 321 158 382
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0068 0.0031033 0.0068 324 159 385
2 0.0091 0.0034830 0.0091 325 160 387
3 0.008 0.0313162 0.0084 326 161 389
4 0.11 0.0001941 0.11 327 162 391
5 0.24 0.0000691 0.24 328 163 393
Iter ErrEst Damping Stepsize #Res #Jac #Sol
1 0.0069 0.0030643 0.0069 331 164 396
2 0.0093 0.0033734 0.0093 332 165 398
3 0.0082 0.0306668 0.0086 333 166 400
4 0.16 0.0000882 0.16 334 167 402
5 0.37 0.0000291 0.37 335 168 404
Step Time Stepsize Res Jac Sol Order Tfail NLfail
81 6.6048e+006 65536 out 335 168 404 2 0 15
Time-Dependent Solver 1 in Solver 1: Solution time: 756 s. (12 minutes, 36 seconds)
Thanks for your help guys,
Andrew
2 Replies Last Post 2013年7月5日 GMT-4 14:23
Hello Andrew Prudil
Your Discussion has gone 30 days without a reply. If you still need help with COMSOL and have an on-subscription license, please visit our Support Center for help.
If you do not hold an on-subscription license, you may find an answer in another Discussion or in the Knowledge Base.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hello,
Did you figure out it? Because I'm in the same trouble.
I'm quite a beginner at using COMSOL.
In particular, I would like to ask how you could obtain the value of Non-Linear Solver Log.
I'm using the Time-Dipendent Solver and in the log, I suppose, I can see only info about
"Step - Time - Stepsize - Res - Jac - Sol - Order - Tfail - NLfail".
So, can you help me to figure out where I can find the infotmation about the Non-Linear Solver:
"Iter - ErrEst - Damping - Stepsize - #Res - #Jac - #Sol"
Thanks a lot and sorry to disturb you.
Kind Regards,
Rocco
Did you figure out it? Because I'm in the same trouble.
I'm quite a beginner at using COMSOL.
In particular, I would like to ask how you could obtain the value of Non-Linear Solver Log.
I'm using the Time-Dipendent Solver and in the log, I suppose, I can see only info about
"Step - Time - Stepsize - Res - Jac - Sol - Order - Tfail - NLfail".
So, can you help me to figure out where I can find the infotmation about the Non-Linear Solver:
"Iter - ErrEst - Damping - Stepsize - #Res - #Jac - #Sol"
Thanks a lot and sorry to disturb you.
Kind Regards,
Rocco
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi,
I can sort of help. To see what I was seeing you need to switch the solver log to detailed mode. This is an option under the "advanced settings" of the time-dependent solver tree.
I never did figure it out what some of the solver log stuff means but I know where you can look if you are curious. The comsol manual states that when using the BDF time-stepping you are actually using something called IDA from Lawarence Livermore National Labs. It has incorperated into comsol as one of the time-dependent solver options.
You can read start reading about IDA at computation.llnl.gov/casc/sundials/documentation/ida_guide.pdf
When I investigated it further I sort of decided it didn't matter because I don't have any control over most of it through COMSOL.
I can sort of help. To see what I was seeing you need to switch the solver log to detailed mode. This is an option under the "advanced settings" of the time-dependent solver tree.
I never did figure it out what some of the solver log stuff means but I know where you can look if you are curious. The comsol manual states that when using the BDF time-stepping you are actually using something called IDA from Lawarence Livermore National Labs. It has incorperated into comsol as one of the time-dependent solver options.
You can read start reading about IDA at computation.llnl.gov/casc/sundials/documentation/ida_guide.pdf
When I investigated it further I sort of decided it didn't matter because I don't have any control over most of it through COMSOL.