youainti
|
40ccb4e84e
|
Fixed debris issue and setup plots
|
4 years ago |
|
youainti
|
f17ba05d7f
|
implemented save and reload functionality.
|
4 years ago |
|
youainti
|
f8d3643329
|
Finished presentation, got congergent social planner problem.
|
4 years ago |
|
will king
|
333bc5000c
|
working convergence, with a saved model
|
4 years ago |
|
youainti
|
be47bfc2f1
|
added notes on the things to finalize other than conclusion.
|
4 years ago |
|
youainti
|
81bc58ba71
|
Got most of the presentation finished
|
4 years ago |
|
youainti
|
bf57b47138
|
Proofread for seminar. Updated mathematic notation
|
4 years ago |
|
youainti
|
3f15afab76
|
Cleaned up results area, although it is still missing things.
|
4 years ago |
|
youainti
|
c2dd67c744
|
finished rewriting computational methods section.
|
4 years ago |
|
youainti
|
d762e32cf7
|
Removed social planner's references to euler equations.
|
4 years ago |
|
youainti
|
4a189109e6
|
Removed euler equation references from operator's problem
|
4 years ago |
|
youainti
|
f5447b116c
|
Outlined presentation and began work on it.
|
4 years ago |
|
will king
|
56a89717ab
|
outline of story and ending of presentation
|
4 years ago |
|
youainti
|
cf9a581bec
|
Almost completed redesign of the system. Needs tested.
|
4 years ago |
|
youainti
|
90478868e1
|
Working branched policy operator and in the middle of cleaning up how loss function etc work
|
4 years ago |
|
youainti
|
76b8cfdc8f
|
Experimenting with flux chains and parallel to get some background knowledge on how to construct the operator policy function
|
4 years ago |
|
youainti
|
c3b4b3e8b4
|
Currently working Social Planner's problem. but I need to get operators problem working.
|
4 years ago |
|
youainti
|
5d60a23270
|
current work on making operator level maximization as opposed to just the planner level.
|
4 years ago |
|
youainti
|
2414e15735
|
properly implemented social planner's problem
|
5 years ago |
|
youainti
|
259a5c942f
|
just committing some changes I have locally to keep them separate from the main work going on in BellmanResidualMinimization.jl
|
5 years ago |
|
youainti
|
f47dd09a8b
|
Got basic planner problem converging.
|
5 years ago |
|
youainti
|
55911694b7
|
Fixed loss function.
|
5 years ago |
|
youainti
|
6dab0f3b4d
|
created the basic bellman residual minimization approach, still needs work.
|
5 years ago |
|
youainti
|
63135d8298
|
Finished investigation of using Maliar et al's approach. Currently not feasible.
|
5 years ago |
|
youainti
|
66a237cd19
|
current work, including a whole lot of julia stuff (Pluto.jl FTW)
|
5 years ago |
|
youainti
|
d5635622cc
|
current work, can't get convergence as I'd like
|
5 years ago |
|
youainti
|
90a4a56baf
|
current work on differentiating the profit function
|
5 years ago |
|
youainti
|
1ad55eb603
|
isolated issue with convergence of EstimandNN to the Partials NN, and moved previous work on dimension from SimplifiedApproach0 to NeuralNetworkSpecification
|
5 years ago |
|
youainti
|
1595bd2241
|
got choice and partials functions to use correct dimensionality
|
5 years ago |
|
youainti
|
d8fff40288
|
Got most of the neural network stuff working. Some parameter updates to manage in NeuralNetworkSpecifications
|
5 years ago |
|
youainti
|
7b51044057
|
I've got the neural net with both launches and partials written, although some internals need redone because I misunderstood ReLU. Those misunderstandings are documented in the Background.
|
5 years ago |
|
youainti
|
2657f4d821
|
got most of the new interfaces/abstractions written, and they appear to be working. Have the basics of a NN outlined. Have added a document with infomration about neural networks and PyTorch in particular. Wrote my own Upscaling module.
|
5 years ago |
|
youainti
|
d0bcdd87c0
|
saving work to be able work on it at home. Wrote most of the new abstractions (concrete and abstract classes) over states and model definition. Still missing some functions.
|
5 years ago |
|
youainti
|
cd994a640e
|
Fixed neural network stuff, and came up with an example training loop. First implementation of optimality conditions completed.
|
5 years ago |
|
youainti
|
d1840a9ed7
|
Got a basic optimality condition function working
|
5 years ago |
|
youainti
|
21b91cd4e8
|
Changed to have launches as a global function
|
5 years ago |
|
youainti
|
9335ce1295
|
verified working state in connect_trans* and moved the next steps from that file to README
|
5 years ago |
|
youainti
|
28dcb5a359
|
have addressed the merge issue by forcing successful_recursing to remote version as recorded in connect_transition_to_optimality. This should have working versions of everything.
|
5 years ago |
|
youainti
|
92cb373b06
|
updated a couple of files (one of which will be overwritten in a soon to exist pull/merge), and added a readme
|
5 years ago |
Will King
|
ef550d6098
|
got transition recursion working, and created a notebook for the next step: connecting transition to optimality and creating a loss function
|
5 years ago |
|
youainti
|
3e8b8f5284
|
reviewed basic neural net (minor change) and changed the destruction (satellite loss) function in successfull recursion
|
5 years ago |
|
Will King
|
de54b5767c
|
Got to a good point to share, have incorporated all of Tom Marsh's suggestions.
|
5 years ago |
|
youainti
|
fb5e70820c
|
Almost finished last of suggested changes
|
5 years ago |
|
youainti
|
0c941a4335
|
Rearranged main layout, part way through updating sections, added sections. Updated current status in readme.
|
5 years ago |
|
youainti
|
9d082f4667
|
Removed section on suvival rates
|
5 years ago |
|
youainti
|
af4783827f
|
quick note
|
5 years ago |
|
youainti
|
322fd02fa2
|
todays work
|
5 years ago |
|
youainti
|
9f65785327
|
Began incorporating Tom Marsh's suggestions
|
5 years ago |
|
youainti
|
55f8a0b00d
|
changes to recursion attempt
|
5 years ago |
|
youainti
|
b40a6af274
|
finished fixing references
|
5 years ago |