Add Melanie's CES code - EKI, GPEmulator, MCMC, and Truth #14
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This code was developed for application to Cloudy, a microphysics toy model, using Ollie's existing CES code. The idea is to take this code as a starting point for developing a general-purpose CES code that is fully "disentangled" from Cloudy.
bors bot
added a commit
that referenced
this pull request
Nov 18, 2022
196: Delete old Manifest affecting local docs build r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Related to #195 <!--- list of proposed tasks in the PR, move to "Content" on completion - Proposed task --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Remove Manifest file <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: odunbar <47412152+odunbar@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
Jan 20, 2023
202: remove redefinition warnings r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Closes #201 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - adds import for `GaussianProcesses` predict method - removes manifests for examples - updates `pyimport` to `pyimport_conda` call <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Jan 23, 2023
199: Modify how to load prior distributions from edmf data r=szy21 a=szy21 <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> This PR modifies how prior distributions are loaded to be consistent with recent changes in CalibrateEDMF. It also updates the EDMF data in `ent-det-calibration.zip` and `ent-det-tked-tkee-stab-calibration`. ## Example output `exp_name = ent-det-calibration`  `exp_name = ent-det-tked-tkee-stab-calibration`  <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: szy21 <11598433+szy21@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
Jan 30, 2023
205: Orad/remove gp r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Closes #204 and updates the Lorenz example to be compatible with the version in EKP v0.14 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Updated the example interface - Split `Lorenz_example.jl` into model files (`GModel.jl`, `GModel_common.jl`) and split CES into `calibrate.jl` calibration file, and `emulate_sample.jl`. <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Apr 5, 2023
208: fix testing bugs r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> No direct issue. Bugs found during other testing, - scikit learn leads to conda issues (i.e. trying to install a non-existent package (`libstdcxx-ngnothing`) presumably due to some typo internally, or "broken SAT solver" message due to pycosat) - `MarkovChainMonteCarlo`s `sample()` export leads to test breaking in some julia versions - GP test fails due to random seed changes ## Content <!--- specific tasks that are currently complete - Solution implemented --> - removes `@sk_learn` macro, which ignored the installed Conda/PyCall and created its own (buggy) installation of conda. Instead, always uses `PyCall` - adds `MCMC.sample` qualifier in tests - loosens GP tolerances for more robust testing under different random seeds <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Apr 6, 2023
208: fix testing bugs r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> No direct issue. Bugs found during other testing, - scikit learn leads to conda issues (i.e. trying to install a non-existent package (`libstdcxx-ngnothing`) presumably due to some typo internally, or "broken SAT solver" message due to pycosat) - `MarkovChainMonteCarlo`s `sample()` export leads to test breaking in some julia versions - GP test fails due to random seed changes - buildkite fails for scikit-learn ## Content <!--- specific tasks that are currently complete - Solution implemented --> - removes `@sk_learn` macro, which ignored the installed Conda/PyCall and created its own (buggy) installation of conda. Instead, always uses `PyCall` - adds `MCMC.sample` qualifier in tests - loosens GP tolerances for more robust testing under different random seeds - bug due to changed conda / python paths in buildkite between instantiate project and running plot_GP example. Unifying these paths solves the issue <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Apr 6, 2023
209: condensed compat changes r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> - [x] I have read and checked the items on the review checklist. Co-authored-by: Oliver Dunbar <47412152+odunbar@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
Apr 25, 2023
211: Update LICENSE r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Closes #210 Co-authored-by: Oliver Dunbar <47412152+odunbar@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
May 1, 2023
212: Rename LICENSE to NOTICE r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## To-do <!--- list of proposed tasks in the PR, move to "Content" on completion - Proposed task --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: Oliver Dunbar <47412152+odunbar@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
May 1, 2023
213: Create LICENSE r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## To-do <!--- list of proposed tasks in the PR, move to "Content" on completion - Proposed task --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: Oliver Dunbar <47412152+odunbar@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
May 5, 2023
194: [WIP] Random Feature-based CES r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Adds the ability to use (Scalar and vector-valued) RF with uncertainty in place of GP within CES. using `RandomFeatures.jl` Closes #164 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Interfaces with the currently registered RandomFeatures.jl - adds `ScalarRandomFeatureInterface` as a `MachineLearningTool` - adds `VectorRandomFeatureInterface` as a `MachineLearningTool` - new example `examples/Emulator/RandomFeature/optimize_and_plot_RF.jl` - new example `examples/Emulator/RandomFeature/vector_optimize_and_plot_RF.jl` - new example `examples/Lorenz/calibrate.jl` - new example `examples/Lorenz/emulate_sample.jl` The current implementation has Scalar RF replacing (exactly) the GP, whereas Vector RF does no SVD, and therefore learns the output space correlations. The hyperparameter learning is more involved, so to reduce some cost I learn the cholesky factors of an input and output covariance of the feature distribution, currently described by a MatrixVariate Normal distribution. - new example `examples/GCM/emulate_sample_script.jl` though *currently just the emulation!* In this example we have 4 options: (Note in all cases we train on cholesky factors for the input variables) 1. `GPR` trains an `output_dim`-length vector of scalar GPRs, 2. `Scalar RFR SVD` replaces the vector of scalar GPRs, with a vector of scalar RFRs, 3. `Vector RFR SVD Diagonal` assumes a diagonalized output in the vector problem (i.e. still in the setting of a system of Scalar RFs & GPs but only train one object) 4. `Vector RFR SVD nondiagonal` still applies the SVD, but does not assume that the resulting output must be diagonal. It therefore learns cholesky factors of the output 4. `Vector RFR nondiagonal` does not apply SVD, nor assumes the output is diagonal. It learns the cholesky factors of the direct output. ### Emulating an R^2 to R^2 function (150 data points) 1) SVD + Scalar GP (diag in) results <img src="https://user-images.githubusercontent.com/47412152/192404099-be8d1241-2dd4-4263-ba2a-31de94763abb.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/192404100-62d72ccc-2b36-4ba9-ad36-38bcdf4b9f0f.png" width="300"> 2) SVD + Scalar RF (nondiag in) results <img src="https://user-images.githubusercontent.com/47412152/230235711-6bb0557e-8914-4a43-8f91-f5a144659edc.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230235715-54fb7d5e-fa24-4528-a3fc-27ff7e9aceb8.png" width="300"> 3) SVD + vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230229962-c7eefa25-3a57-467c-8ca1-c9ef7b3dbb3e.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230229964-fcfcbe7c-73cd-4837-8db7-5eaa339eaec6.png" width="300"> 4) SVD + vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230124-bb50e4db-8ba7-4570-930e-b6504936a1b5.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230128-2422e85a-1100-4422-a52c-8fd32183b7f0.png" width="300"> 5) vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230230033-618dcfa8-99a7-4462-b31f-e9adf302dc14.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230040-7dad6ce8-b5dc-4057-9693-a71e0a72379c.png" width="300"> 6) vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230167-25377c56-c622-493d-bd0e-b98fc672189a.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230169-23c659eb-4218-4b27-9625-94a7fd44a941.png" width="300"> ### Emulating GCM data R^2 -> R^96, evaluated at a test point #### SVD + Scalar RF results <img src="https://user-images.githubusercontent.com/47412152/219200986-9a5f74e4-5e2a-48cf-8e26-5d66de2e751c.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200996-f5b88c6e-8b51-4df0-acac-187a6a786a78.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201000-afac8c7a-aa9a-4400-93e8-75f560954bba.png" width="150"> #### SVD + Vector RF (restrict to diagonal) results [hparam learnt with 202 features]) <img src="https://user-images.githubusercontent.com/47412152/219200373-7b0e2713-c3db-4891-9012-6852381266b5.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219252222-f75ba2c8-b11a-42eb-aa1c-f2308a775041.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200387-d328a6b6-5f64-4eda-bc7e-3a122054c2a5.png" width="150"> #### SVD + vector RF results (full non-diagonal [hparam learnt with 608 features]) <img src="https://user-images.githubusercontent.com/47412152/220444681-20b3ef41-5347-4406-afd6-ffa8cfc1e1b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444689-92f24b34-6937-4e19-b733-2d1b263ca9f7.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444685-70f88334-3c32-438e-8534-e4ca0a1c24d2.png" width="150"> #### No-SVD, with vector RF results (full non-diagonal) + standardize each data-type by median <img src="https://user-images.githubusercontent.com/47412152/235567696-4c5665b1-33db-4c83-a554-f003e5e015b6.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567703-0fae36f9-e49b-4cfc-b1b4-16a313af51b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567706-1d9513f7-ec37-4fad-b824-a3c62b7759d0.png" width="150"> #### SVD + GP results <img src="https://user-images.githubusercontent.com/47412152/219201341-13acb758-a444-4e05-98c9-ed6975dbd094.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201347-d5c13d9f-3d63-456b-8059-fea0f31346a0.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201351-fa2a4a61-bf75-4292-8549-8f304f05cebe.png" width="150"> ### Full CES test (with "E" emulating an R^2 -> R^12 forward map) 250 data points 1) SVD + Scalar GP (diag in) results 2) SVD + Scalar RF (diag in) results 3) SVD + Scalar RF (nondiag in) results 4) SVD + vector RF (nondiag in, diag out) results 5) SVD + vector RF (nondiag in, nondiag out) results 6) vector RF (nondiag in, diag out) results 7) vector RF (nondiag in, nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/236000320-bbf88ee3-6de7-4e8e-8797-13f48696337a.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364924-c41dd024-e56e-4506-ad19-2fc324d0db61.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364926-f74d9467-7083-4fb9-8c89-b6ea4ab8aad3.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364930-a0424381-ccc5-472c-b85f-e24aec7319f0.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364932-3cb2e6c6-0f5b-4544-9916-cddfcbd3c882.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364927-b05e1bc8-0430-4db8-97dc-29b0fce0f305.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364929-5e78ceb3-07e9-4230-95b1-2e638dab1ce3.png" width="175"> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevent documentation. --> Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
May 5, 2023
194: [WIP] Random Feature-based CES r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Adds the ability to use (Scalar and vector-valued) RF with uncertainty in place of GP within CES. using `RandomFeatures.jl` Closes #164 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Interfaces with the currently registered RandomFeatures.jl - adds `ScalarRandomFeatureInterface` as a `MachineLearningTool` - adds `VectorRandomFeatureInterface` as a `MachineLearningTool` - new example `examples/Emulator/RandomFeature/optimize_and_plot_RF.jl` - new example `examples/Emulator/RandomFeature/vector_optimize_and_plot_RF.jl` - new example `examples/Lorenz/calibrate.jl` - new example `examples/Lorenz/emulate_sample.jl` The current implementation has Scalar RF replacing (exactly) the GP, whereas Vector RF does no SVD, and therefore learns the output space correlations. The hyperparameter learning is more involved, so to reduce some cost I learn the cholesky factors of an input and output covariance of the feature distribution, currently described by a MatrixVariate Normal distribution. - new example `examples/GCM/emulate_sample_script.jl` though *currently just the emulation!* In this example we have 4 options: (Note in all cases we train on cholesky factors for the input variables) 1. `GPR` trains an `output_dim`-length vector of scalar GPRs, 2. `Scalar RFR SVD` replaces the vector of scalar GPRs, with a vector of scalar RFRs, 3. `Vector RFR SVD Diagonal` assumes a diagonalized output in the vector problem (i.e. still in the setting of a system of Scalar RFs & GPs but only train one object) 4. `Vector RFR SVD nondiagonal` still applies the SVD, but does not assume that the resulting output must be diagonal. It therefore learns cholesky factors of the output 4. `Vector RFR nondiagonal` does not apply SVD, nor assumes the output is diagonal. It learns the cholesky factors of the direct output. ### Emulating an R^2 to R^2 function (150 data points) 1) SVD + Scalar GP (diag in) results <img src="https://user-images.githubusercontent.com/47412152/192404099-be8d1241-2dd4-4263-ba2a-31de94763abb.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/192404100-62d72ccc-2b36-4ba9-ad36-38bcdf4b9f0f.png" width="300"> 2) SVD + Scalar RF (nondiag in) results <img src="https://user-images.githubusercontent.com/47412152/230235711-6bb0557e-8914-4a43-8f91-f5a144659edc.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230235715-54fb7d5e-fa24-4528-a3fc-27ff7e9aceb8.png" width="300"> 3) SVD + vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230229962-c7eefa25-3a57-467c-8ca1-c9ef7b3dbb3e.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230229964-fcfcbe7c-73cd-4837-8db7-5eaa339eaec6.png" width="300"> 4) SVD + vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230124-bb50e4db-8ba7-4570-930e-b6504936a1b5.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230128-2422e85a-1100-4422-a52c-8fd32183b7f0.png" width="300"> 5) vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230230033-618dcfa8-99a7-4462-b31f-e9adf302dc14.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230040-7dad6ce8-b5dc-4057-9693-a71e0a72379c.png" width="300"> 6) vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230167-25377c56-c622-493d-bd0e-b98fc672189a.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230169-23c659eb-4218-4b27-9625-94a7fd44a941.png" width="300"> ### Emulating GCM data R^2 -> R^96, evaluated at a test point #### SVD + Scalar RF results <img src="https://user-images.githubusercontent.com/47412152/219200986-9a5f74e4-5e2a-48cf-8e26-5d66de2e751c.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200996-f5b88c6e-8b51-4df0-acac-187a6a786a78.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201000-afac8c7a-aa9a-4400-93e8-75f560954bba.png" width="150"> #### SVD + Vector RF (restrict to diagonal) results [hparam learnt with 202 features]) <img src="https://user-images.githubusercontent.com/47412152/219200373-7b0e2713-c3db-4891-9012-6852381266b5.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219252222-f75ba2c8-b11a-42eb-aa1c-f2308a775041.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200387-d328a6b6-5f64-4eda-bc7e-3a122054c2a5.png" width="150"> #### SVD + vector RF results (full non-diagonal [hparam learnt with 608 features]) <img src="https://user-images.githubusercontent.com/47412152/220444681-20b3ef41-5347-4406-afd6-ffa8cfc1e1b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444689-92f24b34-6937-4e19-b733-2d1b263ca9f7.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444685-70f88334-3c32-438e-8534-e4ca0a1c24d2.png" width="150"> #### No-SVD, with vector RF results (full non-diagonal) + standardize each data-type by median <img src="https://user-images.githubusercontent.com/47412152/235567696-4c5665b1-33db-4c83-a554-f003e5e015b6.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567703-0fae36f9-e49b-4cfc-b1b4-16a313af51b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567706-1d9513f7-ec37-4fad-b824-a3c62b7759d0.png" width="150"> #### SVD + GP results <img src="https://user-images.githubusercontent.com/47412152/219201341-13acb758-a444-4e05-98c9-ed6975dbd094.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201347-d5c13d9f-3d63-456b-8059-fea0f31346a0.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201351-fa2a4a61-bf75-4292-8549-8f304f05cebe.png" width="150"> ### Full CES test (with "E" emulating an R^2 -> R^12 forward map) 250 data points 1) SVD + Scalar GP (diag in) results 2) SVD + Scalar RF (diag in) results 3) SVD + Scalar RF (nondiag in) results 4) SVD + vector RF (nondiag in, diag out) results 5) SVD + vector RF (nondiag in, nondiag out) results 6) vector RF (nondiag in, diag out) results 7) vector RF (nondiag in, nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/236000320-bbf88ee3-6de7-4e8e-8797-13f48696337a.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364924-c41dd024-e56e-4506-ad19-2fc324d0db61.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364926-f74d9467-7083-4fb9-8c89-b6ea4ab8aad3.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364930-a0424381-ccc5-472c-b85f-e24aec7319f0.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364932-3cb2e6c6-0f5b-4544-9916-cddfcbd3c882.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364927-b05e1bc8-0430-4db8-97dc-29b0fce0f305.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364929-5e78ceb3-07e9-4230-95b1-2e638dab1ce3.png" width="175"> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevent documentation. --> Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
May 5, 2023
194: [WIP] Random Feature-based CES r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Adds the ability to use (Scalar and vector-valued) RF with uncertainty in place of GP within CES. using `RandomFeatures.jl` Closes #164 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Interfaces with the currently registered RandomFeatures.jl - adds `ScalarRandomFeatureInterface` as a `MachineLearningTool` - adds `VectorRandomFeatureInterface` as a `MachineLearningTool` - new example `examples/Emulator/RandomFeature/optimize_and_plot_RF.jl` - new example `examples/Emulator/RandomFeature/vector_optimize_and_plot_RF.jl` - new example `examples/Lorenz/calibrate.jl` - new example `examples/Lorenz/emulate_sample.jl` The current implementation has Scalar RF replacing (exactly) the GP, whereas Vector RF does no SVD, and therefore learns the output space correlations. The hyperparameter learning is more involved, so to reduce some cost I learn the cholesky factors of an input and output covariance of the feature distribution, currently described by a MatrixVariate Normal distribution. - new example `examples/GCM/emulate_sample_script.jl` though *currently just the emulation!* In this example we have 4 options: (Note in all cases we train on cholesky factors for the input variables) 1. `GPR` trains an `output_dim`-length vector of scalar GPRs, 2. `Scalar RFR SVD` replaces the vector of scalar GPRs, with a vector of scalar RFRs, 3. `Vector RFR SVD Diagonal` assumes a diagonalized output in the vector problem (i.e. still in the setting of a system of Scalar RFs & GPs but only train one object) 4. `Vector RFR SVD nondiagonal` still applies the SVD, but does not assume that the resulting output must be diagonal. It therefore learns cholesky factors of the output 4. `Vector RFR nondiagonal` does not apply SVD, nor assumes the output is diagonal. It learns the cholesky factors of the direct output. ### Emulating an R^2 to R^2 function (150 data points) 1) SVD + Scalar GP (diag in) results <img src="https://user-images.githubusercontent.com/47412152/192404099-be8d1241-2dd4-4263-ba2a-31de94763abb.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/192404100-62d72ccc-2b36-4ba9-ad36-38bcdf4b9f0f.png" width="300"> 2) SVD + Scalar RF (nondiag in) results <img src="https://user-images.githubusercontent.com/47412152/230235711-6bb0557e-8914-4a43-8f91-f5a144659edc.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230235715-54fb7d5e-fa24-4528-a3fc-27ff7e9aceb8.png" width="300"> 3) SVD + vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230229962-c7eefa25-3a57-467c-8ca1-c9ef7b3dbb3e.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230229964-fcfcbe7c-73cd-4837-8db7-5eaa339eaec6.png" width="300"> 4) SVD + vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230124-bb50e4db-8ba7-4570-930e-b6504936a1b5.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230128-2422e85a-1100-4422-a52c-8fd32183b7f0.png" width="300"> 5) vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230230033-618dcfa8-99a7-4462-b31f-e9adf302dc14.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230040-7dad6ce8-b5dc-4057-9693-a71e0a72379c.png" width="300"> 6) vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230167-25377c56-c622-493d-bd0e-b98fc672189a.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230169-23c659eb-4218-4b27-9625-94a7fd44a941.png" width="300"> ### Emulating GCM data R^2 -> R^96, evaluated at a test point #### SVD + Scalar RF results <img src="https://user-images.githubusercontent.com/47412152/219200986-9a5f74e4-5e2a-48cf-8e26-5d66de2e751c.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200996-f5b88c6e-8b51-4df0-acac-187a6a786a78.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201000-afac8c7a-aa9a-4400-93e8-75f560954bba.png" width="150"> #### SVD + Vector RF (restrict to diagonal) results [hparam learnt with 202 features]) <img src="https://user-images.githubusercontent.com/47412152/219200373-7b0e2713-c3db-4891-9012-6852381266b5.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219252222-f75ba2c8-b11a-42eb-aa1c-f2308a775041.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200387-d328a6b6-5f64-4eda-bc7e-3a122054c2a5.png" width="150"> #### SVD + vector RF results (full non-diagonal [hparam learnt with 608 features]) <img src="https://user-images.githubusercontent.com/47412152/220444681-20b3ef41-5347-4406-afd6-ffa8cfc1e1b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444689-92f24b34-6937-4e19-b733-2d1b263ca9f7.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444685-70f88334-3c32-438e-8534-e4ca0a1c24d2.png" width="150"> #### No-SVD, with vector RF results (full non-diagonal) + standardize each data-type by median <img src="https://user-images.githubusercontent.com/47412152/235567696-4c5665b1-33db-4c83-a554-f003e5e015b6.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567703-0fae36f9-e49b-4cfc-b1b4-16a313af51b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567706-1d9513f7-ec37-4fad-b824-a3c62b7759d0.png" width="150"> #### SVD + GP results <img src="https://user-images.githubusercontent.com/47412152/219201341-13acb758-a444-4e05-98c9-ed6975dbd094.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201347-d5c13d9f-3d63-456b-8059-fea0f31346a0.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201351-fa2a4a61-bf75-4292-8549-8f304f05cebe.png" width="150"> ### Full CES test (with "E" emulating an R^2 -> R^12 forward map) 250 data points 1) SVD + Scalar GP (diag in) results 2) SVD + Scalar RF (diag in) results 3) SVD + Scalar RF (nondiag in) results 4) SVD + vector RF (nondiag in, diag out) results 5) SVD + vector RF (nondiag in, nondiag out) results 6) vector RF (nondiag in, diag out) results 7) vector RF (nondiag in, nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/236000320-bbf88ee3-6de7-4e8e-8797-13f48696337a.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364924-c41dd024-e56e-4506-ad19-2fc324d0db61.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364926-f74d9467-7083-4fb9-8c89-b6ea4ab8aad3.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364930-a0424381-ccc5-472c-b85f-e24aec7319f0.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364932-3cb2e6c6-0f5b-4544-9916-cddfcbd3c882.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364927-b05e1bc8-0430-4db8-97dc-29b0fce0f305.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364929-5e78ceb3-07e9-4230-95b1-2e638dab1ce3.png" width="175"> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevent documentation. --> Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
May 5, 2023
194: [WIP] Random Feature-based CES r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Adds the ability to use (Scalar and vector-valued) RF with uncertainty in place of GP within CES. using `RandomFeatures.jl` Closes #164 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Interfaces with the currently registered RandomFeatures.jl - adds `ScalarRandomFeatureInterface` as a `MachineLearningTool` - adds `VectorRandomFeatureInterface` as a `MachineLearningTool` - new example `examples/Emulator/RandomFeature/optimize_and_plot_RF.jl` - new example `examples/Emulator/RandomFeature/vector_optimize_and_plot_RF.jl` - new example `examples/Lorenz/calibrate.jl` - new example `examples/Lorenz/emulate_sample.jl` The current implementation has Scalar RF replacing (exactly) the GP, whereas Vector RF does no SVD, and therefore learns the output space correlations. The hyperparameter learning is more involved, so to reduce some cost I learn the cholesky factors of an input and output covariance of the feature distribution, currently described by a MatrixVariate Normal distribution. - new example `examples/GCM/emulate_sample_script.jl` though *currently just the emulation!* In this example we have 4 options: (Note in all cases we train on cholesky factors for the input variables) 1. `GPR` trains an `output_dim`-length vector of scalar GPRs, 2. `Scalar RFR SVD` replaces the vector of scalar GPRs, with a vector of scalar RFRs, 3. `Vector RFR SVD Diagonal` assumes a diagonalized output in the vector problem (i.e. still in the setting of a system of Scalar RFs & GPs but only train one object) 4. `Vector RFR SVD nondiagonal` still applies the SVD, but does not assume that the resulting output must be diagonal. It therefore learns cholesky factors of the output 4. `Vector RFR nondiagonal` does not apply SVD, nor assumes the output is diagonal. It learns the cholesky factors of the direct output. ### Emulating an R^2 to R^2 function (150 data points) 1) SVD + Scalar GP (diag in) results <img src="https://user-images.githubusercontent.com/47412152/192404099-be8d1241-2dd4-4263-ba2a-31de94763abb.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/192404100-62d72ccc-2b36-4ba9-ad36-38bcdf4b9f0f.png" width="300"> 2) SVD + Scalar RF (nondiag in) results <img src="https://user-images.githubusercontent.com/47412152/230235711-6bb0557e-8914-4a43-8f91-f5a144659edc.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230235715-54fb7d5e-fa24-4528-a3fc-27ff7e9aceb8.png" width="300"> 3) SVD + vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230229962-c7eefa25-3a57-467c-8ca1-c9ef7b3dbb3e.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230229964-fcfcbe7c-73cd-4837-8db7-5eaa339eaec6.png" width="300"> 4) SVD + vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230124-bb50e4db-8ba7-4570-930e-b6504936a1b5.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230128-2422e85a-1100-4422-a52c-8fd32183b7f0.png" width="300"> 5) vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230230033-618dcfa8-99a7-4462-b31f-e9adf302dc14.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230040-7dad6ce8-b5dc-4057-9693-a71e0a72379c.png" width="300"> 6) vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230167-25377c56-c622-493d-bd0e-b98fc672189a.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230169-23c659eb-4218-4b27-9625-94a7fd44a941.png" width="300"> ### Emulating GCM data R^2 -> R^96, evaluated at a test point #### SVD + Scalar RF results <img src="https://user-images.githubusercontent.com/47412152/219200986-9a5f74e4-5e2a-48cf-8e26-5d66de2e751c.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200996-f5b88c6e-8b51-4df0-acac-187a6a786a78.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201000-afac8c7a-aa9a-4400-93e8-75f560954bba.png" width="150"> #### SVD + Vector RF (restrict to diagonal) results [hparam learnt with 202 features]) <img src="https://user-images.githubusercontent.com/47412152/219200373-7b0e2713-c3db-4891-9012-6852381266b5.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219252222-f75ba2c8-b11a-42eb-aa1c-f2308a775041.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200387-d328a6b6-5f64-4eda-bc7e-3a122054c2a5.png" width="150"> #### SVD + vector RF results (full non-diagonal [hparam learnt with 608 features]) <img src="https://user-images.githubusercontent.com/47412152/220444681-20b3ef41-5347-4406-afd6-ffa8cfc1e1b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444689-92f24b34-6937-4e19-b733-2d1b263ca9f7.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444685-70f88334-3c32-438e-8534-e4ca0a1c24d2.png" width="150"> #### No-SVD, with vector RF results (full non-diagonal) + standardize each data-type by median <img src="https://user-images.githubusercontent.com/47412152/235567696-4c5665b1-33db-4c83-a554-f003e5e015b6.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567703-0fae36f9-e49b-4cfc-b1b4-16a313af51b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567706-1d9513f7-ec37-4fad-b824-a3c62b7759d0.png" width="150"> #### SVD + GP results <img src="https://user-images.githubusercontent.com/47412152/219201341-13acb758-a444-4e05-98c9-ed6975dbd094.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201347-d5c13d9f-3d63-456b-8059-fea0f31346a0.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201351-fa2a4a61-bf75-4292-8549-8f304f05cebe.png" width="150"> ### Full CES test (with "E" emulating an R^2 -> R^12 forward map) 250 data points 1) SVD + Scalar GP (diag in) results 2) SVD + Scalar RF (diag in) results 3) SVD + Scalar RF (nondiag in) results 4) SVD + vector RF (nondiag in, diag out) results 5) SVD + vector RF (nondiag in, nondiag out) results 6) vector RF (nondiag in, diag out) results 7) vector RF (nondiag in, nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/236000320-bbf88ee3-6de7-4e8e-8797-13f48696337a.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364924-c41dd024-e56e-4506-ad19-2fc324d0db61.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364926-f74d9467-7083-4fb9-8c89-b6ea4ab8aad3.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364930-a0424381-ccc5-472c-b85f-e24aec7319f0.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364932-3cb2e6c6-0f5b-4544-9916-cddfcbd3c882.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364927-b05e1bc8-0430-4db8-97dc-29b0fce0f305.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364929-5e78ceb3-07e9-4230-95b1-2e638dab1ce3.png" width="175"> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevent documentation. --> Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
May 5, 2023
194: Random Feature-based CES r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Adds the ability to use (Scalar and vector-valued) RF with uncertainty in place of GP within CES. using `RandomFeatures.jl` Closes #164 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Interfaces with the currently registered RandomFeatures.jl - adds `ScalarRandomFeatureInterface` as a `MachineLearningTool` - adds `VectorRandomFeatureInterface` as a `MachineLearningTool` - new example `examples/Emulator/RandomFeature/optimize_and_plot_RF.jl` - new example `examples/Emulator/RandomFeature/vector_optimize_and_plot_RF.jl` - new example `examples/Lorenz/calibrate.jl` - new example `examples/Lorenz/emulate_sample.jl` The current implementation has Scalar RF replacing (exactly) the GP, whereas Vector RF does no SVD, and therefore learns the output space correlations. The hyperparameter learning is more involved, so to reduce some cost I learn the cholesky factors of an input and output covariance of the feature distribution, currently described by a MatrixVariate Normal distribution. - new example `examples/GCM/emulate_sample_script.jl` though *currently just the emulation!* In this example we have 4 options: (Note in all cases we train on cholesky factors for the input variables) 1. `GPR` trains an `output_dim`-length vector of scalar GPRs, 2. `Scalar RFR SVD` replaces the vector of scalar GPRs, with a vector of scalar RFRs, 3. `Vector RFR SVD Diagonal` assumes a diagonalized output in the vector problem (i.e. still in the setting of a system of Scalar RFs & GPs but only train one object) 4. `Vector RFR SVD nondiagonal` still applies the SVD, but does not assume that the resulting output must be diagonal. It therefore learns cholesky factors of the output 4. `Vector RFR nondiagonal` does not apply SVD, nor assumes the output is diagonal. It learns the cholesky factors of the direct output. ### Emulating an R^2 to R^2 function (150 data points) 1) SVD + Scalar GP (diag in) results <img src="https://user-images.githubusercontent.com/47412152/192404099-be8d1241-2dd4-4263-ba2a-31de94763abb.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/192404100-62d72ccc-2b36-4ba9-ad36-38bcdf4b9f0f.png" width="300"> 2) SVD + Scalar RF (nondiag in) results <img src="https://user-images.githubusercontent.com/47412152/230235711-6bb0557e-8914-4a43-8f91-f5a144659edc.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230235715-54fb7d5e-fa24-4528-a3fc-27ff7e9aceb8.png" width="300"> 3) SVD + vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230229962-c7eefa25-3a57-467c-8ca1-c9ef7b3dbb3e.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230229964-fcfcbe7c-73cd-4837-8db7-5eaa339eaec6.png" width="300"> 4) SVD + vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230124-bb50e4db-8ba7-4570-930e-b6504936a1b5.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230128-2422e85a-1100-4422-a52c-8fd32183b7f0.png" width="300"> 5) vector RF (diag out) results <img src="https://user-images.githubusercontent.com/47412152/230230033-618dcfa8-99a7-4462-b31f-e9adf302dc14.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230040-7dad6ce8-b5dc-4057-9693-a71e0a72379c.png" width="300"> 6) vector RF (nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/230230167-25377c56-c622-493d-bd0e-b98fc672189a.png" width="300"> <img src="https://user-images.githubusercontent.com/47412152/230230169-23c659eb-4218-4b27-9625-94a7fd44a941.png" width="300"> ### Emulating GCM data R^2 -> R^96, evaluated at a test point #### SVD + Scalar RF results <img src="https://user-images.githubusercontent.com/47412152/219200986-9a5f74e4-5e2a-48cf-8e26-5d66de2e751c.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200996-f5b88c6e-8b51-4df0-acac-187a6a786a78.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201000-afac8c7a-aa9a-4400-93e8-75f560954bba.png" width="150"> #### SVD + Vector RF (restrict to diagonal) results [hparam learnt with 202 features]) <img src="https://user-images.githubusercontent.com/47412152/219200373-7b0e2713-c3db-4891-9012-6852381266b5.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219252222-f75ba2c8-b11a-42eb-aa1c-f2308a775041.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219200387-d328a6b6-5f64-4eda-bc7e-3a122054c2a5.png" width="150"> #### SVD + vector RF results (full non-diagonal [hparam learnt with 608 features]) <img src="https://user-images.githubusercontent.com/47412152/220444681-20b3ef41-5347-4406-afd6-ffa8cfc1e1b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444689-92f24b34-6937-4e19-b733-2d1b263ca9f7.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/220444685-70f88334-3c32-438e-8534-e4ca0a1c24d2.png" width="150"> #### No-SVD, with vector RF results (full non-diagonal) + standardize each data-type by median <img src="https://user-images.githubusercontent.com/47412152/235567696-4c5665b1-33db-4c83-a554-f003e5e015b6.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567703-0fae36f9-e49b-4cfc-b1b4-16a313af51b8.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/235567706-1d9513f7-ec37-4fad-b824-a3c62b7759d0.png" width="150"> #### SVD + GP results <img src="https://user-images.githubusercontent.com/47412152/219201341-13acb758-a444-4e05-98c9-ed6975dbd094.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201347-d5c13d9f-3d63-456b-8059-fea0f31346a0.png" width="150"> <img src="https://user-images.githubusercontent.com/47412152/219201351-fa2a4a61-bf75-4292-8549-8f304f05cebe.png" width="150"> ### Full CES test (with "E" emulating an R^2 -> R^12 forward map) 250 data points 1) SVD + Scalar GP (diag in) results 2) SVD + Scalar RF (diag in) results 3) SVD + Scalar RF (nondiag in) results 4) SVD + vector RF (nondiag in, diag out) results 5) SVD + vector RF (nondiag in, nondiag out) results 6) vector RF (nondiag in, diag out) results 7) vector RF (nondiag in, nondiag out) results <img src="https://user-images.githubusercontent.com/47412152/236000320-bbf88ee3-6de7-4e8e-8797-13f48696337a.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364924-c41dd024-e56e-4506-ad19-2fc324d0db61.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364926-f74d9467-7083-4fb9-8c89-b6ea4ab8aad3.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364930-a0424381-ccc5-472c-b85f-e24aec7319f0.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364932-3cb2e6c6-0f5b-4544-9916-cddfcbd3c882.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364927-b05e1bc8-0430-4db8-97dc-29b0fce0f305.png" width="175"> <img src="https://user-images.githubusercontent.com/47412152/236364929-5e78ceb3-07e9-4230-95b1-2e638dab1ce3.png" width="175"> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevent documentation. --> Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Jun 13, 2023
218: Compatibility with new EKP, version and updated Lorenz example r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> - `get_logpdf` -> `logpdf` so that MCMC tests pass - updated the EKP Project toml. Also add compats for ProgressBars, StableRNGs - added option to set `"scheduler" => XYZ` in `optimizer_options` for RF emulators - removed Lorenz bugs in observational noise - added a scaling tuner for easily expanding/narrowing priors in RF (currently one for input and one for output space) - added some diagnostic information to compare the prior support and optimal parameters to aid with the tuning. <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Jun 13, 2023
218: Compatibility with new EKP, version and updated Lorenz example r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> - `get_logpdf` -> `logpdf` so that MCMC tests pass - updated the EKP Project toml. Also add compats for ProgressBars, StableRNGs - added option to set `"scheduler" => XYZ` in `optimizer_options` for RF emulators - removed Lorenz bugs in observational noise - added a scaling tuner for easily expanding/narrowing priors in RF (currently one for input and one for output space) - added some diagnostic information to compare the prior support and optimal parameters to aid with the tuning. <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Jul 12, 2023
223: Extend `AdvancedMH.transition` to CES types only r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Resolves #221 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - extends `AdvancedMH.transition` to CES types only. ### Also - Pinning matplotlib version in buildkite to prevent buildkite error. <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Jul 12, 2023
224: Update version to 0.2.1 r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## To-do <!--- list of proposed tasks in the PR, move to "Content" on completion - Proposed task --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: Oliver Dunbar <47412152+odunbar@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
Jul 18, 2023
226: Update Project.toml to 0.3.0 .. overdue for a while r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## To-do <!--- list of proposed tasks in the PR, move to "Content" on completion - Proposed task --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: Oliver Dunbar <47412152+odunbar@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
Jul 21, 2023
227: Low rank normalization r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Closes #225 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Based on the idea that the covariance defines geometrically a linear transformation from white noise to the data noise, ensemble covariance ``C`` can be viewed as ``C = R S S R^{-1}`` where ``R`` is a rotation, and ``S`` is the sqrt of the singular values of ``C``. The transformation from white data to the actual data samples is ``C^{1/2} = RS`` and it's "whitening" inverse is ``C^{-1/2} = S^{-1}R^{-1} = S^{-1}R^T`` as ``R`` is a rotation. For rank deficient ``C``, we take the normalization to be instead ``Sinv R^T`` where ``Sinv`` is Diagonal with the first ``rank(C)`` entries equal to ``S^{-1}``, and zero otherwise. - Added tests to check the covariances become close to the identity in full or low-dim subspace after normalization - Changed an input-dim consistency check to account for new dimension change <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Oct 6, 2023
220: [WIP] Refactor and extend RF interface for more flexible kernels. r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose Addresses points in #215 In particular: - Closes #228 - Closes #229 - Closes #230 <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## To-do <!--- list of proposed tasks in the PR, move to "Content" on completion - Proposed task --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Adds MCMC stage into `examples/GCM` and improves running interface - Adds Shrinkage estimation into cross-validation for RF hyperparameter optimization. - Adds Kernel interface for RF, where one specifies either a `Separable` or `Nonseparable` Kernel followed by the structure of the covariance to be used to sample the features. - `Nonseparable(cov)`: define one covariance structure on the `p * d`-dim space, - `Separable(in_cov,out_cov)`: define one `d`-dim covariance for inputs and one `p`-dim covariance for outputs - Choose covariance structures for `cov`,`in_cov`,`out_cov` from - `OneDimFactor()`: Must be selected for 1D covariances - `DiagonalFactor(eps)`: Diagonal covariance plus `eps*I` - `CholeskyFactor(eps)`: Cholesky representation `L*L^T` plus eps*I` - `LowRankFactor(rank, eps)`: Symmetric factorization `W*W^T` where `W = 1+UDU'` D diag, U rectangular, plus `eps*I` - `HierarchicalLowRankFactor(rank, eps)`: Symmetric factorization `W*W^T` where `W = 1+UXU'` U rectangular, plus `eps*I` X cholesky factored - Rebalancing the loss function seems to get far more robust results for vector RF <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Oct 11, 2023
232: RF emulator documentation r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Closes #233 ## TODO - sort API docs for RF and ensure docstrings all present ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Documentation can be found at [this page](https://clima.github.io/CalibrateEmulateSample.jl/previews/PR232/) <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Oct 12, 2023
235: added RF to buildkite r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> Closes #234 ## Content <!--- specific tasks that are currently complete - Solution implemented --> - adds Emulator (scalar_optimize_and_plot_rf.jl & vector_optimize_and_plot_rf.jl) into buildkite. <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Oct 13, 2023
236: Update Project.toml r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## To-do <!--- list of proposed tasks in the PR, move to "Content" on completion - Proposed task --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: Oliver Dunbar <47412152+odunbar@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
Oct 18, 2023
237: bugfix r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ProgressBars have patched the "Key 9 not found" bug, so removing our local patch <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [x] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Oct 24, 2023
238: add ishigami example & compatible with new EKP r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> - Adds a new example that explores off the training points. Also apparently this is a challenging emulation problem as the GP emulator training fails with a default kernel and internal training methods. - Example also has an analytic form of the "sensitivity" to it's parameters, only by learning the right function on the domain can one recover the right sensitivities. ## Content <!--- specific tasks that are currently complete - Solution implemented --> - The example in `examples/Emulator/Ishigami` - Also added optimizer options "accelerator" and "n_feature_opt" removing the complex defaulting from before - edits to buildkite to stop it taking an old EKP version <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Nov 1, 2023
246: Update [compat] with new requirements r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## To-do <!--- list of proposed tasks in the PR, move to "Content" on completion - Proposed task --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: Oliver Dunbar <47412152+odunbar@users.noreply.github.com>
bors bot
added a commit
that referenced
this pull request
Nov 3, 2023
249: typo fix r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## To-do <!--- list of proposed tasks in the PR, move to "Content" on completion - Proposed task --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
bors bot
added a commit
that referenced
this pull request
Nov 3, 2023
249: Update to new EKP (remove local posdef_correct) & typo fix r=odunbar a=odunbar <!--- THESE LINES ARE COMMENTED --> ## Purpose <!--- One sentence to describe the purpose of this PR, refer to any linked issues: #14 -- this will link to issue 14 Closes #2 -- this will automatically close issue 2 on PR merge --> ## Content <!--- specific tasks that are currently complete - Solution implemented --> - Scalar weight needed a multiplication to make consistent with theory & vector case. - Removes `posdef_correct` which is causing test failures due to new EKP export of a function with the same name. <!--- Review checklist I have: - followed the codebase contribution guide: https://clima.github.io/ClimateMachine.jl/latest/Contributing/ - followed the style guide: https://clima.github.io/ClimateMachine.jl/latest/DevDocs/CodeStyle/ - followed the documentation policy: https://github.com/CliMA/policies/wiki/Documentation-Policy - checked that this PR does not duplicate an open PR. In the Content, I have included - relevant unit tests, and integration tests, - appropriate docstrings on all functions, structs, and modules, and included relevant documentation. --> ---- - [ ] I have read and checked the items on the review checklist. Co-authored-by: odunbar <odunbar@caltech.edu>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This code was developed for application to Cloudy, a microphysics toy model, using Ollie's existing CES code. The idea is to take this code as a starting point for developing a general-purpose CES code that is fully "disentangled" from Cloudy.