Skip to content

Smc merge#212

Open
florianhartig wants to merge 50 commits intomasterfrom
SMC-Merge
Open

Smc merge#212
florianhartig wants to merge 50 commits intomasterfrom
SMC-Merge

Conversation

@florianhartig
Copy link
Owner

No description provided.

florianhartig and others added 30 commits July 13, 2018 16:08
It has occurred that getSampleDistance crashed with option "BH" because of very small values in the inverse covariance matrix (see https://stackoverflow.com/questions/22134398/mahalonobis-distance-in-r-error-system-is-computationally-singular). The argument tol determines which (very small) values should be considered equal to zero.
Several modifications to the particle resampling step:
- generate a sequence of intermediary distributions between prior and posterior
- use weights attributed to particles
- calculate Effective Sample Size (ESS) to determine when particles should be resampled
- dynamically adapt the number of iterations
simplification
introduced general diagnostics option
- Recalculation of previous intermediary distribution for resampled particles
- mutate as a separate function
- extra mutate steps at the end of run
Mutation is only done after resampling
…nction

Tolerance parameter (passed to solve) is now set to the smallest number that can be represented on the machine.
- Automatic selection of scaling factors
- Snooker update
The two output terms represent the contributions of means and covariance to total distance
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants