Invalid uncertainties for highly correlated parameters

Hi,

at the moment i am doing a measurement (or actual multiple measurements) where i have two highly anti-correlated parameters.

Specifically i have these parameters.
N_tot = 250.053 +/- 16.6388 (limited)
R_c = 2.68118 +/- 1.08419 (limited)
f_W0 = 0.310684 +/- 0.655279 (limited)
f_Z0 = 0.299184 +/- 0.64911 (limited)
with their uncertainties before hesse and minos.

for the parameters i have set hard bounds of:
N_tot: [0,2500]
R_c : [0,5]
f_W0: [0,1]
f_Z0: [0,1]
that come from physical considerations.

Here are is an example post fit correlation plot

Depending on the pseudo data that i fit this to either one or both parameters get invalid minimums.

Here is the part of the log where it tries to get the f_W0 errors.

Minuit2Minimizer::GetMinosError for parameter 2  f_W0 using max-calls 2000, tolerance 1
MnFunctionCross: parameter 0 set to -0.343554
MnSeedGenerator: for initial parameters FCN = -515.4510394003
MnSeedGenerator: Initial state:   - FCN =  -515.4510394003 Edm =      9.73552 NCalls =     11
MnSeedGenerator: Negative G2 found - new state:   - FCN =  -515.4510394003 Edm =      7.26966 NCalls =     11
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -515.4510394003 Edm =      7.26966 NCalls =     11
VariableMetric: Iteration #   0 - FCN =  -515.4510394003 Edm =      7.26966 NCalls =     11
VariableMetric: Iteration #   1 - FCN =  -521.4728404213 Edm =    0.0102232 NCalls =     19
VariableMetric: Iteration #   2 - FCN =  -521.4815929396 Edm =  7.04251e-06 NCalls =     26
MnFunctionCross: parameter 0 set to 5.55112e-17
MnSeedGenerator: for initial parameters FCN = -521.4815929396
MnSeedGenerator: Initial state:   - FCN =  -521.4815929396 Edm =  6.90865e-06 NCalls =      9
MnSeedGenerator: Negative G2 found - new state:   - FCN =  -521.4815929396 Edm =  6.49545e-06 NCalls =      9
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -521.4815929396 Edm =  6.49545e-06 NCalls =      9
VariableMetric: Iteration #   0 - FCN =  -521.4815929396 Edm =  6.49545e-06 NCalls =      9
VariableMetric: Iteration #   1 - FCN =  -521.4815994777 Edm =  7.77104e-10 NCalls =     16
MnFunctionCross: parameter 0 set to 0.654239
MnSeedGenerator: for initial parameters FCN = -494.9418758621
MnSeedGenerator: Initial state:   - FCN =  -494.9418758621 Edm =      1781.75 NCalls =     11
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -494.9418758621 Edm =      1781.75 NCalls =     11
VariableMetric: Iteration #   0 - FCN =  -494.9418758621 Edm =      1781.75 NCalls =     11
VariableMetric: Iteration #   1 - FCN =  -525.5371272243 Edm =      1.74369 NCalls =     29
VariableMetric: Iteration #   2 - FCN =  -526.0781211432 Edm =     0.118481 NCalls =     39
VariableMetric: Iteration #   3 - FCN =  -526.2314824381 Edm =     0.027502 NCalls =     49
VariableMetric: Iteration #   4 - FCN =  -526.2700076465 Edm =   0.00192152 NCalls =     58
VariableMetric: Iteration #   5 - FCN =   -526.272904085 Edm =   0.00068266 NCalls =     66
VariableMetric: Iteration #   6 - FCN =  -526.2738723361 Edm =  3.54075e-05 NCalls =     74
MnFunctionCross: parameter 0 set to 0.687609
MnSeedGenerator: for initial parameters FCN = -525.8668114997
MnSeedGenerator: Initial state:   - FCN =  -525.8668114997 Edm =   0.00148695 NCalls =      7
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -525.8668114997 Edm =   0.00148695 NCalls =      7
VariableMetric: Iteration #   0 - FCN =  -525.8668114997 Edm =   0.00148695 NCalls =      7
VariableMetric: Iteration #   1 - FCN =  -525.8680983704 Edm =  1.65737e-05 NCalls =     15
Info in <Minuit2>: MnMinos could not find Lower Value for Parameter  : par_name = f_W0
MnFunctionCross: parameter 0 set to 0.964923
MnSeedGenerator: for initial parameters FCN = -508.1906705745
MnSeedGenerator: Initial state:   - FCN =  -508.1906705745 Edm =     0.697507 NCalls =     11
MnSeedGenerator: Negative G2 found - new state:   - FCN =   -518.220374132 Edm =     0.116042 NCalls =     41
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =   -518.220374132 Edm =     0.116042 NCalls =     41
VariableMetric: Iteration #   0 - FCN =   -518.220374132 Edm =     0.116042 NCalls =     41
VariableMetric: Iteration #   1 - FCN =  -518.3325128181 Edm =  0.000380873 NCalls =     48
VariableMetric: Iteration #   2 - FCN =    -518.33285135 Edm =  1.69704e-09 NCalls =     56
MnFunctionCross: parameter 0 set to 0.637804
MnSeedGenerator: for initial parameters FCN = -526.4200226397
MnSeedGenerator: Initial state:   - FCN =  -526.4200226397 Edm =    0.0129311 NCalls =      9
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -526.4200226397 Edm =    0.0129311 NCalls =      9
VariableMetric: Iteration #   0 - FCN =  -526.4200226397 Edm =    0.0129311 NCalls =      9
VariableMetric: Iteration #   1 - FCN =  -526.4341208549 Edm =  0.000800544 NCalls =     17
VariableMetric: Iteration #   2 - FCN =  -526.4358426168 Edm =  9.61531e-08 NCalls =     25
MnFunctionCross: parameter 0 set to 0.634475
MnSeedGenerator: for initial parameters FCN = -526.465620497
MnSeedGenerator: Initial state:   - FCN =   -526.465620497 Edm =  2.96187e-05 NCalls =      9
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =   -526.465620497 Edm =  2.96187e-05 NCalls =      9
VariableMetric: Iteration #   0 - FCN =   -526.465620497 Edm =  2.96187e-05 NCalls =      9
VariableMetric: Iteration #   1 - FCN =  -526.4656508325 Edm =  4.70918e-08 NCalls =     16
MnFunctionCross: parameter 0 set to 0.628594
MnSeedGenerator: for initial parameters FCN = -526.5157496104
MnSeedGenerator: Initial state:   - FCN =  -526.5157496104 Edm =  0.000114849 NCalls =      9
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -526.5157496104 Edm =  0.000114849 NCalls =      9
VariableMetric: Iteration #   0 - FCN =  -526.5157496104 Edm =  0.000114849 NCalls =      9
VariableMetric: Iteration #   1 - FCN =  -526.5158661952 Edm =  3.16208e-08 NCalls =     16
Minos:  Invalid lower error for parameter f_W0
Minos: Lower error for parameter f_W0  :  -0.654239
Minos: Upper error for parameter f_W0  :  0.317623

in a slightly different setup i get something like this:

Minuit2Minimizer::GetMinosError for parameter 3  f_WZLT using max-calls 12000, tolerance 1
MnFunctionCross: parameter 0 set to -0.697167
MnSeedGenerator: for initial parameters FCN = -418.3355447846
MnSeedGenerator: Initial state:   - FCN =  -418.3355447846 Edm =      4720.43 NCalls =     93
MnSeedGenerator: Negative G2 found - new state:   - FCN =  -418.3355447846 Edm =      3.52679 NCalls =     93
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -418.3355447846 Edm =      3.52679 NCalls =     93
VariableMetric: Iteration #   0 - FCN =  -418.3355447846 Edm =      3.52679 NCalls =     93
Info in <Minuit2>: VariableMetricBuilder: no improvement in line search
VariableMetric: Iteration #   1 - FCN =  -418.3355447846 Edm =      3.52679 NCalls =     98
Info in <Minuit2>: VariableMetricBuilder: iterations finish without convergence.
Info in <Minuit2>: VariableMetricBuilder : edm = 14.1072
Info in <Minuit2>:             requested : edmval = 0.0005
Info in <Minuit2>: VariableMetricBuilder: INVALID function minimum - edm is above tolerance, : edm = 3.52679
Info in <Minuit2>: VariableMetricBuilder: Required tolerance  is 10 x edmval  : edmval = 0.0005
Info in <Minuit2>: MnMinos could not find Lower Value for Parameter  : par_name = f_WZLT
MnFunctionCross: parameter 0 set to 0.820094
MnSeedGenerator: for initial parameters FCN = -389.1211953927
MnSeedGenerator: Initial state:   - FCN =  -389.1211953927 Edm =            0 NCalls =     93
MnSeedGenerator: Negative G2 found - new state:   - FCN =  -389.1211953927 Edm =            0 NCalls =     93
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -389.1211953927 Edm =            0 NCalls =     93
VariableMetric: Iteration #   0 - FCN =  -389.1211953927 Edm =            0 NCalls =     93
MnFunctionCross: parameter 0 set to 0.440779
MnSeedGenerator: for initial parameters FCN = -446.5055263959
MnSeedGenerator: Initial state:   - FCN =  -446.5055263959 Edm =       113245 NCalls =     93
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -446.5055263959 Edm =       113245 NCalls =     93
VariableMetric: Iteration #   0 - FCN =  -446.5055263959 Edm =       113245 NCalls =     93
VariableMetric: Iteration #   1 - FCN =   -518.261482895 Edm =      56540.8 NCalls =    154
VariableMetric: Iteration #   2 - FCN =  -520.7580146987 Edm =      586.983 NCalls =    209
VariableMetric: Iteration #   3 - FCN =  -522.4948308502 Edm =      9.39696 NCalls =    259
VariableMetric: Iteration #   4 - FCN =  -522.5301500962 Edm =      4.18302 NCalls =    308
VariableMetric: Iteration #   5 - FCN =   -522.558431035 Edm =      2.90101 NCalls =    357
VariableMetric: Iteration #   6 - FCN =  -522.6439823471 Edm =      7.61813 NCalls =    405
VariableMetric: Iteration #   7 - FCN =  -522.7534376083 Edm =      1.03794 NCalls =    453
VariableMetric: Iteration #   8 - FCN =  -522.7924650526 Edm =      1.21882 NCalls =    501
VariableMetric: Iteration #   9 - FCN =   -522.800969863 Edm =     0.763302 NCalls =    550
VariableMetric: Iteration #  10 - FCN =  -522.8359927976 Edm =      2.48262 NCalls =    598
VariableMetric: Iteration #  11 - FCN =  -522.8763799712 Edm =      1.18746 NCalls =    646
VariableMetric: Iteration #  12 - FCN =   -522.889016906 Edm =     0.128825 NCalls =    694
VariableMetric: Iteration #  13 - FCN =  -522.8940088352 Edm =     0.771101 NCalls =    742
VariableMetric: Iteration #  14 - FCN =  -522.9676073514 Edm =      4.43429 NCalls =    790
VariableMetric: Iteration #  15 - FCN =  -523.0045719823 Edm =      0.19883 NCalls =    838
VariableMetric: Iteration #  16 - FCN =  -523.0058690827 Edm =    0.0487472 NCalls =    887
VariableMetric: Iteration #  17 - FCN =  -523.0121273303 Edm =      1.18096 NCalls =    935
VariableMetric: Iteration #  18 - FCN =  -523.1002040278 Edm =      1.13832 NCalls =    984
VariableMetric: Iteration #  19 - FCN =  -523.1073667355 Edm =     0.105426 NCalls =   1032
VariableMetric: Iteration #  20 - FCN =  -523.1088128246 Edm =     0.184908 NCalls =   1080
VariableMetric: Iteration #  21 - FCN =  -523.2842016659 Edm =      27.8444 NCalls =   1129
VariableMetric: Iteration #  22 - FCN =   -523.957649437 Edm =      9.72063 NCalls =   1182
VariableMetric: Iteration #  23 - FCN =    -524.01394942 Edm =      0.64482 NCalls =   1230
VariableMetric: Iteration #  24 - FCN =  -524.0281727752 Edm =      2.16844 NCalls =   1278
VariableMetric: Iteration #  25 - FCN =  -524.5464606214 Edm =      4.83489 NCalls =   1328
VariableMetric: Iteration #  26 - FCN =  -524.5764799808 Edm =     0.595418 NCalls =   1376
VariableMetric: Iteration #  27 - FCN =  -524.5958914246 Edm =      2.76933 NCalls =   1424
VariableMetric: Iteration #  28 - FCN =  -524.8589171519 Edm =      15.6508 NCalls =   1473
VariableMetric: Iteration #  29 - FCN =  -524.9479972405 Edm =     0.278672 NCalls =   1521
VariableMetric: Iteration #  30 - FCN =  -525.1204347027 Edm =     0.151263 NCalls =   1570
VariableMetric: Iteration #  31 - FCN =   -525.350910357 Edm =    0.0297723 NCalls =   1620
VariableMetric: Iteration #  32 - FCN =  -525.3824981768 Edm =   0.00127668 NCalls =   1668
VariableMetric: Iteration #  33 - FCN =  -525.3834009816 Edm =  1.62587e-05 NCalls =   1716
MnFunctionCross: parameter 0 set to 0.43762
MnSeedGenerator: for initial parameters FCN = -525.4155815417
MnSeedGenerator: Initial state:   - FCN =  -525.4155815417 Edm =  0.000406428 NCalls =     49
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -525.4155815417 Edm =  0.000406428 NCalls =     49
VariableMetric: Iteration #   0 - FCN =  -525.4155815417 Edm =  0.000406428 NCalls =     49
VariableMetric: Iteration #   1 - FCN =  -525.4160169755 Edm =  4.27228e-05 NCalls =     97
MnFunctionCross: parameter 0 set to 0.330849
MnSeedGenerator: for initial parameters FCN = -525.7624892986
MnSeedGenerator: Initial state:   - FCN =  -525.7624892986 Edm =     0.541618 NCalls =     49
MnSeedGenerator: Initial state:   - FCN =  -525.4155815417 Edm =  0.000406428 NCalls =     49
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -525.4155815417 Edm =  0.000406428 NCalls =     49
VariableMetric: Iteration #   0 - FCN =  -525.4155815417 Edm =  0.000406428 NCalls =     49
VariableMetric: Iteration #   1 - FCN =  -525.4160169755 Edm =  4.27228e-05 NCalls =     97
MnFunctionCross: parameter 0 set to 0.330849
MnSeedGenerator: for initial parameters FCN = -525.7624892986
MnSeedGenerator: Initial state:   - FCN =  -525.7624892986 Edm =     0.541618 NCalls =     49
MnSeedGenerator: Negative G2 found - new state:   - FCN =  -525.8080631154 Edm =     0.351401 NCalls =    104
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -525.8080631154 Edm =     0.351401 NCalls =    104
VariableMetric: Iteration #   0 - FCN =  -525.8080631154 Edm =     0.351401 NCalls =    104
VariableMetric: Iteration #   1 - FCN =  -526.1242855397 Edm =     0.237102 NCalls =    152
VariableMetric: Iteration #   2 - FCN =  -526.2887854693 Edm =    0.0260461 NCalls =    200
VariableMetric: Iteration #   3 - FCN =  -526.3120018483 Edm =   0.00260996 NCalls =    248
VariableMetric: Iteration #   4 - FCN =  -526.3152057263 Edm =  4.57748e-05 NCalls =    296
MnFunctionCross: parameter 0 set to 0.306733
MnSeedGenerator: for initial parameters FCN = -526.4403212432
MnSeedGenerator: Initial state:   - FCN =  -526.4403212432 Edm =    0.0241512 NCalls =     49
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -526.4403212432 Edm =    0.0241512 NCalls =     49
VariableMetric: Iteration #   0 - FCN =  -526.4403212432 Edm =    0.0241512 NCalls =     49
VariableMetric: Iteration #   1 - FCN =  -526.4651395987 Edm =    0.0001002 NCalls =     96
MnFunctionCross: parameter 0 set to 0.298181
MnSeedGenerator: for initial parameters FCN = -526.5100685931
MnSeedGenerator: Initial state:   - FCN =  -526.5100685931 Edm =   0.00391207 NCalls =     49
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -526.5100685931 Edm =   0.00391207 NCalls =     49
VariableMetric: Iteration #   0 - FCN =  -526.5100685931 Edm =   0.00391207 NCalls =     49
VariableMetric: Iteration #   1 - FCN =  -526.5139052471 Edm =  9.11503e-06 NCalls =     96
Minos:  Invalid lower error for parameter f_WZLT
Minos: Lower error for parameter f_WZLT  :  -0.75863
Minos: Upper error for parameter f_WZLT  :  0.235945

and this

Minuit2Minimizer::GetMinosError for parameter 4  f_WZTL using max-calls 12000, tolerance 1
MnFunctionCross: parameter 0 set to -0.702676
MnSeedGenerator: for initial parameters FCN = -446.510846723
MnSeedGenerator: Initial state:   - FCN =   -446.510846723 Edm =       2716.4 NCalls =     51
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =   -446.510846723 Edm =       2716.4 NCalls =     51
VariableMetric: Iteration #   0 - FCN =   -446.510846723 Edm =       2716.4 NCalls =     51
VariableMetric: Iteration #   1 - FCN =  -514.6782408225 Edm =      126.826 NCalls =    105
VariableMetric: Iteration #   2 - FCN =  -517.5910437663 Edm =      1.12998 NCalls =    153
VariableMetric: Iteration #   3 - FCN =  -524.4738630028 Edm =      1.47257 NCalls =    203
VariableMetric: Iteration #   4 - FCN =  -526.9779524397 Edm =    0.0299066 NCalls =    251
VariableMetric: Iteration #   5 - FCN =  -527.0103656139 Edm =   0.00037157 NCalls =    299
VariableMetric: Iteration #   6 - FCN =  -527.0113510945 Edm =  0.000348233 NCalls =    347
VariableMetric: Iteration #   7 - FCN =  -527.0123202832 Edm =  1.80309e-05 NCalls =    395
Info in <Minuit2>: MnMinos Parameter is at Lower limit. : par_name = f_WZTL
MnFunctionCross: parameter 0 set to 0.802602
MnSeedGenerator: for initial parameters FCN = 790.6008390092
MnSeedGenerator: Initial state:   - FCN =   790.6008390092 Edm =            0 NCalls =     93
MnSeedGenerator: Negative G2 found - new state:   - FCN =   790.6008390092 Edm =            0 NCalls =     93
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =   790.6008390092 Edm =            0 NCalls =     93
VariableMetric: Iteration #   0 - FCN =   790.6008390092 Edm =            0 NCalls =     93
MnFunctionCross: parameter 0 set to 0.426283
MnSeedGenerator: for initial parameters FCN = -365.7478834353
MnSeedGenerator: Initial state:   - FCN =  -365.7478834353 Edm =  2.34468e+07 NCalls =     89
VariableMetric: start iterating until Edm is < 0.0005
VariableMetric: Initial state   - FCN =  -365.7478834353 Edm =  2.34468e+07 NCalls =     89
VariableMetric: Iteration #   0 - FCN =  -365.7478834353 Edm =  2.34468e+07 NCalls =     89
Info in <Minuit2>: VariableMetricBuilder: no improvement in line search
VariableMetric: Iteration #   1 - FCN =  -365.7478834353 Edm =  2.34468e+07 NCalls =    100
Info in <Minuit2>: VariableMetricBuilder: iterations finish without convergence.
Info in <Minuit2>: VariableMetricBuilder : edm = 2.34468e+07
Info in <Minuit2>:             requested : edmval = 0.0005
Info in <Minuit2>: VariableMetricBuilder: INVALID function minimum - edm is above tolerance, : edm = 2.34468e+07
Info in <Minuit2>: VariableMetricBuilder: Required tolerance  is 10 x edmval  : edmval = 0.0005
Info in <Minuit2>: MnMinos could not find Upper Value for Parameter  : par_name = f_WZTL
Minos:  Parameter : f_WZTL  is at Lower limit.
Minos: Lower error for parameter f_WZTL  :  -0.0499633
Minos:  Invalid upper error for parameter f_WZTL
Minos: Upper error for parameter f_WZTL  :  0.752639

for these parameters:
N_tot = 250.053 +/- 17.6416 (limited)
f_WZLL = 0.249221 +/- 0.0978734 (limited)
f_WZLT = 0.0614636 +/- 0.936316 (limited)
f_WZTL = 0.0499633 +/- 0.956574 (limited)

Is there anything i can do to help the fit get the proper uncertainties here, where for these parameters they are dominated by the (physical) boundary conditions?

Hi,
I don’t think the problem is caused by some numerical issues, because the line search failed when running the minimization for finding the minimum error. Are you using the option Offset(true) in RooFit that improves the likelihood numerical error ?

Lorenzo

The line search fail only happens in the secod scenario, not for te first.

And i havent checked this option for the second case yet, but for the first one it did not help with the uncertainties. but it did break our scripts for determining the significance, so i do not think this is really an options. but i will still have a look at enabling it for the second scenario.

Edit:
i checked for the second case and it actually does help. so it might make sense to do 2 fits there.

once with offset to get the proper 1sigma uncertainties for every parameter and once without to get the significance of te parameter of interest.

still having trouble with the first ones.

Hi,
Correct, the failure in the first scenario are due to some negative G2, I suspect this is due to the highly correlated parameter, but I would need to have the code to reproduce it and investigate better to understand the issue. As a possible test you can try to fix one of the highly correlated parameter and run Minos again to see if it works in that case

Cheers

Lorenzo

If i fix one of the parameters then the uncertainties for the other one get calculated just fine.
but i cant really do that.

the behaviour i would want is probably that the lower uncertainty of each parameter is so that value-uncertainty hits the lower bound (0) and the upper uncertainties should be slightly larger than that because for the most part they are driven by the anticorrelation but the statistical uncertainty adds another small bit.

is there anway settings on the fit side that i can use to help guide it there? because i dont think i have any room elsewhere.

Hi,
Which ROOT version are you using ? You can try eventually to use “Minuit” as minimiser in case it makes a difference.
Another possibility if you can’t fit a parameter is to apply a transformation to them to decrease the correlation.

Lorenzo

Another thing you can do is trying performing a scan of the profile likelihood.
For example for your problematic parameter, f_W0, you fix it and scan the value between 0,1 and you compute for some points (e.g. 20 or more if running the fit is fast), the minimum of the likelihood where the parameter f_W0 is fixed to those points. If you plot the resulting function you can have a better idea what is happening and compute the Minos error by interpolation.

Lorenzo

current root version is 6.08.06 and we have those lines in the code:

  RooMinimizer m(*nll);
  m.setMinimizerType("Minuit2");
  m.minimize("Minuit2","minimize");
  m.hesse();
  m.minos();

And i actually do have some likelihood scans for these parameters:


And if i combine them into one paramter that is the average of the two then it looks fine.

but that doesnt really work with the physics/interpretation of the measurement we are going for.

This makes sense. Are these scan performed already with a fine grid ? If not you can try increase the number of points close to the sigma lines crossing to have a better estimation of the errors.

its 20 points between the manually set min and max.

but i dont really want get these manually whenever i change anything.

Since it does these fine if i do it like that shouldnt it be possible for them to be determined automatically? i guess the problem are the starting points from where it tries to get these. is there anyway to change those?

Hi,
Yes I agree it should work if done automatically. I would need to have your workspace and your code doing the fitting to understand exactly what is happening.

Lorenzo

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.