使用 data.tables 时循环

Question

I have the following (representative) dataset (a very small portion of the actual dataset)

structure(list(Time = c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
1, 1, 1, 1, 1, 1), AgentID = 1:40, State = c(59L, 28L, 84L, 11L, 
5L, 8L, 14L, 71L, 47L, 7L, 84L, 95L, 91L, 92L, 99L, 34L, 70L, 
37L, 55L, 96L, 46L, 38L, 71L, 2L, 61L, 13L, 73L, 26L, 44L, 59L, 
52L, 53L, 42L, 66L, 23L, 11L, 42L, 77L, 38L, 48L), Action = c(-1L, 
-1L, 1L, -1L, 1L, 1L, 1L, -1L, -1L, -1L, -1L, 1L, 1L, 1L, 1L, 
-1L, 1L, 1L, -1L, -1L, 1L, 1L, -1L, 1L, 1L, -1L, -1L, 1L, -1L, 
-1L, 1L, -1L, -1L, 1L, 1L, -1L, -1L, 1L, 1L, 1L), N = c(40L, 
40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 
40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 
40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L
), SimulationID = c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L), discountFactor = c(0, 
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0), i_phase = c(1L, 
-1L, -1L, 1L, 1L, -1L, 1L, 1L, -1L, 1L, 1L, 1L, 1L, 1L, 1L, -1L, 
-1L, 1L, 1L, 1L, 1L, -1L, -1L, 1L, -1L, -1L, -1L, -1L, -1L, -1L, 
1L, -1L, -1L, -1L, -1L, 1L, -1L, 1L, 1L, -1L), i_antiPhase = c(-1, 
1, 1, -1, -1, 1, -1, -1, 1, -1, -1, -1, -1, -1, -1, 1, 1, -1, 
-1, -1, -1, 1, 1, -1, 1, 1, 1, 1, 1, 1, -1, 1, 1, 1, 1, -1, 1, 
-1, -1, 1), totalCount = c(40L, 40L, 40L, 40L, 40L, 40L, 40L, 
40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 
40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 40L, 
40L, 40L, 40L, 40L, 40L, 40L, 40L), phaseCount = c(20L, 20L, 
20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 
20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 
20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L, 20L), 
    phaseCountVar = c(1.24104938271605, 1.15579357351509, 1.15579357351509, 
    1.24104938271605, 1.24104938271605, 1.15579357351509, 1.24104938271605, 
    1.24104938271605, 1.15579357351509, 1.24104938271605, 1.24104938271605, 
    1.24104938271605, 1.24104938271605, 1.24104938271605, 1.24104938271605, 
    1.15579357351509, 1.15579357351509, 1.24104938271605, 1.24104938271605, 
    1.24104938271605, 1.24104938271605, 1.15579357351509, 1.15579357351509, 
    1.24104938271605, 1.15579357351509, 1.15579357351509, 1.15579357351509, 
    1.15579357351509, 1.15579357351509, 1.15579357351509, 1.24104938271605, 
    1.15579357351509, 1.15579357351509, 1.15579357351509, 1.15579357351509, 
    1.24104938271605, 1.15579357351509, 1.24104938271605, 1.24104938271605, 
    1.15579357351509)), row.names = c(NA, -40L), class = c("data.table", 
"data.frame"), .internal.selfref = <pointer: 0x5619bf50c210>)

where i_phase, i_antiPhase == +/-1.

What I want to do is something like:

a[a, antiPhaseAgents:=list(i.AgentID[i_phase==x.i_antiPhase]), on=.(Time,SimulationID,N,SimulationID), by=.(x.AgentID,Time,SimulationID,N,discountFactor)]

In other words, for a given (Time,SimulationID, N, discountFactor) , for all different AgentID s in x , seperately, find AgentID s in i whose i_phase is THE i_antiPhase of the AgentID (from x ) in consideration.

Of course, the syntax in the above code doesn't work, so I'm looking for ways to implement the above operation.

Note: Pure data.table solutions are preferred.

Answer 1

If I understood your question correctly:

a[a,.(x.AgentID,i.AgentID,Time,SimulationID,N,discountFactor,x.i_antiPhase), on=.(Time,SimulationID,N, i_phase=i_antiPhase),allow.cartesian=T][
   ,.(antiPhaseAgents=list(i.AgentID)),by=.(x.AgentID,Time,SimulationID,N,discountFactor)]

    x.AgentID Time SimulationID  N discountFactor       antiPhaseAgents
 1:         2    1            1 40              0  1, 4, 5, 7, 8,10,...
 2:         3    1            1 40              0  1, 4, 5, 7, 8,10,...
 3:         6    1            1 40              0  1, 4, 5, 7, 8,10,...
 4:         9    1            1 40              0  1, 4, 5, 7, 8,10,...
 5:        16    1            1 40              0  1, 4, 5, 7, 8,10,...
 6:        17    1            1 40              0  1, 4, 5, 7, 8,10,...