Skip to content

Plots for prior and posterior samples #21

@V-Rang

Description

@V-Rang

Attaching plots for samples taken from prior and posterior for two examples - example/poisson_dirichlet_example.py and example/sfsi_toy_gaussian.py

commit: 6a30e9d

Poisson Dirichlet example:

The base case is
noise_variance = 1e-4
prior_param = {"gamma": 0.03, "delta": 0.3}

Scaling factor, s is varied as 2,4,6,78,10 where modifications are made as:
noise_variance = (1/s^2) *base noise_variance
gamma = s * base gamma
delta = s * base delta
3 samples are generated for each value of scaling factor.

true parameter:

Alt text

s = 2

posterior samples

Alt text Alt text Alt text

s = 4

posterior samples

Alt text Alt text Alt text

s = 6

posterior samples

Alt text Alt text Alt text

s = 8

posterior samples

Alt text Alt text Alt text

s = 10

posterior samples

Alt text Alt text Alt text

qpact example:

The base case is
noise_variance = 1e-6
prior_param = {"gamma": 0.04, "delta": 0.8}

Scaling factor, s is varied as 2,4,6,78,10 where modifications are made as:
noise_variance = (1/s^2) *base noise_variance
gamma = s * base gamma
delta = s * base delta
3 samples are generated for each value of scaling factor.

true parameter:

Alt text

s = 2

posterior samples

Alt text Alt text Alt text

s = 4

posterior samples

Alt text Alt text Alt text

s = 6

posterior samples

Alt text Alt text Alt text

s = 8

posterior samples

Alt text Alt text Alt text

s = 10

posterior samples

Alt text Alt text Alt text

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions