1 | \secA{Notes and hints on the use of \duchamp} |
---|
2 | \label{sec-notes} |
---|
3 | |
---|
4 | In using \duchamp, the user has to make a number of decisions about |
---|
5 | the way the program runs. This section is designed to give the user |
---|
6 | some idea about what to choose. |
---|
7 | |
---|
8 | The main choice is whether or not to use the wavelet |
---|
9 | reconstruction. The main benefits of this are the marked reduction in |
---|
10 | the noise level, leading to regularly-shaped detections, and good |
---|
11 | reliability for faint sources. The main drawback with its use is the |
---|
12 | long execution time: to reconstruct a $170\times160\times1024$ |
---|
13 | (\hipass) cube often requires three iterations and takes about 20-25 |
---|
14 | minutes to run completely. Note that this is for the three-dimensional |
---|
15 | reconstruction: using \texttt{reconDim=1} makes the reconstruction |
---|
16 | quicker (the full program then takes about 6 minutes), but it is still |
---|
17 | the largest part of the time. |
---|
18 | |
---|
19 | The searching part of the procedure is much quicker: searching an |
---|
20 | un-reconstructed cube leads to execution times of only a couple of |
---|
21 | minutes. Alternatively, using the ability to read in previously-saved |
---|
22 | reconstructed arrays makes running the reconstruction more than once a |
---|
23 | more feasible prospect. |
---|
24 | |
---|
25 | On the positive side, the shape of the detections in a cube that has |
---|
26 | been reconstructed will be much more regular and smooth -- the ragged |
---|
27 | edges that objects in the raw cube possess are smoothed by the removal |
---|
28 | of most of the noise. This enables better determination of the shapes |
---|
29 | and characteristics of objects. |
---|
30 | |
---|
31 | A further point to consider when using the reconstruction is that if |
---|
32 | the two-dimensional reconstruction is chosen (\texttt{reconDim=2}), it |
---|
33 | can be susceptible to edge effects. If the valid area in the cube (\ie |
---|
34 | the part that is not BLANK) has non-rectangular edges, the convolution |
---|
35 | can produce artefacts in the reconstruction that mimic the edges and |
---|
36 | can lead (depending on the selection threshold) to some spurious |
---|
37 | sources. Caution is advised with such data -- the user is advised to |
---|
38 | check carefully the reconstructed cube for the presence of such |
---|
39 | artefacts. Note, however, that the 1- and 3-dimensional |
---|
40 | reconstructions are \emph{not} susceptible in the same way, since the |
---|
41 | spectral direction does not generally exhibit these BLANK edges, and |
---|
42 | so we recommend the use of either of these. |
---|
43 | |
---|
44 | If one chooses the reconstruction method, a further decision is |
---|
45 | required on the signal-to-noise cutoff used in determining acceptable |
---|
46 | wavelet coefficients. A larger value will remove more noise from the |
---|
47 | cube, at the expense of losing fainter sources, while a smaller value |
---|
48 | will include more noise, which may produce spurious detections, but |
---|
49 | will be more sensitive to faint sources. Values of less than about |
---|
50 | $3\sigma$ tend to not reduce the noise a great deal and can lead to |
---|
51 | many spurious sources (although this will depend on the nature of the |
---|
52 | cube). |
---|
53 | |
---|
54 | When it comes to searching, the FDR method produces more reliable |
---|
55 | results than simple sigma-clipping, particularly in the absence of |
---|
56 | reconstruction. However, it does not work in exactly the way one |
---|
57 | would expect for a given value of \texttt{alpha}. For instance, |
---|
58 | setting fairly liberal values of \texttt{alpha} (say, 0.1) will often |
---|
59 | lead to a much smaller fraction of false detections (\ie much less |
---|
60 | than 10\%). This is the effect of the merging algorithms, that combine |
---|
61 | the sources after the detection stage, and reject detections not |
---|
62 | meeting the minimum pixel or channel requirements. It is thus better |
---|
63 | to aim for larger \texttt{alpha} values than those derived from a |
---|
64 | straight conversion of the desired false detection rate. |
---|
65 | |
---|
66 | Finally, as \duchamp\ is still undergoing development, there are some |
---|
67 | elements that are not fully developed. In particular, it is not as |
---|
68 | clever as I would like at avoiding interference. The ability to place |
---|
69 | requirements on the minimum number of channels and pixels partially |
---|
70 | circumvents this problem, but work is being done to make \duchamp\ |
---|
71 | smarter at rejecting signals that are clearly (to a human eye at |
---|
72 | least) interference. See the following section for further |
---|
73 | improvements that are planned. |
---|