Table 2 Estimated variance components of two Markov chain Monte Carlo algorithms and REML based on 10 simulation replicates

From: Bayesian adaptive Markov chain Monte Carlo estimation of genetic parameters

 

σa2

σd2

σe2

 

Avg

Min

Max

Avg

Min

Max

Avg

Min

Max

Learning phase

 Mean

778.04

659.95

873.01

593.67

512.86

698.77

3031.13

2845.09

3184.00

 Median

767.32

627.37

831.31

584.36

503.76

690.93

3057.02

2894.05

3191.20

 Mode

728.63

518.87

879.81

568.09

515.13

658.08

3008.63

2737.08

3282.07

Adapted phase

 Mean

774.70

732.99

839.36

577.07

545.11

620.89

3065.10

2901.00

3184.05

 Median

764.48

718.31

827.60

574.23

529.86

615.08

3069.23

2907.08

3191.20

 Mode

779.02

747.16

838.08

580.98

531.31

634.40

3103.76

2928.00

3284.50

 REML estimates

731.27

588.17

881.47

555.99

527.02

629.06

3064.94

2928.00

3174.91

 True values

800.00

  

600.00

  

3025.00

  
  1. Abbreviation: REML, residual maximum likelihood.
  2. True simulated values are also shown. Average, minimum and maximum values of the posterior (mean, median and mode) and REML estimates are calculated for variance components from the analyses of 10 simulation replicates.