Removed numba, made all pyfftw imports conditional

Author: bbfrederick

Dockerfile.safe

@@ -77,10 +77,8 @@ RUN mamba install -y python \
                      h5py \
                      "tensorflow>=2.4.0" \
                      pyqtgraph \
-                     pyfftw \
                      pandas \
-                     versioneer \
-                     numba; sync && \
+                     versioneer; sync && \
     chmod -R a+rX /usr/local/miniconda; sync && \
     chmod +x /usr/local/miniconda/bin/*; sync && \
     conda-build purge-all; sync && \

capcalc/filter.py

@@ -23,23 +23,27 @@
 
 """
 import sys
+import warnings
 
 import matplotlib.pyplot as plt
 import numpy as np
-import pyfftw
 
-# from numba import jit
+with warnings.catch_warnings():
+    warnings.simplefilter("ignore")
+    try:
+        import pyfftw
+    except ImportError:
+        pyfftwpresent = False
+    else:
+        pyfftwpresent = True
+
 from scipy import fftpack, ndimage, signal
 from scipy.signal import savgol_filter
 
-# import warnings
-
+if pyfftwpresent:
+    fftpack = pyfftw.interfaces.scipy_fftpack
+    pyfftw.interfaces.cache.enable()
 
-# fftpack = pyfftw.interfaces.scipy_fftpack
-# pyfftw.interfaces.cache.enable()
-
-# ---------------------------------------- Global constants -------------------------------------------
-donotusenumba = True
 
 # ----------------------------------------- Conditional imports ---------------------------------------
 try:
@@ -49,23 +53,6 @@
 except ImportError:
     memprofilerexists = False
 
-
-# ----------------------------------------- Conditional jit handling ----------------------------------
-def conditionaljit():
-    def resdec(f):
-        global donotusenumba
-        if donotusenumba:
-            return f
-        return jit(f, nopython=False)
-
-    return resdec
-
-
-def disablenumba():
-    global donotusenumba
-    donotusenumba = True
-
-
 # --------------------------- Filtering functions -------------------------------------------------
 # NB: No automatic padding for precalculated filters
 
@@ -177,7 +164,8 @@ def ssmooth(xsize, ysize, zsize, sigma, inputdata):
 
 
 # - butterworth filters
-@conditionaljit()
+
+
 def dolpfiltfilt(Fs, upperpass, inputdata, order, padlen=20, cyclic=False, debug=False):
     r"""Performs a bidirectional (zero phase) Butterworth lowpass filter on an input vector
     and returns the result.  Ends are padded to reduce transients.
@@ -235,7 +223,6 @@ def dolpfiltfilt(Fs, upperpass, inputdata, order, padlen=20, cyclic=False, debug
     ).astype(np.float64)
 
 
-@conditionaljit()
 def dohpfiltfilt(Fs, lowerpass, inputdata, order, padlen=20, cyclic=False, debug=False):
     r"""Performs a bidirectional (zero phase) Butterworth highpass filter on an input vector
     and returns the result.  Ends are padded to reduce transients.
@@ -292,7 +279,6 @@ def dohpfiltfilt(Fs, lowerpass, inputdata, order, padlen=20, cyclic=False, debug
     )
 
 
-@conditionaljit()
 def dobpfiltfilt(Fs, lowerpass, upperpass, inputdata, order, padlen=20, cyclic=False, debug=False):
     r"""Performs a bidirectional (zero phase) Butterworth bandpass filter on an input vector
     and returns the result.  Ends are padded to reduce transients.
@@ -419,7 +405,6 @@ def getlpfftfunc(Fs, upperpass, inputdata, debug=False):
     return transferfunc
 
 
-@conditionaljit()
 def dolpfftfilt(Fs, upperpass, inputdata, padlen=20, cyclic=False, debug=False):
     r"""Performs an FFT brickwall lowpass filter on an input vector
     and returns the result.  Ends are padded to reduce transients.
@@ -462,7 +447,6 @@ def dolpfftfilt(Fs, upperpass, inputdata, padlen=20, cyclic=False, debug=False):
     return unpadvec(fftpack.ifft(inputdata_trans).real, padlen=padlen)
 
 
-@conditionaljit()
 def dohpfftfilt(Fs, lowerpass, inputdata, padlen=20, cyclic=False, debug=False):
     r"""Performs an FFT brickwall highpass filter on an input vector
     and returns the result.  Ends are padded to reduce transients.
@@ -505,7 +489,6 @@ def dohpfftfilt(Fs, lowerpass, inputdata, padlen=20, cyclic=False, debug=False):
     return unpadvec(fftpack.ifft(inputdata_trans).real, padlen=padlen)
 
 
-@conditionaljit()
 def dobpfftfilt(Fs, lowerpass, upperpass, inputdata, padlen=20, cyclic=False, debug=False):
     r"""Performs an FFT brickwall bandpass filter on an input vector
     and returns the result.  Ends are padded to reduce transients.
@@ -555,7 +538,8 @@ def dobpfftfilt(Fs, lowerpass, upperpass, inputdata, padlen=20, cyclic=False, de
 
 
 # - fft trapezoidal filters
-@conditionaljit()
+
+
 def getlptrapfftfunc(Fs, upperpass, upperstop, inputdata, debug=False):
     r"""Generates a trapezoidal lowpass transfer function.
 
@@ -608,7 +592,6 @@ def getlptrapfftfunc(Fs, upperpass, upperstop, inputdata, debug=False):
     return transferfunc
 
 
-@conditionaljit()
 def getlptransfunc(Fs, inputdata, upperpass=None, upperstop=None, type="brickwall", debug=False):
     if upperpass is None:
         print("getlptransfunc: upperpass must be specified")
@@ -693,7 +676,6 @@ def gethptransfunc(Fs, inputdata, lowerstop=None, lowerpass=None, type="brickwal
     return transferfunc
 
 
-@conditionaljit()
 def dolptransfuncfilt(
     Fs,
     inputdata,
@@ -757,7 +739,6 @@ def dolptransfuncfilt(
     return unpadvec(fftpack.ifft(inputdata_trans).real, padlen=padlen)
 
 
-@conditionaljit()
 def dohptransfuncfilt(
     Fs,
     inputdata,
@@ -827,7 +808,6 @@ def dohptransfuncfilt(
     return unpadvec(fftpack.ifft(inputdata_trans).real, padlen=padlen)
 
 
-@conditionaljit()
 def dobptransfuncfilt(
     Fs,
     inputdata,
@@ -908,7 +888,6 @@ def dobptransfuncfilt(
     return unpadvec(fftpack.ifft(inputdata_trans).real, padlen=padlen)
 
 
-@conditionaljit()
 def dolptrapfftfilt(Fs, upperpass, upperstop, inputdata, padlen=20, cyclic=False, debug=False):
     r"""Performs an FFT filter with a trapezoidal lowpass transfer
     function on an input vector and returns the result.  Ends are padded to reduce transients.
@@ -955,7 +934,6 @@ def dolptrapfftfilt(Fs, upperpass, upperstop, inputdata, padlen=20, cyclic=False
     return unpadvec(fftpack.ifft(inputdata_trans).real, padlen=padlen)
 
 
-@conditionaljit()
 def dohptrapfftfilt(Fs, lowerstop, lowerpass, inputdata, padlen=20, cyclic=False, debug=False):
     r"""Performs an FFT filter with a trapezoidal highpass transfer
     function on an input vector and returns the result.  Ends are padded to reduce transients.
@@ -1002,7 +980,6 @@ def dohptrapfftfilt(Fs, lowerstop, lowerpass, inputdata, padlen=20, cyclic=False
     return unpadvec(fftpack.ifft(inputdata_trans).real, padlen=padlen)
 
 
-@conditionaljit()
 def dobptrapfftfilt(
     Fs,
     lowerstop,
@@ -1293,7 +1270,6 @@ def csdfilter(obsdata, commondata, padlen=20, cyclic=False, debug=False):
     return unpadvec(fftpack.ifft(obsdata_trans).real, padlen=padlen)
 
 
-@conditionaljit()
 def arb_pass(
     Fs,
     inputdata,

capcalc/fit.py

@@ -23,22 +23,15 @@
 
 import matplotlib.pyplot as plt
 import numpy as np
-import pyfftw
 import scipy as sp
 import scipy.special as sps
-
-# from numba import jit
 from scipy.signal import find_peaks, hilbert
 
 import capcalc.util as ccalc_util
 
-fftpack = pyfftw.interfaces.scipy_fftpack
-pyfftw.interfaces.cache.enable()
-
 # ---------------------------------------- Global constants -------------------------------------------
 defaultbutterorder = 6
 MAXLINES = 10000000
-donotbeaggressive = True
 
 # ----------------------------------------- Conditional imports ---------------------------------------
 try:
@@ -48,31 +41,6 @@
 except ImportError:
     memprofilerexists = False
 
-donotusenumba = True
-
-
-def conditionaljit():
-    def resdec(f):
-        if donotusenumba:
-            return f
-        return jit(f, nopython=False)
-
-    return resdec
-
-
-def conditionaljit2():
-    def resdec(f):
-        if donotusenumba or donotbeaggressive:
-            return f
-        return jit(f, nopython=False)
-
-    return resdec
-
-
-def disablenumba():
-    global donotusenumba
-    donotusenumba = True
-
 
 # --------------------------- Fitting functions -------------------------------------------------
 def gaussresidualssk(p, y, x):
@@ -108,7 +76,6 @@ def gaussskresiduals(p, y, x):
     return y - gausssk_eval(x, p)
 
 
-@conditionaljit()
 def gaussresiduals(p, y, x):
     """
 
@@ -174,7 +141,6 @@ def gausssk_eval(x, p):
     return p[0] * sp.stats.norm.pdf(t) * sp.stats.norm.cdf(p[3] * t)
 
 
-@conditionaljit()
 def kaiserbessel_eval(x, p):
     """
 
@@ -201,7 +167,6 @@ def kaiserbessel_eval(x, p):
     )
 
 
-@conditionaljit()
 def gauss_eval(x, p):
     """
 
@@ -254,7 +219,6 @@ def risetime_eval_loop(x, p):
     return r
 
 
-@conditionaljit()
 def trapezoid_eval(x, toplength, p):
     """
 
@@ -277,7 +241,6 @@ def trapezoid_eval(x, toplength, p):
         return p[1] * (np.exp(-(corrx - toplength) / p[3]))
 
 
-@conditionaljit()
 def risetime_eval(x, p):
     """
 
@@ -367,7 +330,8 @@ def locpeak(data, samplerate, lastpeaktime, winsizeinsecs=5.0, thresh=0.75, hyst
 
 
 # generate the polynomial fit timecourse from the coefficients
-@conditionaljit()
+
+
 def trendgen(thexvals, thefitcoffs, demean):
     """
 
@@ -416,7 +380,6 @@ def detrend(inputdata, order=1, demean=False):
     return inputdata - thefittc
 
 
-@conditionaljit()
 def findfirstabove(theyvals, thevalue):
     """
 
@@ -640,7 +603,6 @@ def territorydecomp(
     return fitmap, thecoffs, theRs
 
 
-@conditionaljit()
 def refinepeak_quad(x, y, peakindex, stride=1):
     # first make sure this actually is a peak
     ismax = None
@@ -670,7 +632,6 @@ def refinepeak_quad(x, y, peakindex, stride=1):
     return peakloc, peakval, peakwidth, ismax, badfit
 
 
-@conditionaljit2()
 def findmaxlag_gauss(
     thexcorr_x,
     thexcorr_y,
@@ -924,7 +885,6 @@ def findmaxlag_gauss(
     return maxindex, maxlag, maxval, maxsigma, maskval, failreason, fitstart, fitend
 
 
-@conditionaljit2()
 def maxindex_noedge(thexcorr_x, thexcorr_y, bipolar=False):
     """
 
@@ -965,8 +925,6 @@ def maxindex_noedge(thexcorr_x, thexcorr_y, bipolar=False):
     return maxindex, flipfac
 
 
-# disabled conditionaljit on 11/8/16.  This causes crashes on some machines (but not mine, strangely enough)
-@conditionaljit2()
 def findmaxlag_gauss_rev(
     thexcorr_x,
     thexcorr_y,
@@ -1266,7 +1224,6 @@ def findmaxlag_gauss_rev(
     )
 
 
-@conditionaljit2()
 def findmaxlag_quad(
     thexcorr_x,
     thexcorr_y,