diff --git a/extra/basisgen.py b/extra/basisgen.py
deleted file mode 100755
index 1af6501..0000000
--- a/extra/basisgen.py
+++ /dev/null
@@ -1,112 +0,0 @@
-#!/usr/bin/env python
-""" Generates a basis guess based on linear combination """
-
-__author__ = 'Stefano Carrazza'
-__license__ = 'GPL'
-__version__ = '1.0.0'
-__email__ = 'stefano.carrazza@mi.infn.it'
-
-import sys
-import numpy
-from mc2hessian import XGrid, Flavors, LocalPDF, splash
-
-def main(argv):
-    # Get input set name
-    nrep = []
-    pdf_name = ""
-    Q = 1.0
-    if len(argv) < 3: usage()
-    else:
-        pdf_name = argv[0]
-        v = argv[1].split()
-        for i in v:
-            nrep.append(int(i))
-        Q = float(argv[2])
-    print "- Basis generator at", Q, "GeV."
-
-        # Loading basic elements
-    fl = Flavors(3)
-    xgrid = XGrid(1e-5, 1e-1, 25, 25)
-    pdf = LocalPDF(pdf_name, nrep[0], xgrid, fl, Q, "")
-    nx = xgrid.n
-    nf = fl.n
-
-    # Step 1: create pdf covmat
-    print "\n- Building PDF covariance matrix."
-    cov = pdf.pdfcovmat()
-
-    val, vec = numpy.linalg.eigh(cov)
-    sqrtval = numpy.zeros(shape=len(val))
-    for i in range(len(val)):
-        if val[i] > 1e-12:
-            sqrtval[i] = 1.0/val[i]**0.5
-        else: print " [Warning] Removing eigenvalue", i, val[i]
-    sqrtinvcov = numpy.dot(vec, numpy.diag(sqrtval)).dot(vec.T)
-    print " [Done] "
-
-    # Step 2: determine the best an for each replica
-    d = 5
-    p = pdf.n_rep/d
-    print "\n- Splitting system in", p, "systems of", d, "replicas."
-
-    # compute matrices for linear dependence test
-    A = numpy.zeros(shape=(p, nx*nf, d))
-    for t in range(p):
-        for r in range(d):
-            for fl in range(nf):
-                for ix in range(nx):
-                    i = nx*fl + ix
-                    A[t, i, r] = pdf.xfxQ[r+t*d, fl, ix] - pdf.f0[fl, ix]
-        A[t] = sqrtinvcov.dot(A[t])
-
-    # solving the system and saving residuals
-    resA = numpy.zeros(shape=(p,d))
-    for t in range(p):
-        for r in range(d):
-            b = A[t, :, r]
-            a = numpy.delete(A[t], r, axis=1)
-            resA[t, r] = numpy.linalg.lstsq(a,b)[1]
-    print " [Done] "
-
-    # select a fraction of replicas of each subset
-    print "\n- Building the final system."
-    rep = numpy.zeros(shape=p, dtype=numpy.int64)
-    B = numpy.zeros(shape=(nf*nx, p))
-    for t in range(p):
-        s = numpy.argmin(resA[t])
-        rep[t] = s+1+t*d
-        B[:, t] = A[t, :, s]
-
-    # build the final system and solve
-    res = numpy.zeros(p)
-    for r in range(p):
-        b = B[:, r]
-        a = numpy.delete(B, r, axis=1)
-        res[r] = numpy.linalg.lstsq(a, b)[1]
-    print " [Done] "
-
-    # print replicas
-    sres = numpy.sort(res)
-    for n in nrep:
-        eps = sres[len(sres)-1-n]
-        print "\n- Selecting", n, ", cutoff for residuals:", eps
-
-        filename = pdf_name + "_hessian_" + str(n) + ".dat"
-        print "- Printing final", filename, "replicas"
-        r = numpy.where(res > eps)[0]
-        f = open(filename, 'wb')
-        for i in range(len(r)):
-            f.write(str(rep[r[i]]) + "\n")
-        f.close()
-
-    print " [Done]\n\n Now run the mc2hessian script with the custom basis.\n"
-
-
-def usage():
-    print "usage: ./basisgen.py [PDF LHAPDF set] [Number of replicas or string array] [Input energy]"
-    print "output: file with custom basis for the mc2hessian.py script\n"
-    exit()
-
-if __name__ == "__main__":
-    splash()
-    main(sys.argv[1:])