|
100 | 100 |
|
101 | 101 | * *Entropy* is a non-negative number that measures the size of a set of models surrounding the decision-maker's approximating model.
|
102 | 102 |
|
103 |
| - * Entropy is zero when the set includes only the approximating model, indicating that the decision-maker completely trusts the approximating model. |
| 103 | + * Entropy is zero when the set includes only the approximating model, indicating that the decision-maker completely trusts the approximating model. |
104 | 104 |
|
105 |
| - * Entropy is bigger, and the set of surrounding models is bigger, the less the decision-maker trusts the approximating model. |
| 105 | + * Entropy is bigger, and the set of surrounding models is bigger, the less the decision-maker trusts the approximating model. |
106 | 106 |
|
107 | 107 | The shaded region indicates that for **all** models having entropy less than or equal to the number on the horizontal axis, the value obtained will be somewhere within the indicated set of values.
|
108 | 108 |
|
@@ -132,9 +132,9 @@ Below we'll explain in detail how to construct these sets of values for a given
|
132 | 132 |
|
133 | 133 | Here is a hint about the *secret weapons* we'll use to construct these sets
|
134 | 134 |
|
135 |
| - * We'll use some min problems to construct the lower bounds |
| 135 | +* We'll use some min problems to construct the lower bounds |
136 | 136 |
|
137 |
| - * We'll use some max problems to construct the upper bounds |
| 137 | +* We'll use some max problems to construct the upper bounds |
138 | 138 |
|
139 | 139 | We will also describe how to choose :math:`F` to shape the sets of values.
|
140 | 140 |
|
@@ -540,15 +540,16 @@ In particular, the lower bound on the left side of :eq:`rob_bound` is attained w
|
540 | 540 |
|
541 | 541 | To construct the *lower bound* on the set of values associated with all perturbations :math:`{\mathbf w}` satisfying the entropy constraint :eq:`rb_dec` at a given entropy level, we proceed as follows:
|
542 | 542 |
|
543 |
| - * For a given :math:`\theta`, solve the minimization problem :eq:`rb_a2o`. |
| 543 | +* For a given :math:`\theta`, solve the minimization problem :eq:`rb_a2o`. |
544 | 544 |
|
545 |
| - * Compute the minimizer :math:`R_\theta(x_0, F)` and the associated entropy using :eq:`rb_pdt22`. |
| 545 | +* Compute the minimizer :math:`R_\theta(x_0, F)` and the associated entropy using :eq:`rb_pdt22`. |
546 | 546 |
|
547 |
| - * Compute the lower bound on the value function :math:`R_\theta(x_0, F) - \theta \ {\rm ent}` and plot it against :math:`{\rm ent}`. |
| 547 | +* Compute the lower bound on the value function :math:`R_\theta(x_0, F) - \theta \ {\rm ent}` and plot it against :math:`{\rm ent}`. |
548 | 548 |
|
549 |
| - * Repeat the preceding three steps for a range of values of :math:`\theta` to trace out the lower bound. |
| 549 | +* Repeat the preceding three steps for a range of values of :math:`\theta` to trace out the lower bound. |
550 | 550 |
|
551 | 551 | .. note::
|
| 552 | + |
552 | 553 | This procedure sweeps out a set of separating hyperplanes indexed by different values for the Lagrange multiplier :math:`\theta`.
|
553 | 554 |
|
554 | 555 | The Upper Bound
|
@@ -608,13 +609,13 @@ The upper bound on the left side of :eq:`robboundmax` is attained when
|
608 | 609 |
|
609 | 610 | To construct the *upper bound* on the set of values associated all perturbations :math:`{\mathbf w}` with a given entropy we proceed much as we did for the lower bound
|
610 | 611 |
|
611 |
| - * For a given :math:`\tilde \theta`, solve the maximization problem :eq:`rba2omax`. |
| 612 | +* For a given :math:`\tilde \theta`, solve the maximization problem :eq:`rba2omax`. |
612 | 613 |
|
613 |
| - * Compute the maximizer :math:`V_{\tilde \theta}(x_0, F)` and the associated entropy using :eq:`rbpdt223`. |
| 614 | +* Compute the maximizer :math:`V_{\tilde \theta}(x_0, F)` and the associated entropy using :eq:`rbpdt223`. |
614 | 615 |
|
615 |
| - * Compute the upper bound on the value function :math:`V_{\tilde \theta}(x_0, F) + \tilde \theta \ {\rm ent}` and plot it against :math:`{\rm ent}`. |
| 616 | +* Compute the upper bound on the value function :math:`V_{\tilde \theta}(x_0, F) + \tilde \theta \ {\rm ent}` and plot it against :math:`{\rm ent}`. |
616 | 617 |
|
617 |
| - * Repeat the preceding three steps for a range of values of :math:`\tilde \theta` to trace out the upper bound. |
| 618 | +* Repeat the preceding three steps for a range of values of :math:`\tilde \theta` to trace out the upper bound. |
618 | 619 |
|
619 | 620 |
|
620 | 621 | Reshaping the Set of Values
|
@@ -922,9 +923,9 @@ Here is a brief description of the methods of the class
|
922 | 923 | * ``robust_rule()`` and ``robust_rule_simple()`` both solve for the
|
923 | 924 | triple :math:`\hat F, \hat K, \hat P`, as described in equations :eq:`rb_oc_ih` -- :eq:`rb_kd` and the surrounding discussion
|
924 | 925 |
|
925 |
| - * ``robust_rule()`` is more efficient |
| 926 | + * ``robust_rule()`` is more efficient |
926 | 927 |
|
927 |
| - * ``robust_rule_simple()`` is more transparent and easier to follow |
| 928 | + * ``robust_rule_simple()`` is more transparent and easier to follow |
928 | 929 |
|
929 | 930 | * ``K_to_F()`` and ``F_to_K()`` solve the decision problems
|
930 | 931 | of :ref:`agent 1 <rb_a1>` and :ref:`agent 2 <rb_a2>` respectively
|
|
0 commit comments