Consider reorganising the model

Hi,

I have successfully run a random regression model fitting legendre polynomials of order 1 for both random effects of sire and cow(permanent environmental effect). However, when I use the same model layout for legendre polynomial of order 2 for both random effects I have the message to consider reorganising the model.

I have tried fitting sire as order 2 and the permanent environmental effect as order 1 but I get the same comment of considering reorganising the model.

The starting values are taken from the previous analysis with random effects fitted with legendre polynomials of order 1, and I have tried different analyses changing parameters to be fixed, unbounded, and positive, but I keep getting the same error message.

I wondered if anyone could suggest what I can do.

Thanks, Tracey

 Sire 1654 !P
 Cow 43952 !I
 Lactation 3
 HTD 193060 !I
 SeasonSubClass 151 !I
 DIM !-154.5 !/150
 CalvingYear 13 !I
 lnSCC !=SCC1000 !^0 !M0
SCC_Pedigree_07102009.ped
SCC_SireData_AllLact_07102009.dat !CSV !MAXIT 300 !STEP 0.001 !MVREMOVE
lnSCC ~ mu leg(DIM,4).Lactation SeasonSubClass CalvingYear,
!r leg(DIM,2).Lactation.Sire leg(DIM,2).Lactation.Cow,
!f HTD #mv
12 1 2
26688 0 0 !s2=0.999152 #lactation 1 dim 4 to 24
34017 0 0 !s2=0.633912 #lactation 1 dim 25 to 49
279948 0 0 !s2=0.398534 #lactation 1 dim 50 to 249
71752 0 0 !s2=0.320376 #lactation 1 dim 250 to 305
27272 0 0 !s2=1.04453 #lactation 2 dim 4 to 24
34546 0 0 !s2=0.771882 #lactation 2 dim 25 to 49
280210 0 0 !s2=0.443768 #lactation 2 dim 50 to 249
71632 0 0 !s2=0.289394 #lactation 2 dim 250 to 305
26964 0 0 !s2=1.19397 #lactation 3 dim 4 to 24
33975 0 0 !s2=0.864685 #1actation 3 dim 25 to 49
270069 0 0 !s2=0.502304 #lactation 3 dim 50 to 249
63271 0 0 !s2=0.313208 #lactation 3 dim 250 to 305
leg(DIM,2).Lactation.Sire
9 0 US !GP #!GU #!+45
0.0297869
0.0291059 0.0466367
0.0 0.0 0.01
0.0289319 0.0481556 0.0 0.0562011
0.00159064 0.00496842 0.01 0.00297813 0.00460538
0.0 0.0 0.0 0.0 0.0 0.01
-0.00172770 0.000306669 0.0 -0.00210902 0.00292977 0.0 0.00806909
-0.000717143 0.00131430 0.0 -0.000618479 0.00144014 0.0 0.00827325 0.0102263
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.01
Sire 0 AINV
leg(DIM,2).Lactation.Cow
9 0 US !GP #!GU !+45
0.751512
0.258720 0.866191
0.0 0.0 0.01
0.234040 0.446013 0.0 1.09465
0.00968714 0.0395773 0.0 0.0275260 0.117216
0.0 0.0 0.0 0.0 0.0 0.01
-0.0130884 -0.0642843 0.0 0.0152998 0.01644203 0.0 0.180386
-0.0219540 -0.0223845 0.0 -0.107697 0.0118643 0.0 0.0489351 0.224197
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.01
Cow 0 0

parts of the asr file

  Sire 1654 !P
  Cow 43952 !I
  HTD 193060 !I
  SeasonSubClass 151 !I
  DIM !-154.5 !/150
  CalvingYear 13 !I
  SCC1000 !M0
  lnSCC !=SCC1000 !^0 !M0
 Reading pedigree file SCC_Pedigree_07102009.ped : skipping 0 lines
 Using an adapted version of Meuwissen & Luo GSE 1992 305-313: Specify !METHOD 1 for column method.
 PEDIGREE [SCC_Pedigree_07102009.ped ] has 5789 identities, 18805 Non zero elements
 QUALIFIERS: !CSV !MAXIT 300 !STEP 0.001 !MVREMOVE
 Reading SCC_SireData_AllLact_07102009.dat FREE FORMAT skipping 0 lines

 Univariate analysis of lnSCC
 Using 1220344 records of 1229522 read

 Notice: Model term leg(DIM,4) is not the same term as leg(DIM,2) but may be
          equivalent. Check the knot points reported in the .res file

  Model term Size #miss #zero MinNon0 Mean MaxNon0
   1 Sire !P 5789 0 0 2770. 4967. 5789.
   2 Cow 43952 0 0 1 21870.3217 43952
   3 Lactation 3 0 0 1 1.9851 3
   4 HTD 193060 0 0 1 78133.0359 193060
   5 SeasonSubClass 151 0 0 1 45.6440 151
   6 DIM 0 0 -1.003 -0.1367E-01 1.003
   7 CalvingYear 13 0 0 1 4.8086 13
   8 Het1 0648943 0.3000E-01 6.951 100.0
   9 Recomb1 0509660 0.3000E-01 6.697 50.02
  10 SCC 0 0 1.000 159.0 0.2311E+05
  11 SCC1000 0 0 1000. 0.1590E+06 0.2311E+08
  12 lnSCC Variate 0 0 6.908 11.14 16.96
  13 mu 1
  14 leg(DIM,4) 5 0 0 -1.003 -.1367E-01 1.003
  15 leg(DIM,4).Lactation 15 14 leg(DIM,4): 5 3 Lactation : 3
  16 leg(DIM,2) 3 0 0 -1.003 -.1367E-01 1.003
  17 Lactation.Sire 17367 3 Lactatio: 3 1 Sire : 5789
  18 leg(DIM,2).Lactation.Si 52101 16 leg(DIM,: 3 17 Lactation.Sire :17367
  19 Lactation.Cow 131856 3 Lactatio: 3 2 Cow :43952
  20 leg(DIM,2).Lactation.Co395568 16 leg(DIM,: 3 19 Lactation.Cow :*****
  26688 identity
   26688 units in section 1; initial variance 0.99915
  34017 identity
   34017 units in section 2; initial variance 0.63391
 279948 identity
  279948 units in section 3; initial variance 0.39853
  71752 identity
   71752 units in section 4; initial variance 0.32038
  27272 identity
   27272 units in section 5; initial variance 1.0445
  34546 identity
   34546 units in section 6; initial variance 0.77188
 280210 identity
  280210 units in section 7; initial variance 0.44377
  71632 identity
   71632 units in section 8; initial variance 0.28939
  26964 identity
   26964 units in section 9; initial variance 1.1940
  33975 identity
   33975 units in section 10; initial variance 0.86469
 270069 identity
  270069 units in section 11; initial variance 0.50230
  63271 identity
   63271 units in section 12; initial variance 0.31321
 Warning: US matrix was not positive definite: MODIFIED
      9 UnStructure 0.0298 0.0131 0.0466 0.0045 0.0045 0.0100
    0.0131 0.0217 0.0045 0.0562 0.0007 0.0022 0.0045 0.0013
    0.0046 0.0000 -0.0000 0.0000 -0.0000 -0.0000 0.0100 -0.0008
    0.0001 0.0045 -0.0010 0.0013 0.0045 0.0081 -0.0003 0.0006
    0.0045 -0.0003 0.0006 0.0045 0.0037 0.0102 0.0045 0.0045
    0.0045 0.0045 0.0045 0.0045 0.0045 0.0045 0.0100
   5789 Ainverse
 Structure for leg(DIM,2).Lactation.Sire has 52101 levels defined
 Warning: leg(DIM,2).Lactation.Sire is a 3 x 17367 interaction.
          Are you sure the G structure matches it? 9 x 5789
          Lactation.Sire is a 3 x 5789 interaction.
 Warning: US matrix was not positive definite: MODIFIED
      9 UnStructure 0.7515 0.2561 0.8662 0.0099 0.0099 0.0100
    0.2317 0.4416 0.0099 1.0947 0.0096 0.0392 0.0099 0.0273
    0.1172 0.0099 0.0099 0.0098 0.0099 0.0099 0.0100 -0.0130
   -0.0636 0.0099 0.0151 0.0163 0.0099 0.1804 -0.0217 -0.0222
    0.0099 -0.1066 0.0117 0.0099 0.0484 0.2242 0.0099 0.0099
    0.0099 0.0099 0.0099 0.0099 0.0099 0.0099 0.0099
  43952 identity
 Structure for leg(DIM,2).Lactation.Cow has 395568 levels defined
 Warning: leg(DIM,2).Lactation.Cow is a 3 x131856 interaction.
          Are you sure the G structure matches it? 9 x 43952 ( = 395568)
          Lactation.Cow is a 3 x 43952 interaction.
 Forming 640909 equations: 180 dense.
 Initial updates will be shrunk by factor 0.001
 Notice: LogL values are reported relative to a base of -580000.00

 ** Notice **: Rscore: ASReml prefers I x US R structures for MultiVariate data 6 2

   1 LogL=-8262.56 S2= 1.0000 1027373 df
 Warning: - The estimation was ABORTED because a fault occured.
          The following results are probably partly wrong
               but are printed to help you locate possible causes
          The Standard Errors of variance components and solutions
               and derived quantities are probably wrong.
          Use the DL command line options to get details in the .asl file.
           Rerun using !MAXIT to stop the job earlier
               to get correct standard errors.
 Analysis of Variance NumDF F_inc
  15 leg(DIM,4).Lactation 14 1038.47
   5 SeasonSubClass 149 14.26
   7 CalvingYear 8 13.77
  18 leg(DIM,2).Lactation.Sire 52101 effects fitted ( 18 are zero)
   4 HTD 192800 effects fitted (+ 260 singular)
  20 leg(DIM,2).Lactation.Cow 395568 effects fitted ( 2583 are zero)
 Fault 9 Consider reorganising the model

  Model specification: TERM LEVELS GAMMAS
 CalvingYear 13
 SeasonSubClass 151
 leg(DIM,4).Lactation 15
 mu 1
 Lactation 3
 Sire 5789
 Cow 43952
 leg(DIM,2).Lactation.Sire 52101
 HTD 193060
 leg(DIM,2).Lactation.Cow 395568

  Last line read was: Record Yhat Residual Hat
    20 91229522****** 512000
 Finished: 06 Nov 2009 20:08:27.512 Consider reorganising the model

-------------------- m2f --------------------

Sent using Mail2Forum (http://www.mail2forum.com).

Read this topic online here:
http://www.vsni.co.uk/forum/viewtopic.php?p=1072#1072

-------------------- m2f --------------------

Attachments:
http://www.vsni.co.uk/forum/files/scc_asremlforum_103.txt
Received on Thu Dec 02 2009 - 11:56:24 EST