the new kinematics don't seem to help :(

prog/python/qqgg/monte_carlo.py

@@ -124,15 +124,16 @@ def find_upper_bound(f, interval, **kwargs):
 
 def find_upper_bound_vector(f, interval):
     result = shgo(_negate(f), bounds=interval, options=dict(maxfev=100))
+
     if not result.success:
         raise RuntimeError("Could not find an upper bound.", result)
 
-    upper_bound = -result.fun + 0.1
+    upper_bound = -result.fun
     return upper_bound
 
 
 def sample_unweighted_vector(
-    f, interval, seed=None, upper_bound=None, report_efficiency=False, num=None
+    f, interval, seed=None, upper_bound=None, report_efficiency=False
 ):
     dimension = len(interval)
     interval = np.array([_process_interval(i) for i in interval])

prog/python/qqgg/parton_density_function_stuff.org

@@ -230,62 +230,6 @@ nicer.
 
 #+RESULTS:
 
-** Test Driving
-Now, let's try it out.
-#+begin_src jupyter-python :exports both :results raw drawer
-  dist, x_limits = get_xs_distribution_with_pdf(
-      diff_xs, averaged_tchanel_q2, e_proton
-  )
-#+end_src
-
-#+RESULTS:
-
-Let's plot it for some random values 😃.
-#+begin_src jupyter-python :exports both :results raw drawer
-  fig, ax = set_up_plot()
-  pts = np.linspace(*interval_cosθ, 1000)
-
-  ax.plot(pts, [dist([cosθ, 0.3, 0.3]) for cosθ in pts])
-#+end_src
-
-#+RESULTS:
-:RESULTS:
-| <matplotlib.lines.Line2D | at | 0x7f0bd18b0700> |
-[[file:./.ob-jupyter/4b7773815fe4943f422b5943cc67e48b7a75cc23.png]]
-:END:
-
-Having set both x to the same value, we get a symmetric distribution as expected.
-Just the magnitude is a little startling! The value 1/3 is intentional!
-
-Now we gonna take some samples!
-But first we have to find an upper bound, which is expensive!
-
-#+begin_src jupyter-python :exports both :results raw drawer
-  intervals = [interval_cosθ, [.01, 1], [.01, 1]]
-  upper_bound = monte_carlo.find_upper_bound_vector(dist, intervals)
-  upper_bound
-#+end_src
-
-#+RESULTS:
-: 5721.40648474465
-
-Beware!, this is darn slow, becaus the efficiency is soooo low.
-#+begin_src jupyter-python :exports both :results raw drawer
-
-  sample_momenta(
-      100,
-      dist,
-      intervals,
-      e_proton,
-      upper_bound=upper_bound,
-      proc="auto",
-      cache="cache/pdf/samp_costh_test",
-  )[1]
-#+end_src
-
-#+RESULTS:
-: 0.0004240744427426794
-
 ** Switching Horses: Sampling η
 We set up a new distribution.
 #+begin_src jupyter-python :exports both :results raw drawer
@@ -308,8 +252,8 @@ Plotting it, we can see that the variance is reduced.
 
 #+RESULTS:
 :RESULTS:
-| <matplotlib.lines.Line2D | at | 0x7f0bcf65abe0> |
-[[file:./.ob-jupyter/5f2e010bf22bb8e157d5327258268cf8a465510d.png]]
+| <matplotlib.lines.Line2D | at | 0x7f3574d07820> |
+[[file:./.ob-jupyter/5597ca6056db11908cfca64c2090d67e3b94cc9e.png]]
 :END:
 
 Lets plot how the pdf looks.
@@ -323,7 +267,7 @@ Lets plot how the pdf looks.
 
 #+RESULTS:
 :RESULTS:
-| <matplotlib.lines.Line2D | at | 0x7f0bd1831100> |
+| <matplotlib.lines.Line2D | at | 0x7f3572b7b8b0> |
 [[file:./.ob-jupyter/b92f0c4b2c9f2195ae14444748fcdb7708d81c19.png]]
 :END:
 
@@ -332,27 +276,22 @@ Now we sample some events. Doing this in parallel helps. We let the os
 figure out the cpu mapping.
 
 #+begin_src jupyter-python :exports both :results raw drawer
-  intervals_η = [interval_η, [.01, 1], [.01, 1]]
+  intervals_η = [interval_η, [.1, 1], [.1, 1]]
 
   result, eff = monte_carlo.sample_unweighted_array(
-      num_samples,
+      10000,
       dist_η,
       interval=intervals_η,
       proc="auto",
       report_efficiency=True,
-      cache="cache/pdf/huge",
+      #cache="cache/pdf/huge",
   )
   result
+  eff
 #+end_src
 
 #+RESULTS:
-: array([[-1.43205911,  0.34762794,  0.01802916],
-:        [-2.13582434,  0.02851933,  0.07415878],
-:        [ 1.79962451,  0.01835771,  0.2725591 ],
-:        ...,
-:        [ 1.1782848 ,  0.1113013 ,  0.01554592],
-:        [-1.27942179,  0.12550189,  0.05389335],
-:        [ 2.22491947,  0.04076429,  0.13610809]])
+: 0.003007891162077376
 
 
 
@@ -369,6 +308,6 @@ Let's look at a histogramm of eta samples.
 
 #+RESULTS:
 :RESULTS:
-| <Figure | size | 432x288 | with | 1 | Axes> | <matplotlib.axes._subplots.AxesSubplot | at | 0x7f0bcd5a1be0> |
-[[file:./.ob-jupyter/b55eed7143c116ac0471a510bd62174a42e1ac31.png]]
+| <Figure | size | 432x288 | with | 1 | Axes> | <matplotlib.axes._subplots.AxesSubplot | at | 0x7f35728b26a0> |
+[[file:./.ob-jupyter/ec474fc3576110c487c7fb31403cbab0a063efa9.png]]
 :END: