      update  step  correct-short   correct-long    wrong-short     wrong-long  E[len | wrong]  E[len | correct]
    unbiased     0         0.2500         0.2500         0.2500         0.2500            5.00              5.00
    unbiased    20         0.4340         0.4340         0.0660         0.0660            5.00              5.00
    unbiased   100         0.4887         0.4887         0.0113         0.0113            5.00              5.00
    unbiased   400         0.4974         0.4974         0.0026         0.0026            5.00              5.00
      length     0         0.2500         0.2500         0.2500         0.2500            5.00              5.00
      length    20         0.4281         0.2529         0.1301         0.1890            5.55              4.23
      length   100         0.8073         0.1166         0.0233         0.0527            6.16              2.76
      length   400         0.9424         0.0411         0.0043         0.0122            6.43              2.25
         std     0         0.2500         0.2500         0.2500         0.2500            5.00              5.00
         std    20         0.4829         0.4829         0.0171         0.0171            5.00              5.00
         std   100         0.4992         0.4992         0.0008         0.0008            5.00              5.00
         std   400         0.5000         0.5000         0.0000         0.0000            5.00              5.00
        both     0         0.2500         0.2500         0.2500         0.2500            5.00              5.00
        both    20         0.6159         0.2013         0.0643         0.1184            5.89              3.48
        both   100         0.9508         0.0357         0.0035         0.0100            6.46              2.22
        both   400         0.9931         0.0059         0.0002         0.0008            6.66              2.04

expected gradient components at the uniform policy (up to the common factor 1/4)
    unbiased: [ 0.5  0.5 -0.5 -0.5]
      length: [ 0.25    0.0625 -0.25   -0.0625]
         std: [ 1.  1. -1. -1.]
        both: [ 0.5    0.125 -0.5   -0.125]
