diff --git a/aipy/amp.py b/aipy/amp.py
index 43ed5b9..9e19efc 100644
--- a/aipy/amp.py
+++ b/aipy/amp.py
@@ -125,7 +125,7 @@ def response(self, xyz):
 class Beam2DGaussian(phs.Beam):
     """Representation of a 2D Gaussian beam pattern, with default setting for
     a flat beam."""
-    def __init__(self, freqs, xwidth=np.Inf, ywidth=np.Inf):
+    def __init__(self, freqs, xwidth=np.inf, ywidth=np.inf):
         """xwidth = angular width (radians) in EW direction
         ywidth = angular width (radians) in NS direction"""
         phs.Beam.__init__(self, freqs)
diff --git a/aipy/deconv.py b/aipy/deconv.py
index 722a75d..76d6627 100644
--- a/aipy/deconv.py
+++ b/aipy/deconv.py
@@ -77,7 +77,7 @@ def recenter(a, c):
     return a2
 
 def lsq(im, ker, mdl=None, area=None, gain=.1, tol=1e-3, maxiter=200,
-        lower=lo_clip_lev, upper=np.Inf, verbose=False):
+        lower=lo_clip_lev, upper=np.inf, verbose=False):
     """This simple least-square fitting procedure for deconvolving an image
     saves computing by assuming a diagonal pixel-pixel gradient of the fit.
     In essence, this assumes that the convolution kernel is a delta-function.
@@ -132,7 +132,7 @@ def f(x):
     return x, info
 
 def maxent(im, ker, var0, mdl=None, gain=.1, tol=1e-3, maxiter=200,
-        lower=lo_clip_lev, upper=np.Inf, verbose=False):
+        lower=lo_clip_lev, upper=np.inf, verbose=False):
     """Maximum entropy deconvolution (MEM) (see Cornwell and Evans 1984
     "A Simple Maximum Entropy Deconvolution Algorithm" and Sault 1990
     "A Modification of the Cornwell and Evans Maximum Entropy Algorithm")
@@ -197,7 +197,7 @@ def dot(x, y): return (x*y/-gg_J).sum()
     return b_i, info
 
 def maxent_findvar(im, ker, var=None, f_var0=.6, mdl=None, gain=.1, tol=1e-3,
-        maxiter=200, lower=lo_clip_lev, upper=np.Inf, verbose=False,
+        maxiter=200, lower=lo_clip_lev, upper=np.inf, verbose=False,
         maxiterok=False):
     """This frontend to maxent tries to find a variance for which maxent will
     converge.  If the starting variance (var) is not specified, it will be
@@ -237,7 +237,7 @@ def maxent_findvar(im, ker, var=None, f_var0=.6, mdl=None, gain=.1, tol=1e-3,
     if verbose: print('Done with MEM.')
     return cl, info
 
-def anneal(im, ker, mdl=None, maxiter=1000, lower=lo_clip_lev, upper=np.Inf,
+def anneal(im, ker, mdl=None, maxiter=1000, lower=lo_clip_lev, upper=np.inf,
         cooling=lambda i,x: 1e+1*(1-np.cos(i/50.))*(x**2), verbose=False):
     """Annealing takes a non-deterministic approach to deconvolution by
     randomly perturbing the model and selecting perturbations that improve the
diff --git a/scripts/cl_img.py b/scripts/cl_img.py
index c66c99d..9d65f85 100755
--- a/scripts/cl_img.py
+++ b/scripts/cl_img.py
@@ -78,7 +78,7 @@ def to_fits(prefix,ftag,i,kwds):
             level = float(opts.rewgt.split('(')[-1][:-1])
             abms = np.abs(bms)
             thresh = abms.max() * level
-            divisor = abms.clip(thresh, np.Inf)
+            divisor = abms.clip(thresh, np.inf)
             uvs /= divisor; bms /= divisor
         elif opts.rewgt.startswith('radial'):
             #x,y = np.indices(dim.shape)
diff --git a/scripts/combine_freqs.py b/scripts/combine_freqs.py
index 9c88074..5530355 100755
--- a/scripts/combine_freqs.py
+++ b/scripts/combine_freqs.py
@@ -54,7 +54,7 @@ def mfunc(uv, p, d, f):
             d = d.sum(axis=1)
             f.shape = (opts.nchan, nchan/opts.nchan)
             f = np.logical_not(f).astype(np.int_).sum(axis=1)
-            d /= f.clip(1,np.Inf)
+            d /= f.clip(1,np.inf)
             if opts.careful_flag: f = np.where(f < nchan/opts.nchan, 1, 0)
             elif opts.dont_flag: f = np.where(f < 1, 1, 0)
             else: f = np.where(f <= nchan/opts.nchan/2, 1, 0)
diff --git a/scripts/fitmdl.py b/scripts/fitmdl.py
index 8d1e186..2b0530c 100755
--- a/scripts/fitmdl.py
+++ b/scripts/fitmdl.py
@@ -54,7 +54,7 @@
 aa.set_jultime(t)
 cat.compute(aa)
 del(uv)
-if opts.maxiter < 0: opts.maxiter = np.Inf
+if opts.maxiter < 0: opts.maxiter = np.inf
 
 # Figure out parameters to fit
 prms, prm_dict, shkeys = {}, {}, []
@@ -234,7 +234,7 @@ def fit_func(prms):
         fit_func, prm_list,
         #args=(args, opts.decimate, opts.decphs),
         full_output=1, disp=0,
-        maxfun=opts.maxiter, maxiter=np.Inf,
+        maxfun=opts.maxiter, maxiter=np.inf,
         ftol=opts.ftol, xtol=opts.xtol
     )
     prms,score = rv[:2]
@@ -252,7 +252,7 @@ def fit_func(prms):
         fit_func, prm_list,
         args=(args, opts.decimate, opts.decphs),
         full_output=1, disp=0,
-        maxfun=opts.maxiter, maxiter=np.Inf,
+        maxfun=opts.maxiter, maxiter=np.inf,
         ftol=opts.ftol, xtol=opts.xtol
     )
     prms,score = rv[:2]
@@ -283,7 +283,7 @@ def fit_func(prms):
                 fit_func, prm_list,
                 args=([uvfile], decimate, opts.decimate*cnt + opts.decphs),
                 full_output=1, disp=0,
-                maxfun=opts.maxiter, maxiter=np.Inf,
+                maxfun=opts.maxiter, maxiter=np.inf,
                 ftol=opts.ftol, xtol=opts.xtol
             )
             prms,score = rv[:2]
diff --git a/scripts/xtalk3.py b/scripts/xtalk3.py
index dd0c38a..ecdfecb 100755
--- a/scripts/xtalk3.py
+++ b/scripts/xtalk3.py
@@ -91,7 +91,7 @@
             guess[bl] += np.where(f, 0, d)
             cnt[bl] += np.logical_not(f)
         del(uv)
-        for bl in guess: xtalk[bl] = guess[bl] / np.clip(cnt[bl], 1, np.Inf)
+        for bl in guess: xtalk[bl] = guess[bl] / np.clip(cnt[bl], 1, np.inf)
     if opts.outfile:
         xfile = '%f.xtalk.npz' % jd
         print('Writing', xfile)