write_pomdpx
write_pomdpx(P, O, R, gamma, b = rep(1/dim(O)[1], dim(O)[1]), file = "input.pomdpx", digits = 12, digits2 = 12, format = "f")
| P | transition matrix |
|---|---|
| O | observation matrix |
| R | reward |
| gamma | discount factor |
| b | initial belief |
| file | pomdpx file to create |
| digits | precision to round to before normalizing. Leave at 4 since sarsop seems unable to do more? |
| digits2 | precision to write solution to. Leave at 10, since normalizing requires additional precision |
| format | floating point format, because sarsop parser doesn't seem to know scientific notation |