### Table 1. De nition Introduction Phase

2002

"... In PAGE 9: ...nnermost (i.e., ti does not contain operation-rooted subterms) such that each one of them shares at least a common variable with at least one more subterm in T 2. Apply the DEFINITION INTRODUCTION RULE to generate : Rdef = (fnew(x) ! T) where fnew is a new function symbol not appearing R, and x is the set of variables of T OUTPUT: Definition Rule (Eureka) Rdef In order to avoid these risks, our approach generates eureka de nition following a very simple strategy ( Table1 ) that obtains similar levels of generality than pre- vious approaches and covers most practical cases. The main advantages are that the analysis is terminating, easy and quickly, and does not perform redundant calculus (like unfolding steps, instantiations, etc.... In PAGE 9: ...) that properly corresponds to subsequent phases. As illustrated by step 1 in Example 2, our eureka generator proceeds as the algorithm in Table1 shows. Observe that the input of the algorithm is the original program R and a selected rule R 2 R which de nition is intended to be 4 This fact is observed during the so-called quot;program extraction phase quot; in [20].... In PAGE 13: ...contrast with the previous case, where an optimal de nition of fnew is obtained, now the process tries with rules in R in order to replace and reuse as much as possible the optimized de nition of fnew into the original program R. In particular, we know that there exists at least a rule R = (l ! r) 2 R (the one considered in Table1 to generate de eureka Rdef) that veri es TEST(R; Rdef)=2. Hence, similarly as done before, we can apply an abstraction and folding steps to it, obtaining the new rule: l ! r[zj]Pj where hz1; : : : ; zni = fnew(x): where the call to fnew enhances the nal de nition of the old function symbol that roots l.... ..."

### Table 3: The introduction of new roles in the tax preparation business (adapted from [19])

### Table 1: Adaptive Scheduling Algorithms.

2000

Cited by 64

### Table 1: Adaptive Scheduling Algorithms.

2000

Cited by 64

### Table 1: Parameters of the adaptive algorithm.

2006

"... In PAGE 6: ...2 Parameter search Now that we have chosen a parameterized adaptive algo- rithm and have a means of generating bidder behavior, we are ready to search for the set of parameters that results in the best expected performance. (For reference, all parame- ters are described in Table1 .) For any given set of parameter values, we can obtain an estimate of the expected revenue from an episode by generating a population of bidders as described in Section 3.... ..."

Cited by 2

### Table 1: Parameters of the adaptive algorithm.

"... In PAGE 6: ...2 Parameter search Now that we have chosen a parameterized adaptive algo- rithm and have a means of generating bidder behavior, we are ready to search for the set of parameters that results in the best expected performance. (For reference, all parame- ters are described in Table1 .) For any given set of parameter values, we can obtain an estimate of the expected revenue from an episode by generating a population of bidders as described in Section 3.... ..."

### TABLE I ADAPTATION ALGORITHM SUMMARY

2003

Cited by 6

### TABLE II ADAPTING COEFFICIENTS OF ALGORITHM

### Table 1: Adaptive QR algorithm.

### TABLE 5. Adaptive algorithm perfor-

2007