[R] inconsistency in pbmclapply...

[akshay kulkarni]

dear members,
                             I am using parallel processing with pbmclapply. below is the code. LYG1[[1]] got from pbmclapply is not the same as LYG1[[1]] got from direct application of auto.arima. what may be wrong?

              debug at <tmp>#20: LYG1 <- pbmclapply(LYGH, FUN = auto.arima, mc.cores = detectCores())
Browse[2]>
  |=========================================                   |  68%, ETA 00:21
debug at <tmp>#21: LYG2 <- pbmclapply(LYGH, FUN = ets, mc.cores = detectCores())
Browse[2]> LYG1[[1]]
[1] "Error in ts(x) : 'ts' object must have one or more observations\n"
attr(,"class")
[1] "try-error"
attr(,"condition")
<simpleError in ts(x): 'ts' object must have one or more observations>
Warning message:
In pbmclapply(LYGH, FUN = auto.arima, mc.cores = detectCores()) :
  scheduled cores encountered errors in user code
Browse[2]> LYGH[[1]]
 [1] 0.700000000 0.400000000 0.300000000 0.150000000 0.250000000 0.950000000
 [7] 1.000000000 0.300000000 0.650000000 0.200000000 0.600000000 0.100000000
[13] 0.001412873 1.550000000 0.150000000 0.300000000 0.450000000 0.350000000
[19] 0.150000000 2.350000000 0.250000000 0.100000000 3.700000000 3.950000000
[25] 3.050000000 0.900000000 0.400000000 1.050000000 1.100000000 1.950000000
[31] 2.000000000 0.650000000 0.700000000 0.250000000 5.250000000 0.800000000
[37] 0.001412873
Browse[2]> auto.arima(LYGH[[1]])
Series: LYGH[[1]]
ARIMA(0,1,0)

sigma^2 estimated as 2.303:  log likelihood=-66.1
AIC=134.2   AICc=134.31   BIC=135.78

very many thanks for your time and effort.....
yours sincerely,
AKSHAY M KULKARNI

	[[alternative HTML version deleted]]