A Fast Method to

Predict Distributions of

Binary Black Hole Masses

Based on Gaussian Process Regression

A Fast Method to Predict Distributions of Binary Black Hole Masses Based of Gaussian Process Regression

Introduction

Case Study Data

Quantify a Distribution: Gaussian Mixture Model

Gaussian Mixture Model with 2 Components

Z=0.0001

Gaussian Mixture Model with 3 Components

Z=0.0001

Gaussian Mixture Model with 2 Components

Z=0.0003

Gaussian Mixture Model with 2 Components

Z=0.001

Gaussian Mixture Model with 2 Components

Z=0.004

Gaussian Mixture Model with 2 Components

Z=0.01

Gaussian Mixture Model with 2 Components

Z=0.02

Gaussian Mixture Model with Components

Z=0.03

<

>

Interpolate Parameters: Gaussian Process

GP with linear scale

GP with linear scale

GP with linear scale

GP with linear scale

GP with linear scale

GP with linear scale

<

>

GP with semi-log scale

GP with semi-log scale

GP with semi-log scale

GP with semi-log scale

GP with semi-log scale

GP with semi-log scale

<

>

Reconstruct Distributions

Construct the predicted distribution based on estimated parameters using four methods

Construct the predicted distribution based on estimated parameters using four methods

Construct the predicted distribution based on estimated parameters using four methods

Construct the predicted distribution based on estimated parameters using four methods

Construct the predicted distribution based on estimated parameters using four methods

<

>

Model Evaluation

Future Prospects