Spaces:

MilesCranmer
/

PySR

Running

App Files Files Community

MilesCranmer commited on Sep 28, 2020

Commit

44d381d

•

1 Parent(s): 5908dc9

Update todo

Browse files

Files changed (1) hide show

README.md +2 -2

README.md CHANGED Viewed

@@ -282,6 +282,8 @@ pd.DataFrame, Results dataframe, giving complexity, MSE, and equations
 - [x] Print score alongside MSE: \delta \log(MSE)/\delta \log(complexity)
 - [x] Calculating the loss function - there is duplicate calculations happening.
 - [x] Declaration of the weights array every iteration
 - [ ] Add true multi-node processing, with MPI, or just file sharing. Multiple populations per core.
     - Ongoing in cluster branch
 - [ ] Consider allowing multi-threading turned off, for faster testing (cache issue on travis). Or could simply fix the caching issue there.
@@ -297,7 +299,6 @@ pd.DataFrame, Results dataframe, giving complexity, MSE, and equations
 - [ ] Implement more parts of the original Eureqa algorithms: https://www.creativemachineslab.com/eureqa.html
 - [ ] Experiment with freezing parts of model; then we only append/delete at end of tree.
 - [ ] Sympy printing
-- [ ] Sympy evaluation
 - [ ] Consider adding mutation for constant<->variable
 - [ ] Hierarchical model, so can re-use functional forms. Output of one equation goes into second equation?
 - [ ] Use NN to generate weights over all probability distribution conditional on error and existing equation, and train on some randomly-generated equations
@@ -305,7 +306,6 @@ pd.DataFrame, Results dataframe, giving complexity, MSE, and equations
      - Not sure if possible, as binary trees are the real bottleneck.
 - [ ] Performance:
     - Use an enum for functions instead of storing them?
-    - Threaded recursion?
 - [ ] Idea: use gradient of equation with respect to each operator (perhaps simply add to each operator) to tell which part is the most "sensitive" to changes. Then, perhaps insert/delete/mutate on that part of the tree?
 - [ ] For hierarchical idea: after running some number of iterations, do a search for "most common pattern". Then, turn that subtree into its own operator.
 - [ ] Additional degree operators?

 - [x] Print score alongside MSE: \delta \log(MSE)/\delta \log(complexity)
 - [x] Calculating the loss function - there is duplicate calculations happening.
 - [x] Declaration of the weights array every iteration
+- [x] Sympy evaluation
+- [x] Threaded recursion
 - [ ] Add true multi-node processing, with MPI, or just file sharing. Multiple populations per core.
     - Ongoing in cluster branch
 - [ ] Consider allowing multi-threading turned off, for faster testing (cache issue on travis). Or could simply fix the caching issue there.
 - [ ] Implement more parts of the original Eureqa algorithms: https://www.creativemachineslab.com/eureqa.html
 - [ ] Experiment with freezing parts of model; then we only append/delete at end of tree.
 - [ ] Sympy printing
 - [ ] Consider adding mutation for constant<->variable
 - [ ] Hierarchical model, so can re-use functional forms. Output of one equation goes into second equation?
 - [ ] Use NN to generate weights over all probability distribution conditional on error and existing equation, and train on some randomly-generated equations
      - Not sure if possible, as binary trees are the real bottleneck.
 - [ ] Performance:
     - Use an enum for functions instead of storing them?
 - [ ] Idea: use gradient of equation with respect to each operator (perhaps simply add to each operator) to tell which part is the most "sensitive" to changes. Then, perhaps insert/delete/mutate on that part of the tree?
 - [ ] For hierarchical idea: after running some number of iterations, do a search for "most common pattern". Then, turn that subtree into its own operator.
 - [ ] Additional degree operators?