bnelearn.tests.test_util_loss_estimator module¶
Testing correctness of util_loss estimator for a number of settings. Estimates the potential benefit of deviating from the current energy, as:
\[util\_loss(v_i) = Max_{b_i}[ E_{b_{-i}}[u(v_i,b_i,b_{-i})] ]\]
\[util\_loss\_max = Max_{v_i}[ util\_loss(v_i) ]\]
\[util\_loss\_expected = E_{v_i}[ util\_loss(v_i) ]\]
- param agent:
1
- param bid_profile:
\((batch\_size * n\_player * n\_items)\)
- param bid_i:
\((bid\_size * n\_items)\)
- return:
util_loss_max
- return:
util_loss_expected
bid_i always used as val_i and only using truthful bidding
- bnelearn.tests.test_util_loss_estimator.test_ex_interim_util_loss_estimator_fpsb_bne()[source]¶
Test the util_loss in BNE of fpsb. - ex interim util_loss should be close to zero