Used SciPyWrapper for AreWeThereYet notebook#1
Conversation
Changed the notebook to route both IID and LD sampling through SciPyWrapper. The notebook now declares the wait time marginals once and lets SciPyWrapper handle the unit square to distribution transform instead of manually calling each marginal's ppf on Sobol coordinates. Had to use a PPF only adapter for W2 because the taxi wait time has a zero inflated distribution. There is a chance the wait is exactly zero, and otherwise the wait follows an exponential distribution. For this notebook, we only need the PPF to turn uniform samples into taxi wait times. The PDF method is not needed and caused problems because W2 is not a regular continuous distribution. Kept make_zie for W2 instead of QMCPy ZeroInflatedExpUniform because the notebook model needs two separate independent waiting times: W1 for the train wait and W2 for the taxi wait. make_zie gives only the taxi wait distribution, while ZeroInflatedExpUniform is already a full 2D joint true measure. That joint model does not match this notebook's setup as directly.
|
@copilot resolve the merge conflicts in this pull request A student submitted a pull request to MATH565Fall2025 that changes a notebook. The PR cannot be merged because the Please resolve this locally and update the student's PR branch. Requirements:
|
|
@copilot resolve the merge conflicts in this pull request |
Here is the full proposed resolution — nothing has been committed or pushed yet. 1. Conflicting commits before resolution
The branch's earlier merge ( 2. Resolved
|
Changed the notebook to route both IID and LD sampling through SciPyWrapper. The notebook now declares the wait time marginals once and lets SciPyWrapper handle the unit square to distribution transform instead of manually calling each marginal's ppf on Sobol coordinates.
Had to use a PPF only adapter for W2 because the taxi wait time has a zero inflated distribution. There is a chance the wait is exactly zero, and otherwise the wait follows an exponential distribution. For this notebook, we only need the PPF to turn uniform samples into taxi wait times. The PDF method is not needed and caused problems because W2 is not a regular continuous distribution.
Kept make_zie for W2 instead of QMCPy ZeroInflatedExpUniform because the notebook model needs two separate independent waiting times: W1 for the train wait and W2 for the taxi wait. make_zie gives only the taxi wait distribution, while ZeroInflatedExpUniform is already a full 2D joint true measure. That joint model does not match this notebook's setup as directly.