Changeset 2ccb775 in sasmodels

Timestamp:

Nov 9, 2016 1:50:01 PM (8 years ago)

Author:

jhbakker

Branches:

master, core_shell_microgels, costrafo411, magnetic_model, ticket-1257-vesicle-product, ticket_1156, ticket_1265_superball, ticket_822_more_unit_tests

Children:

46d9f48

Parents:

1a6cd57

Message:

Experimenting with Hankel transform optimizations

Location:

sasmodels

Files:

• resolution.py (modified) (2 diffs)
• sesans.py (modified) (2 diffs)

sasmodels/resolution.py

@@ -59 +59 @@
 
 class SESANS1D(Resolution):
-
+    def __init__(self, data, H0, H, q_calc):
+        self.q = data.x
+        self.H0 = H0
+        self.H = H
+        self.data=data
+        self.q_calc = q_calc
+    def apply(self, theory):
+        return sesans.hankeltrafo(self.H0, self.H, theory)
+
+"""
     def __init__(self, data, q_calc):
         self.q = data.x
@@ -67 +76 @@
     def apply(self, theory):
         return sesans.transform(self.data, self.q_calc, theory, None, None)
-
+"""
 class Pinhole1D(Resolution):
     r"""

sasmodels/sesans.py

@@ -54 +54 @@
         # data.has_z_acceptance
         return [q]
+
+def hankeltrafo(H0, H, Iq_calc):
+    G0 = np.dot(H0, Iq_calc)
+    G=[]
+    for i in range(len(H[0, :])):
+        Gtmp = np.dot(H[:, i], Iq_calc)
+        G.append(Gtmp)
+    P = G - G0
+    return P  # This is the logarithmic P, not the linear one as found in Andersson2008
+
+"""
+class HankelTransform(object):
+    def __init__(self,q_calc, SElength):
+        dq = q_calc[1]-q_calc[0]
+        self.H0 = dq / (2 * pi) * q_calc
+        self.H = dq / (2 * pi) * besselj(0, np.outer(q_calc, SElength))
+
+    def apply(self, Iq_calc):
+        G0 = np.dot(self.H0, Iq_calc)
+        G = np.dot(self.H, Iq_calc)
+        return G - G0 #This is the logarithmic G, not the linear one as found in Andersson2008
+"""
 
 def transform(data, q_calc, Iq_calc, qmono, Iq_mono):
@@ -208 +230 @@
         integral = besselj(0, q*SElength_i)*Iq*q
         G[i] = np.sum(integral)
-    G0 = np.sum(Iq*q) #relation to ksi? For generalization to all models
+    #G0 = np.sum(Iq*q) #relation to ksi? For generalization to all models
 
     # [m^-1] step size in q, needed for integration