### Table I Content of Base A, C, G, and T at Three Codon Positions

### Table 3. Architectural details. Parameter Brief Explanation

### Table 1: The refinement algorithm for landmark sets on source and target geometries; see Section 6.2 for detailed explanation.

"... In PAGE 6: ...arget geometries; see Section 6.2 for detailed explanation. possible in typically two or three iterations.1 The algorithm outline is shown in Table1 . To be able to refine our standard set of land- marks automatically, we interpret the landmarks as the vertices of a triangle mesh, which we will call a feature mesh in the following.... ..."

### Table 1: The refinement algorithm for landmark sets on source and target geometries; see Section 6.2 for detailed explanation.

"... In PAGE 6: ...arget geometries; see Section 6.2 for detailed explanation. possible in typically two or three iterations.1 The algorithm outline is shown in Table1 . To be able to refine our standard set of land- marks automatically, we interpret the landmarks as the vertices of a triangle mesh, which we will call a feature mesh in the following.... ..."

### Table 2.3: Steps of selection and results for the 0-! selections. Detailed explanations are given in the text.

in CB note 302

### TABLE I THIS TABLE SHOWS THE COMPLETE LEARNING ALGORITHM FOR OPERATIONAL SPACE CONTROL. SEE TEXT OF DETAILED EXPLANATIONS.

### Table 1: Statistics for estimations of the largest Lyapunov exponent of different dynam- ical systems. All exponents are measured in bits. See text for detailed explanations.

2001

"... In PAGE 7: ... We generated a time series of length 1000 and applied the method described in the previous subsection for embedding di- mensions m = 1; 5; 10; 15. The results are summarized in Table1 . For all embedding dimensions, the estimated mean value is close to the true Lyapunov exponent = 1:0 which is contained in the corresponding confidence intervals.... In PAGE 9: ...hese volatility clusters are clearly visible in Fig. 3. First, we estimate the largest Lyapunov exponent (with confidence intervals) for the ATX data set for embedding dimensions m = 1; 5; 10; 15. The results are summa- rized in Table1 and depicted on the left-hand side of Fig. 4.... In PAGE 10: ... Again, increases with increasing m. For m = 10 and m = 15, is positive (see also Table1 ). For m = 15, 0 is not even included in the confidence interval.... ..."

Cited by 1

### Table 5. Contributions of prime ideals and zeros to discriminant bounds for some elds. (See Section 6 for detailed explanation.) D n

### Table 1: Statistics for estimations of the largest Lyapunov exponent of di erent dynamical systems. All exponents are measured in bits. See text for detailed explanations.

"... In PAGE 8: ... We generated a time series of length 1000 and applied the method described in the previous subsection for embedding dimensions m = 1; 5; 10; 15. The results are summarized in Table1 . For all embedding dimensions, the estimated mean value is close to the true Lyapunov exponent = 1:0 which is contained in the corresponding con dence intervals.... In PAGE 10: ... From the clouds of points it is very hard to determine in which case the skeleton (the deterministic part) is periodic and in which case it is chaotic. The estimated Lyapunov exponents and con dence intervals (see Table1 ) are also very simi- lar. In other words, the algorithm estimates the Lyapunov exponent very reliably for the disturbed chaotic system, but it also returns a positive value for the disturbed periodic system.... In PAGE 10: ...isible in Fig. 3. First, we estimate the largest Lyapunov exponent (with con dence intervals) for the ATX data set for embedding dimensions m = 1; 5; 10; 15. The results are summarized in Table1 and depicted on the left-hand side of Fig. 4.... ..."

### Table 5. Contributions of prime ideals and zeros to discriminant bounds for some elds. (See Section 6 for detailed explanation.) D n

"... In PAGE 9: ...1. What are the relative contributions of prime ideals and zeros to the explicit formula for minimal discriminants? Table5 suggests that these contributions are of comparable magnitude, but it would be nice to obtain data for higher degree elds. (This should not be too di cult, since the elds of small discriminant found by Martinet in [C3] are given quite explicitly as ray class elds.... In PAGE 10: ...3), and obtain the contribution of the zeros by subtraction. Table5 presents the results of the computation that was carried out for six elds,... ..."