baseline_v1/src/functionality.py

"""A passive scalar-resistor benchmark, not an electronic-device simulator."""
import numpy as np
from scipy.sparse import coo_matrix
from scipy.sparse.linalg import spsolve

def conductance(material):
    sig=np.full(material.shape,1e-8)
    sig[material==0]=1e-7; sig[material==1]=1e-4; sig[material==2]=1.
    fixed=np.zeros(material.shape,bool); fixed[0]=True; fixed[-1]=True
    volt=np.zeros(material.shape); volt[0]=1.
    ids=np.full(material.shape,-1,int); ids[~fixed]=np.arange((~fixed).sum())
    n=int((~fixed).sum()); rows=[]; cols=[]; data=[]; rhs=np.zeros(n)
    diag=np.zeros(n)
    for ax in range(material.ndim):
        a=[slice(None)]*material.ndim; b=a.copy(); a[ax]=slice(None,-1); b[ax]=slice(1,None)
        a,b=tuple(a),tuple(b)
        g=2*sig[a]*sig[b]/(sig[a]+sig[b]); ia,ib=ids[a],ids[b]
        for i,j,vj in [(ia,ib,volt[b]),(ib,ia,volt[a])]:
            inside=i>=0
            np.add.at(diag,i[inside],g[inside])
            ff=inside&(j<0); np.add.at(rhs,i[ff],(g*vj)[ff])
            both=inside&(j>=0)
            rows.extend(i[both]);cols.extend(j[both]);data.extend(-g[both])
    rows.extend(range(n));cols.extend(range(n));data.extend(diag)
    A=coo_matrix((data,(rows,cols)),shape=(n,n)).tocsr()
    volt[~fixed]=spsolve(A,rhs)
    g0=2*sig[0]*sig[1]/(sig[0]+sig[1])
    return float(np.sum(g0*(1-volt[1])))