Unpin scipy & use threadpoolctl in minimize_with_timeout (#2712)

Summary:
Scipy>=1.15 was leading to significant slowdowns in CI due to issues with OpenBLAS configuration (see scipy/scipy#22438). Thanks to suggestions by SciPy maintainers, we've found a solution that resolves the slowdown without modifying and global configurations for OpenBLAS.

This PR wraps `minimize` calls in `minimize_with_timeout` in a `with threadpool_limits(limits=1, user_api="blas")` context, which makes OpenBLAS operate single-threaded within the context. Since PyTorch relies on OpenMP & MKL rather than OpenBLAS, this does not affect the evaluations of the acquisition functions or the MLL.

Pull Request resolved: #2712

Test Plan:
Compare runtimes in the following tutorial workflows
Baseline - with scipy 1.14.1, without `threadpool_limits`: https://github.com/pytorch/botorch/actions/runs/13044166839/job/36391816602
With scipy 1.14.1, with `threadpool_limits`: https://github.com/pytorch/botorch/actions/runs/13044167834/job/36391818438
With scipy 1.15.1, with `threadpool_limits`: https://github.com/pytorch/botorch/actions/runs/13044245918/job/36392026548

Reviewed By: Balandat

Differential Revision: D68874700

Pulled By: saitcakmak

fbshipit-source-id: c6273bd43a8312e215204207bf92a765eb5201ce

botorch/optim/utils/timeout.py

@@ -14,6 +14,7 @@
 import numpy.typing as npt
 from botorch.exceptions.errors import OptimizationTimeoutError
 from scipy import optimize
+from threadpoolctl import threadpool_limits
 
 
 def minimize_with_timeout(
@@ -79,20 +80,23 @@ def wrapped_callback(xk: npt.NDArray) -> None:
 
     try:
         warnings.filterwarnings("error", message="Method .* cannot handle")
-        return optimize.minimize(
-            fun=fun,
-            x0=x0,
-            args=args,
-            method=method,
-            jac=jac,
-            hess=hess,
-            hessp=hessp,
-            bounds=bounds,
-            constraints=constraints,
-            tol=tol,
-            callback=wrapped_callback,
-            options=options,
-        )
+        # To prevent slowdowns after scipy 1.15.
+        # See https://github.com/scipy/scipy/issues/22438.
+        with threadpool_limits(limits=1, user_api="blas"):
+            return optimize.minimize(
+                fun=fun,
+                x0=x0,
+                args=args,
+                method=method,
+                jac=jac,
+                hess=hess,
+                hessp=hessp,
+                bounds=bounds,
+                constraints=constraints,
+                tol=tol,
+                callback=wrapped_callback,
+                options=options,
+            )
     except OptimizationTimeoutError as e:
         msg = f"Optimization timed out after {e.runtime} seconds."
         current_fun, *_ = fun(e.current_x, *args)

requirements.txt

@@ -4,5 +4,6 @@ gpytorch==1.14
 linear_operator==0.6
 torch>=2.0.1
 pyro-ppl>=1.8.4
-scipy<1.15
+scipy
 multipledispatch
+threadpoolctl