r - Merging data.tables uses more than 10 GB RAM

Question

I have two data.tables: DT and meta. When I merge them using DT[meta], memory usage increases by more than 10 GB (and the merge is very slow). What's going wrong? It seems like the merge is successful, but I can only look at single lines, otherwise I run out of memory. DT itself was created by merging two data.tables without any problems.

Edit:

It seems to be a problem with the key. I can do the following without a problem:

DT[,id:=1:nrow(DT)]
meta[,id:=1:nrow(DT)]
setkey(DT,id)
setkey(meta,id)

DT2<-DT[meta]   # Comment from Matthew Dowle:
                # X[Y] (or merge) on a key of 1:nrow(DT) is just a cbind, isn't it? 

unique(DT2[,"Moor_ID",with=F]==DT2[,"Moor_ID.1",with=F])
     Moor_ID
[1,]    TRUE

First data.table:

str(DT)
Classes ‘data.table’ and 'data.frame':  10212 obs. of  55 variables:
 $ DWD_ID                 : chr  "Bremerhav" "Bremerhav" "Bremerhav" "Bremerhav" ...
 $ numdays                : int  1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 ...
 $ days                   : Date, format: "2009-09-01" "2009-09-02" "2009-09-03" "2009-09-04" ...
 $ TBoden_dayAnzahl       : int  0 0 0 0 0 0 0 0 0 0 ...
 $ TBoden_dayMin          : num  NA NA NA NA NA NA NA NA NA NA ...
 $ TBoden_dayMax          : num  NA NA NA NA NA NA NA NA NA NA ...
 $ TBoden_dayMeanAR       : num  NA NA NA NA NA NA NA NA NA NA ...
 $ TBoden_dayStabw        : num  NA NA NA NA NA NA NA NA NA NA ...
 $ TBoden_dayMedian       : num  NA NA NA NA NA NA NA NA NA NA ...
 $ TBoden_dayMeanMM       : num  NA NA NA NA NA NA NA NA NA NA ...
 $ T2m_dayAnzahl          : int  0 0 0 0 0 0 0 0 0 0 ...
 $ T2m_dayMin             : num  15.6 13.8 13.7 12.8 13.5 13.1 13.3 13.8 15.9 13.7 ...
 $ T2m_dayMax             : num  25.6 19.9 18.1 18.1 16.9 18.6 21 25.7 19.3 17.6 ...
 $ T2m_dayMeanAR          : num  19 16.9 15.6 15.2 14.8 ...
 $ T2m_dayStabw           : num  3.409 2.048 1.334 1.726 0.965 ...
 $ T2m_dayMedian          : num  17.2 16.8 15.2 14.8 14.5 ...
 $ T2m_dayMeanMM          : num  20.6 16.9 15.9 15.4 15.2 ...
 $ T10cm_dayAnzahl        : int  0 0 0 0 0 0 0 0 0 0 ...
 $ T10cm_dayMin           : num  14.3 12.6 12.9 12.2 12.7 12 12.8 11.7 15.1 12.2 ...
 $ T10cm_dayMax           : num  27.7 20.9 18.7 18.7 17.4 19.8 22.4 25.9 21.8 18.6 ...
 $ T10cm_dayMeanAR        : num  18.7 16.5 14.9 15.1 14.5 ...
 $ T10cm_dayStabw         : num  4.36 2.84 1.73 2.36 1.54 ...
 $ T10cm_dayMedian        : num  16.1 15.6 14.3 14.2 14 ...
 $ T10cm_dayMeanMM        : num  21 16.8 15.8 15.4 15.1 ...
 $ RF_dayAnzahl           : int  0 0 0 0 0 0 0 0 0 0 ...
 $ RF_dayMin              : num  45 58 73 56 68 62 63 44 65 58 ...
 $ RF_dayMax              : num  94 94 94 93 94 92 84 84 89 84 ...
 $ RF_dayMean             : num  68.6 76.3 78.9 74.4 86.5 ...
 $ RF_dayStabw            : num  17.09 12.53 5.88 9.83 5.62 ...
 $ RF_dayMedian           : num  64.5 74 77.5 76 87.5 77.5 75 63 77 76 ...
 $ Luftdruck_dayMean      : num  100.8 101 99.7 99.9 101.1 ...
 $ es_day                 : num  2.53 1.95 1.82 1.78 1.74 ...
 $ ea_day                 : num  1.57 1.42 1.49 1.27 1.38 ...
 $ defi_day               : num  0.956 0.535 0.327 0.509 0.355 ...
 $ Nebel_dayAnteil        : num  0 0 0 0 0 0 0 0 0 0 ...
 $ Sonnenscheind_dayAnzahl: int  18 18 18 18 18 18 18 18 18 18 ...
 $ Sonnenscheind_daySum   : num  6.63 4.93 1.05 5.82 3.27 ...
 $ julian_day             : int  244 245 246 247 248 249 250 251 252 253 ...
 $ zeta_day               : num  2.81 2.82 2.84 2.86 2.88 ...
 $ maxSonnenscheind       : num  13.9 13.8 13.7 13.6 13.5 ...
 $ R0_day                 : num  2920 2890 2860 2830 2799 ...
 $ Globalstrahlung_dayMean: num  NA NA NA NA NA NA NA NA NA NA ...
 $ RG_day                 : num  13.24 11.19 6.64 12.02 9.03 ...
 $ lambdaET_day           : num  2.45 2.46 2.46 2.46 2.47 ...
 $ sAnstieg_day           : num  0.15 0.122 0.116 0.113 0.111 ...
 $ gamma_day              : num  0.067 0.0669 0.0659 0.0661 0.0668 ...
 $ ETp_TW_day             : num  2.71 2.15 1.28 2.24 1.68 ...
 $ Moor_ID                : chr  "Ahlenmoor" "Ahlenmoor" "Ahlenmoor" "Ahlenmoor" ...
 $ Distanz_in_km          : num  24 24 24 24 24 ...
 $ North                  : num  53.5 53.5 53.5 53.5 53.5 ...
 $ East                   : num  8.58 8.58 8.58 8.58 8.58 ...
 $ Hoehe_in_m             : num  7 7 7 7 7 7 7 7 7 7 ...
 $ Kueste_km              : num  20 20 20 20 20 ...
 $ peatland               : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...
 $ diffmaxt2m             : num  -1.6 0 -0.2 0.1 -0.4 ...
 - attr(*, "sorted")= chr "Moor_ID"
 - attr(*, ".internal.selfref")=<externalptr> 

Second data.table:

str(meta)
Classes ‘data.table’ and 'data.frame':  10212 obs. of  6 variables:
 $ Moor_ID        : chr  "Ahlenmoor" "Ahlenmoor" "Ahlenmoor" "Ahlenmoor" ...
 $ Hoehe_Moor     : num  2.35 2.35 2.35 2.35 2.35 2.35 2.35 2.35 2.35 2.35 ...
 $ Kueste_km      : num  15.7 15.7 15.7 15.7 15.7 ...
 $ WSPsommer_muGOK: num  0.699 0.699 0.699 0.699 0.699 ...
 $ WSPwinter_muGOK: num  0.446 0.446 0.446 0.446 0.446 ...
 $ Moorgroesse_km2: num  59 59 59 59 59 59 59 59 59 59 ...
 - attr(*, ".internal.selfref")=<externalptr> 
 - attr(*, "sorted")= chr "Moor_ID"

Session info:

R version 2.15.1 (2012-06-22)
Platform: x86_64-pc-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=German_Germany.1252  LC_CTYPE=German_Germany.1252    LC_MONETARY=German_Germany.1252 LC_NUMERIC=C                   
[5] LC_TIME=German_Germany.1252    

attached base packages:
[1] grDevices datasets  splines   graphics  stats     tcltk     utils     methods   base     

other attached packages:
[1] reshape_0.8.4    plyr_1.7.1       data.table_1.8.0 svSocket_0.9-53  TinnR_1.0-5      R2HTML_2.2       Hmisc_3.9-3     
[8] survival_2.36-14

loaded via a namespace (and not attached):
[1] cluster_1.14.2 grid_2.15.1    lattice_0.20-6 svMisc_0.9-65  tools_2.15.1 

Answer 1

My bad. The problem was that keys were not unique:

a<-data.table(x=c(1,1),y=c(1,2))
b<-data.table(x=c(1,1),y=c(3,4))
setkey(a,x)
setkey(b,x)
a[b]
     x y y.1
[1,] 1 1   3
[2,] 1 2   3
[3,] 1 1   4
[4,] 1 2   4

It would be nice if data.table could give a warning for that.

---

Update from Matthew

This warning has now been implemented in v1.8.7 :

New argument allow.cartesian ( default FALSE) added to X[Y] and merge(X,Y), #2464. Prevents large allocations due to misspecified joins; e.g., duplicate key values in Y joining to the same group in X over and over again. The word cartesian is used loosely for when more than max(nrow(X),nrow(Y)) rows would be returned. The error message is verbose and includes advice.