use same ludr convention for hotrg2

hotrg2.py

@@ -6,13 +6,11 @@ adapted from Jin Chen's hotrg.jl at https://github.com/wangleiphy/TRG/blob/maste
 
 def gauge(T, Dcut, side):
     if side=='l':
-        M = torch.einsum('xazb,xcwb,wlyn,zmyn->amcl' , (T, T, T, T))
+        M = torch.einsum('axzb,cxwb,lwyn,mzyn->amcl', (T, T, T, T))
     elif side=='r':
-        M = torch.einsum('xbza,xbwc,wlyn,zlym->amcn', (T, T, T, T))
+        M = torch.einsum('bxza,bxwc,lwyn,lzym->amcn', (T, T, T, T))
     D = M.shape[0]
     M = M.contiguous().view(D**2, D**2)
-    #M = ( M + M.t()) / 2.0;
-    #w, v = torch.symeig(M, eigenvectors=True)
 
     U, S, V = torch.svd(M)
     Dnew = min(D**2, Dcut)
@@ -32,18 +30,18 @@ def hotrg(T, D, Dcut, no_iter):
         vr, er = gauge(T, Dcut,'r')
         U = vl if (el < er) else vr
         truncation_error += min(el, er)
-        T = torch.einsum('xaob,amz,omyn,bnw->wxzy',(T, U, T, U))
+        T = torch.einsum('axob,amz,moyn,bnw->xwzy',(T, U, T, U))
        
     trace = 0.0 
     for x in range(T.shape[0]):
         for y in range(T.shape[1]):
-            trace += T[x, y, x, y]
+            trace += T[x, y, y, x]
     lnZ += torch.log(trace)
     return lnZ, truncation_error
 
 if __name__=='__main__':
     K = torch.tensor([0.44])
-    Dcut = 10
+    Dcut = 20
     n = 20
 
     #Boltzmann factor on a bond M=LR^T
@@ -60,7 +58,6 @@ if __name__=='__main__':
     #            |
     #            R^{T}
     
-    # T is a D*D*D*D tensor u l d r
-    T = torch.einsum('ai,aj,ak,al->ijkl', (L, R, R, L)) 
+    T = torch.einsum('ai,aj,ak,al->ijkl', (R, L, R, L)) 
     lnZ, _ = hotrg(T, 2, Dcut, n)
     print (lnZ.item()/2**n)