Changes

Page history
Update home authored by Wei Changfeng's avatar Wei Changfeng
Show whitespace changes
Inline Side-by-side
home.md
View page @ 1437ee57
...@@ -255,13 +255,20 @@ This is due to different normalization items. ...@@ -255,13 +255,20 @@ This is due to different normalization items.
### (i) How to remake: ### (i) How to remake:
In this test we get frequency domain noise-free signals at detector directly. In this test we get frequency domain noise-free signals at detector directly.
What I did: To make my data exactly the same to Hunter's test data, I delete 'adding white Gaussian noise' step, which doesn't show up in Hunter's code.(The method of making test data is written in ‘gen_benchmark_pe.py’) And if I add this step, my data will look so different from Hunter's. What I did: To make my data exactly the same to Hunter's test data, I delete 'adding white Gaussian noise' step, which doesn't show up in Hunter's code.(The method of making test data is written in ‘gen_benchmark_pe.py’) And if I add this step, my data will look so different from Hunter's.
P.S. When Hunter inputs this test data into VItamin, he adds white Gaussian noise additionally, so I shouldn't skip this step.
Fig.1 Hunter's test data (whitened): Fig.1 Hunter's test data (whitened):
![01](uploads/9e822df0dd23fc5a8e40fe2bec2cc5d4/01.png) ![01](uploads/9e822df0dd23fc5a8e40fe2bec2cc5d4/01.png)
Fig.2 my data (whitened): Fig.2 my data (whitened):
![01_remake](uploads/d94383852818a5f3040f20da4f6c5fa2/01_remake.png) ![01_remake](uploads/d94383852818a5f3040f20da4f6c5fa2/01_remake.png)
Fig.3 my data if added white Gaussian noise:
![01_nosiy](uploads/adfcd89f4bd9b2e5bd8e9eea3eaae10f/01_nosiy.png) Since we use the totally same parameters, the signal should also look the same. But they don't, because different version of Bilby call different default PSD file.
After we point to the same file: 'aLIGO_O4_high_asd.txt', Hunter's waveform is like:
Fig.3 Hunter's new test data (whitened):
![01](uploads/8371e5cca305660420f5a65b2acb6ec2/01.png)
Fig.4 my new data (whitened):
![01_remake](uploads/66d45d93f992ca72750079ccf1281b91/01_remake.png)
Now they are the same. This has been verified by ‘the points with the same ordinal have the same value’.
### (ii) What my original data looks like (before whitening and adding Gaussian white nosie): ### (ii) What my original data looks like (before whitening and adding Gaussian white nosie):
Fig.4 The signal's plot in time domain: Fig.4 The signal's plot in time domain:
![bilby_bbh_signal](uploads/e40b70cbfd5df4d665f52adf523f497a/bilby_bbh_signal.png) ![bilby_bbh_signal](uploads/e40b70cbfd5df4d665f52adf523f497a/bilby_bbh_signal.png)
...@@ -269,57 +276,21 @@ I fail to generate a time-domain waveform whose amplitude comes to zero quickly ...@@ -269,57 +276,21 @@ I fail to generate a time-domain waveform whose amplitude comes to zero quickly
In Hunter's test data, the amplitude also doesn't come to zero after merging(as you can see in Fig. 1), and there is a vibration, although it is not obvious. I think whitening step reduces it. In Hunter's test data, the amplitude also doesn't come to zero after merging(as you can see in Fig. 1), and there is a vibration, although it is not obvious. I think whitening step reduces it.
### (iii) Comparison of their performances in VItamin: ### (iii) Comparison of their performances in VItamin:
There are 4 data sets for test data.Except the first one, all the corner plots from my data are extremely similar to Hunter's. Hunter's corner plot is on the top, and mine is below.
*Hunter's corner plot is on the top, and mine is below.
#### (a)
Fig.4 & Fig.5 Fig.4 & Fig.5
![0](uploads/cdac31b163fa6022ac91f1d5e74f405b/0.png) ![vitamin_corner_timeseries-0](uploads/ef20b4b2aec9f767dc2616c879e03a5d/vitamin_corner_timeseries-0.png)
![0_remake_2](uploads/6aa4486b90e3f667e117e856aa531cca/0_remake_2.png) ![vitamin_corner_timeseries-1](uploads/a156a11d0e35d40913e84ceb39e691aa/vitamin_corner_timeseries-1.png)
> mass_1=72.6
> mass_2=39.7
> luminosity_distance=1818
> geocent_time=0.23+1126259642.5
> theta_jn=1.59
> ra=2.26
> dec=0.22
#### (b)
Fig.6 & Fig.7
![1](uploads/3e94af076a7a04f8fe633cc081c998b8/1.png)
![1_remake_2](uploads/a1de775dafd6b1864b1d1752158c4754/1_remake_2.png)
> mass_1=79.5 > mass_1=79.5
> mass_2=59.7 > mass_2=59.7
> luminosity_distance=1327 > luminosity_distance=1945
> geocent_time=0.16+1126259642.5 > t_zero=0.24+1126259642.5
> theta_jn=0.69 > theta_jn=0.44
> ra=2.79 > ra=1.77
> dec=-0.05 > dec=-1.01
#### (c) Their posterior look very similar.
Fig.8 & Fig.9 ### (iv)windowing signal helps...
![2](uploads/d889d5e82185914b6d02aed30a323fc5/2.png)
![2_remake_2](uploads/7d9982185bdf8101d293c4d35c6aec36/2_remake_2.png)
> mass_1=70.3
> mass_2=63.6
> luminosity_distance=1882
> geocent_time=0.16+1126259642.5
> theta_jn=2.54
> ra=1.97
> dec=1.06
#### (d)
Fig.10 & Fig.11
![3](uploads/949727cfc1fc44a842a443f4186c5b40/3.png)
![3_remake_2](uploads/74e88b9fc572edd10fec40294344f124/3_remake_2.png)
> mass_1=67.8
> mass_2=50.8
> luminosity_distance=1493
> geocent_time=0.29+1126259642.5
> theta_jn=1.06
> ra=4.06
> dec=-0.17
### (iv)Does windowing signal helps...
Windowing signal usually is used to make fft. How can I use it if making ifft? Windowing signal usually is used to make fft. How can I use it if making ifft?
... ...
......