@@ -57,7 +57,8 @@ For VItamin, it makes more sense to calculate the JSD between BBH and BH encount
100 BBHs with different random noise were input to VItamin. We calculated JSD for each two t0 samples and got 50 JSDs of t0.
## 3.2 JSD of t0 calculation: BBH vs BH encounter
In the Signal section, Only one kind of BH encounter waveform is showed. Now we decide to expand the test to all proper BH encounters that we have. Therefore, BH encounters with mass ratio of 1, 4, 8, 16 were used except the one with mass ratio of 2, which has two peaks. Naturally, the injection parameters like sky locations should be fixed. So we only adjusted distance, to make the amplitude of each kind of BH encounter can be similar with the BBH in the Signal section.
In the Signals section, Only one kind of BH encounter waveform is showed. Now we decide to expand the test to all proper BH encounters that we have. Therefore, BH encounters with mass ratio of 1, 4, 8, 16 were used except the one with mass ratio of 2, which has two peaks.
Naturally, the injection parameters like sky locations should be fixed. So we only adjusted distance, to make the amplitude of each kind of BH encounter can be similar with the BBH in the Signals section.
> m1: 5000 MPc
> m4: 2500 MPc
> m8: 1500 MPc
...
...
@@ -68,6 +69,7 @@ Fig 3.2.1 t0 JSD distribution
We also injected BBH using BH encounter peak value given by VItamin. From four BH encounters(mass ratio = 1, 4, 8, 16), we got four BBHs. Then these BBHs' VItamin posteriors were calculated JSDs with corresponding BH encounters.
JSD of t0 are 0.1732, 0.2894, 0.0110, and 0.0983, for mass ratio of 1, 4, 8 and 16 respectively. They're represented as blue lines in the Fig 3.2.1.
In my opinion, these four JSD values don't have an obvious distance from the BBH JSD distribution. Remember in the previous test, a big difference between BBH and BH encounter posterior is BH encounter has double peaks on t0. However, new injected BBH (using BH encounter peak values) also have double peaks on t0 expect m4. This makes me confusing.
Therefore, I think it may work if we perform more this kind of tests. We have 100 random noise, but only used one for each BH encounter waveform already.
