oversampling not being done properly

[b1quint]

Philippe created a dummy cube with 36 channels with 100 everywhere in channel 1 and 0 in the 35 others channels.

Within 2 pixels, this gives:

1 200.000000
2 0.000000
3 0.000000
4 0.000000
5 0.000000
105 0.000000
106 0.000000
107 0.000000
108 0.000000

Thus I repeat 3 times the cubes, this gives:

1 200.000000
2 0.000000
35 0.000000
36 0.000000
37 200.000000
38 0.000000
71 0.000000
72 0.000000
73 200.000000
74 0.000000
107 0.000000
108 0.000000

Which is absolutely normal

Thus I oversample 3 times to get a 324 channel-cube, thus I cut it to 108 and I get:

1 177.708984
2 110.835915
3 43.962849
4 0.000000
5 0.000000
6 0.000000
7 0.000000
101 0.000000
102 0.000000
103 0.000000
104 0.000000
105 0.000000
106 43.962849
107 110.835915
108 177.708984

Which is a little bit strange, I should expect something like

1 177.708984
2 110.835915
3 43.962849
4 0.000000
5 0.000000
6 0.000000
7 0.000000
101 0.000000
102 0.000000
103 0.000000
104 0.000000
105 0.000000
106 0
107 43.962849
108 110.835915

From my point of view there is a shift of one channel somewhere…

Another strange thing is that the sum does not give 600 (if you normalize to keep the flux conserved it should give 200)

In [2]: (177.708984 + 110.835915 + 43.962849)*2
Out[2]: 665.015496

Here are the commands I used:

fp_repeat -b 1 -e 1 dummy.fits dummy_108_repeat.fits
fp_oversample dummy_108_repeat.fits dummy_324_oversample.fits 3
fp_cut --n_begin 108 --n_end -108 dummy_324_oversample.fits dummy_108_cut.fits

Could you check please ?

I put attached the dummy.fits cube I used, the 3 files containing the profiles and the zone of 2 pixels I used but it does not matter which zone has been selected.

Cheers