### Table 1: Variation of proportion of image extrema with scale. Fractions give the number of extreme regions the total number of regions.

### Table 2. An extreme case showing the difference between parallel region and the proposed scheme.

### TABLE II Annual mean maximum and minimum temperature ( C), and pre- cipitation (mm) for the observations and the downscaling performed with the control simulation averaged over the MINK area. Also shown are the model produced mean temperature and precipitation averaged over the MINK region. Included are the means, standard deviations and extreme values for each variable

1999

Cited by 1

### Table 6.1 gives the relative compression rates and processing speeds for smooth filling, cosine filling, and region dependent transform coding. The compression rates in the table directly follow from the compression curves given earlier. As one might expect, the encoding process of smooth filling is the fastest. Cosine filling and region dependent transform coding have comparable encoding speeds; however, there is a vast difference in their decoding speeds. Recall that both smooth filling and cosine filling use the 8 x 8 DCT as transform. These two filling methods can be considered to be JPEG compliant because their decoding only requires JPEG decompression. Since the DCT has an efficient hardware implementation, JPEG decompression is extremely fast. On the other hand, for region dependent transform coding, the decoder is as computationally intensive as the encoder. The decoding process for region dependent transform coding is at least an order of magnitude slower than that for the two filling methods.

### (Table 3) from the oxidation of isoprene in the model, consistent with the values derived from the chemical mech- anism (Table 2 and Figure 1), and confirming the robustness of the overall approach. The apparent low yield computed for the SW is due to the effect of smearing in a highly heterogeneous isoprene emission field. The definition of SW and SE quadrants dissects a region of relatively low isoprene emission (SW) and an extremely active isoprene emitting region (SE) (see Figure 4). The related manifes- tation of this smearing is apparent in the SE quadrant, as an occasional enhancement of the HCHO column over areas without isoprene emissions. In section 4, we will use the model slopes of Figure 3 as a transfer function to convert the HCHO columns observed by GOME into isoprene emission fluxes.

2003

Cited by 7

### Table 3. The occurrences of the association (NF-E= gt;atgcaa) in functional catalogue of Glycolysis and gluconeogenesis 4 Discussion This study finds combinations of known TF binding sites and over-represented repetitive oligonucleotides located within the promoter regions of groups of functional related genes. The discovering enormous number of associations makes it extremely difficult to identify those interesting and useful ones. Chi-square significance level is then used to remove those

1960

"... In PAGE 4: ... The association rules mined after applying Chi-square test (Partial). Table3 shows an example of the occurrences of the association, [NF-E= gt;atgcaa], in functional catalogue Glycolysis and gluconeogenesis. The first column in Table 2 is the ORF names; the second one is the locations of known TF binding sites or putative regulatory site in the promoter region.... ..."

Cited by 2

### Table 1 Noise densities and associated Kullback-Leibler distances. where C gt; 0 is the bound on the amplitudes of the signals. This problem has MK + 1 variables and M(M + 1)=2 + MN + 1 inequality constraints. A typical problem would have K = 2; M = 16; N = 50, giving 33 variables and 937 constraints. These nonlinear programming problems all share the characteristic that they have extremely \small quot; feasible regions. 3. Noise distributions. The primary purpose of this paper is to investigate non-Gaussian noise distributions. Following Johnson and Orsak (see [18]), we selected the ve densities found in Table 1 (including the Gaussian density for comparison). These densities are graphed in Figure 1, while the associated Kullback-Leibler distances are found in Figure 2.

1999

Cited by 6

### Table 3: Performance results for the 36 synthetic cases XL, with 8 R10000 processors, 512MB of RAM, and 2MB of level 2 cache per processor. The run time was de ned as the sum of the time that the CPU spends running instructions in carabeamer and the time that the operating system spends running on behalf of carabeamer. The results of Table 3 depend somewhat on the initial number of beams, but they are indicative of carabeamer apos;s performance. Figures 28 displays the 80% and 90% isodose surfaces for each of the tissue geometries when the tightest constraints on the dose distribution were used. In all cases, the dose distribution is skewed away from the critical region. That is, the fallo around the tumor is more rapid on the sides that are adjacent to the critical region. In all cases, the region of high dose matches the tumor volume extremely well.

1999

"... In PAGE 27: ... The 36 cases were obtained by forming all combinations of the 4 geometries, the 3 constraint sets, and the 3 maximum beam set sizes. Table3 summarizes the performance results obtained. Each of the 4 major columns refer to one of the 4 geometrical cases.... ..."

Cited by 12

### Table 3. Performance for the re nement step when edges are marked in a single region of the global mesh

1996

"... In PAGE 12: ... This is because the computational time will increase while the percentage of elements along processor boundaries will decrease (and so too will the communication time). Table3 shows the timings and speedup when edges are marked in a single region of the global mesh. The performance is extremely poor, with speedups of only 5.... ..."

Cited by 17

### Table 1 Background events remaining in the K eqeyg signal region L

"... In PAGE 6: ... L Backgrounds from three other sources are large enough to be estimated in a practical way though they remain extremely small fractions of the signal. These are discussed below and the numbers of events which remain in the signal region from these sources are summarized in Table1 . Backgrounds from the first two sources were obtained from Monte Carlo studies, but have been cross-checked with data.... ..."