Dataset for LearningMatch: Siamese Neural Network Learns the Match Manifold

A Siamese neural network (called LearningMatch) has learned the mapping between the parameters, specifically λ0 (which is proportional to the chirp mass), η (symmetric mass ratio), and equal aligned spin (χ1 = χ2), of two gravitational-wave templates and the match. This respoitory contains the training data required to train the LearningMatch model presented in the paper. 1000000LambdaEtaAlignedSpinTrainingDataset+1500000DiffusedLambdaEtaAlignedSpinTrainingDataset.csv - This is the training dataset that was used to train the LearningMatch model presenetd in the paper. 100000 datapoints were created uisng the normal method and 150000 datapoints were created using the diffused dataset (see paper for more information). 100000LambdaEtaAlignedSpinValidationDataset+150000DiffusedLambdaEtaAlignedSpinValidationDataset.csv - This is the validation dataset that was used to train the LearningMatch model presenetd in the paper. 100000 datapoints were created uisng the normal method and 150000 datapoints were created using the diffused dataset (see paper for more information). Figure 8 & 9 showcased the accuracy of the LearningMatch model and this was the dataset that was used to test the model: 100000LambdaEtaAlignedSpinTestDataset+150000DiffusedLambdaEtaAlignedSpinTestDataset.csv - This is the test dataset that was used to test the LearningMatch model presenetd in the paper. 100000 datapoints were created uisng the normal method and 150000 datapoints were created using the diffused dataset (see paper for more information). Figure 3 showcased that large datasets are required for LearningMatch to learn accurately, and these are the datasets used to create this figure: 10000LambdaEtaAlignedSpinTrainingDataset+15000DiffusedLambdaEtaAlignedSpinTrainingDataset.csv - This is the training dataset that was used to train the LearningMatch model presenetd in the paper. 10000 datapoints were created uisng the normal method and 15000 datapoints were created using the diffused dataset. 1000LambdaEtaAlignedSpinValidationDataset+1500DiffusedLambdaEtaAlignedSpinValidationDataset.csv - This is the validation dataset that was used to train the LearningMatch model presenetd in the paper. 1000 datapoints were created uisng the normal method and 1500 datapoints were created using the diffused dataset. 10000LambdaEtaAlignedSpinValidationDataset+15000DiffusedLambdaEtaAlignedSpinValidationDataset.csv - This is the validation dataset that was used to train the LearningMatch model presenetd in the paper. 10000 datapoints were created uisng the normal method and 15000 datapoints were created using the diffused dataset. 100000LambdaEtaAlignedSpinTrainingDataset+150000DiffusedLambdaEtaAlignedSpinTrainingDataset.csv - This is the training dataset that was used to train the LearningMatch model presenetd in the paper. 100000 datapoints were created uisng the normal method and 150000 datapoints were created using the diffused dataset.孪生神经网络（Siamese neural network）名为LearningMatch，已学习到引力波模板（gravitational-wave templates）的参数与匹配度之间的映射关系，这些参数具体包括与啁啾质量（chirp mass）成正比的λ₀、对称质量比（symmetric mass ratio）η，以及共线等自旋（equal aligned spin，χ₁=χ₂）。本仓库包含了训练该论文中提出的LearningMatch模型所需的训练数据。 1000000LambdaEtaAlignedSpinTrainingDataset+1500000DiffusedLambdaEtaAlignedSpinTrainingDataset.csv：该数据集为论文中训练LearningMatch模型所用的训练集，其中1000000条数据通过常规方法生成，1500000条数据通过扩散数据集（diffused dataset）方法生成（详细信息参见论文）。 100000LambdaEtaAlignedSpinValidationDataset+150000DiffusedLambdaEtaAlignedSpinValidationDataset.csv：该数据集为论文中训练LearningMatch模型所用的验证集，其中100000条数据通过常规方法生成，150000条数据通过扩散数据集方法生成（详细信息参见论文）。 图8与图9展示了LearningMatch模型的精度，以下为测试该模型所用的数据集： 100000LambdaEtaAlignedSpinTestDataset+150000DiffusedLambdaEtaAlignedSpinTestDataset.csv：该数据集为论文中测试LearningMatch模型所用的测试集，其中100000条数据通过常规方法生成，150000条数据通过扩散数据集方法生成（详细信息参见论文）。 图3表明LearningMatch要实现精准学习需要大规模数据集，以下为绘制该图所用的数据集： 10000LambdaEtaAlignedSpinTrainingDataset+15000DiffusedLambdaEtaAlignedSpinTrainingDataset.csv：该数据集为论文中训练LearningMatch模型所用的训练集，其中10000条数据通过常规方法生成，15000条数据通过扩散数据集方法生成。 1000LambdaEtaAlignedSpinValidationDataset+1500DiffusedLambdaEtaAlignedSpinValidationDataset.csv：该数据集为论文中训练LearningMatch模型所用的验证集，其中1000条数据通过常规方法生成，1500条数据通过扩散数据集方法生成。 10000LambdaEtaAlignedSpinValidationDataset+15000DiffusedLambdaEtaAlignedSpinValidationDataset.csv：该数据集为论文中训练LearningMatch模型所用的验证集，其中10000条数据通过常规方法生成，15000条数据通过扩散数据集方法生成。 100000LambdaEtaAlignedSpinTrainingDataset+150000DiffusedLambdaEtaAlignedSpinTrainingDataset.csv：该数据集为论文中训练LearningMatch模型所用的训练集，其中100000条数据通过常规方法生成，150000条数据通过扩散数据集方法生成。