Skip to content

Criticality search method on the Model class #3569

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
paulromano merged 15 commits into from
Oct 28, 2025
Merged

Criticality search method on the Model class #3569

Changes from 1 commit
Commits
Show all changes
15 commits
Select commit Hold shift + click to select a range
e23d476
Implement Model.keff_search method
paulromano Jul 31, 2025
9ab24db
Implement GRsecant algorithm
paulromano Aug 9, 2025
b7d4ef8
Determine number of generations based on r=0.5
paulromano Aug 11, 2025
d48358e
Add one more TODO
paulromano Aug 11, 2025
705b37d
Lots of updates to keff_search and grsecant
paulromano Sep 11, 2025
eec0f94
Inline grsecant in Model.keff_search
paulromano Sep 11, 2025
0072992
Add description of GRsecant in docstring
paulromano Sep 11, 2025
a9ea0aa
Rename print_iterations --> output
paulromano Sep 11, 2025
9dce3df
Add support for keff search with openmc.lib
paulromano Sep 15, 2025
36fccf7
Get rid of b1 argument
paulromano Sep 15, 2025
2a15eb8
Add documentation on stopping criterion
paulromano Sep 15, 2025
0d580ad
Fix two failing tests
paulromano Sep 17, 2025
d49af74
Merge branch 'develop' into model-keff-search
paulromano Oct 13, 2025
7948ba6
Incorporate suggestion from @church89
paulromano Oct 13, 2025
efcb773
Add test for keff_search method
paulromano Oct 13, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
Determine number of generations based on r=0.5
  • Loading branch information
@paulromano
paulromano committed Sep 10, 2025
commit b7d4ef83a1ca17bac1a59ce52eac073fadca704b
31 changes: 18 additions & 13 deletions openmc/model/model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2259,6 +2259,14 @@ class RootResult:
flag: str


# TODO: Consistent use of bracket vs x0, x1
# TODO: Account for min/max search interval (vs initial points)
# TODO: Allow for passing a single x0 value?
# TODO: Clean up use of g0, g1
# TODO: Reporting total number of generations?
# TODO: Fit on sigam vs g is done with last 4 points. Change this?
# TODO: Change order of attributes in RootResult / change return datatype?

def root_scalar_grsecant(
f,
x0: float,
Expand Down Expand Up @@ -2334,21 +2342,20 @@ def eval_at(x: float, g: int):

# ---------- Step 3: choose generations to hit σ_target (Appendix C)

# Use at least two past points for regression; otherwise, assume r≈0.5
if len(gs) >= 4 and np.var(np.log(gs)) > 0.0:
# Perform a curve fit on ln(sigma) = ln(k) - r*ln(g) to solve for
# ln(k) and r.
popt, _ = curve_fit(
lambda ln_g, ln_k, r: ln_k - r*ln_g,
np.log(gs), np.log(ss), absolute_sigma=False
# Use at least two past points for regression
if len(gs) >= 2 and np.var(np.log(gs)) > 0.0:
# Perform a curve fit based on Eq. (C.3) to solve for ln(k). Note
# that unlike the paper, we do not leave r as an undetermined
# parameter and choose r=0.5.
(ln_k,), _ = curve_fit(
lambda ln_g, ln_k: ln_k - 0.5*ln_g,
np.log(gs[-4:]), np.log(ss[-4:]),
)
ln_k, r = popt
k = float(np.exp(ln_k))
else:
r = 0.5
k = float(ss[-1] * gs[-1]**r)
k = float(ss[-1] * math.sqrt(gs[-1]))

g_new = (k / sig_target) ** (1 / r)
g_new = (k / sig_target) ** 2

# Clamp and round up to integer
g_new = max(min_g, math.ceil(g_new))
Expand All @@ -2358,8 +2365,6 @@ def eval_at(x: float, g: int):
# Evaluate at proposed x with chosen effort
f_new, s_new = eval_at(x_new, g_new)

#print(f'param_k={k}, {r=}, {sig_target=}, {g_new=}, {f_new=}, {s_new=}')

# Termination: both criteria (|f| and σ) as in the paper
if (abs(f_new) <= k_tol and s_new <= sigma_final):
return RootResult(x_new, True, it + 1, len(xs), "converged")
Expand Down