ε and Γ_c objectives with Fourier Bounce #1290
Open
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
which was used just to compute R0. probably not worth the memory
… can actually merge it
unalmis
commented
Oct 23, 2024
I can't reproduce. maybe let's wait until things like #1293 get merged before debugging |
This descript copies the comment at #1290 (comment). Due to how we make sure enough resolution is used when we call `eq.compute`, the code I gave above actually computes this quantity by computing all dependencies of the effective ripple on this higher resolution grid: ```python all_dependencies_grid = LinearGrid(rho=rho, M=eq.M_grid, N=eq.N_grid, NFP=eq.NFP, sym=False) ``` Because #1313 was not resolved yet, I had marked `effective ripple` as a quantity that should not be overriden when we call `eq.compute`. That means that once all the input data to the `effective ripple` was computed accurately on the above grid. The computation of the effective ripple itself used this grid ```python eps_eff_grid = LinearGrid(rho=rho, theta=eq.M_grid, zeta=eq.N_grid, NFP=eq.NFP, sym=False) ``` This grid has half as many nodes in poloidal and toroidal directions on which we compute FFTs to do all the integrals. The plot above is the result of this computation against Neo. To generate the Neo plot we converted the equilibrium to boozer coordinates at three times the DESC equilibrium (eq.M, eq.N, etc.) resolution. So the conclusion is that when all the bounce integrals are done using FFTs that are generated from half typical desc resolution, then effective ripple converges to Neo produces from an input with boozer resolution three times the DESC resolution. Now that #1313 is resolved, I have merged that PR into this branch, and marked effective ripple as a flux surface quantity (by saying resolution_requirement="tz"). That means unless `override_grid=False` is given then everything, including all the bounce integrals will be computed on a grid with resolution at least ```python all_dependencies_grid = LinearGrid(rho=rho, M=eq.M_grid, N=eq.N_grid, NFP=eq.NFP, sym=False) ``` The old behavior can be recovered by marking `effective ripple 3/2` compute as something other than requiring theta and zeta resolution. i.e. changing `resolution_requirement="tz"` to `resolution_requirement="z"`. Then the code I gave above (repeated here), will reproduce that plot. ```python eq = get("W7-X") rho = np.linspace(1e-12, 1, 60) # the axis limit IS implemented by the way grid = LinearGrid(rho=rho, theta=eq.M_grid, zeta=eq.N_grid, NFP=eq.NFP, sym=False) num_transit = 64 data = eq.compute( "effective ripple 3/2", grid=grid, theta=Bounce2D.compute_theta(eq, X=32, Y=64, rho=rho), num_transit=num_transit, num_quad=64, num_pitch=200, num_well=20 * num_transit, ) ```
Still can't reproduce. |
unalmis
changed the title
Peak optimization memory usage is 5gb.
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 PR updates the
Gamma_c,Gamma_c Velasco#1042 andeffective ripple#1003 objectives to use the numerical methods from #1119.