blindscrambler 0.1.0__cp39-abi3-macosx_11_0_arm64.whl → 0.1.2__cp39-abi3-macosx_11_0_arm64.whl

blindscrambler/differential/__init__.py

@@ -1,2 +1,2 @@
 from .discrete import diff
-__all__ = ['diff']
+__all__ = ['diff']

blindscrambler/distributions/__init__.py

@@ -0,0 +1,2 @@
+from .cvdistributions import uniform
+__all__ = ["uniform"]

blindscrambler/distributions/cvdistributions.py

@@ -0,0 +1,15 @@
+# Metadata:
+# Author: Syed Raza
+# email: sar0033@uah.edu
+
+# use python secrets to make a normally distributed random samples
+import secrets
+
+def uniform(a: float = 0.0, b: float = 1.0) -> float:
+    """
+    Cryptographically secure sample
+    """
+    # 53 random bits gives 53-bit precision double
+    u = secrets.randbits(53) / (1 << 53) # in [0, 1)
+
+    return a + (b - a) * u

blindscrambler/matrix/__init__.py

@@ -0,0 +1,2 @@
+from .elementary import float_equality, rowswap, rowscale, rowreplacement, rref
+__all__ = ["float_equality", "rowswap", "rowscale", "rowreplacement", "rref"]

blindscrambler/matrix/elementary.py

@@ -0,0 +1,138 @@
+# Metadata:
+# Author: Syed Raza
+# email: sar0033@uah.edu
+
+# import statements
+import torch # type: ignore
+tol = 1e-7
+
+# This is the program to perform matrix operations on pytorch based arrays/matrices
+
+def float_equality(n1, n2):
+    """
+    Params:
+    (1) n1 -- type torch.float32
+        the first number for equality check
+    (2) n2 -- type torch.float32
+        the second number for equality check
+    (3) tol -- the tolerance for float equality
+
+    Returns:
+        true if they are equal
+        False if they are not 
+    """
+    return abs(n1 - n2) < tol
+
+
+def rowswap(M, src, trg):
+    """
+    Params: (1) A Pytorch based matrix -- M, 
+        (2) index of type int the source row -- src, 
+        (3) Index of type int target row -- trg
+
+    Returns: Returns the matrix after swapping the content of source row with the target row
+    """
+    # swap the rows you wanted:
+    M[[src, trg]] = M[[trg, src]]
+
+    return M
+
+def rowscale(M, src, sca):
+    """
+    Params:
+    (1) A Pytorch based matrix -- M
+    (2) index (type int) of the row to be scaled -- src
+    (2) scaling factor of type int -- sca
+    """
+    M[src] *= sca
+
+    return M
+
+def rowreplacement(M, r1, r2, sca1, sca2):
+    """
+    Params: 
+    (1) M -- type Pytorch tensor
+        This is the matrix where we are applying the row replacement
+    (2) r1 -- type int 
+        The index of the first row
+    (3) r2 -- type int 
+        The index of the second row 
+    (4) sca1 -- type int
+        the scaling factor row 1
+    (5) sca2 -- type int 
+        the scalikng factor row 2
+    
+    Returns:
+    The matrix but with finished row replacement
+    """
+    # clone the matrix, so we do not change the matrix in place
+    M1 = M.clone()
+
+    # scale both rows in the clone
+    M1 = rowscale(M1, r1, sca1)
+    M1 = rowscale(M1, r2, sca2)
+
+    # perform row-replacement
+    M[r1] = M1[r1] + M1[r2]
+
+    # return M
+    return M
+
+# A routine for converting matrices into reduced row echelon form:
+def rref(M):
+    """
+    Params: M -- type Pytorch tensor
+        This is the matrix where we are trying to convert into Row Echelon form
+    Returns: M -- type Pytorch tensor
+        The row echelon version of the matix the was inputed. It used the functions generated 
+        in this module: float inequality, rowswap, rowreplacement, and rowscale
+    """
+    # get the rows and the columns
+    nrow = M.shape[0]
+    ncol = M.shape[1]
+    
+    # The workflow is given as follows:
+
+    row = 0
+    for col in range(ncol):
+        if row >= nrow:
+            break
+        # (a) if there is a zero in the row, rowswap with a new row
+        if float_equality(M[row, col], 0): 
+            # go through rest of the rows
+            for k in range(row, nrow):
+                if not float_equality(M[k, col], 0):
+                    rowswap(M, k, row) # this is the row that includes the pivot
+                    break # break rowswapping first time this happens
+        
+        # (b) use row scale to make the pivot element 1 
+        if not float_equality(M[row, col], 0):
+            rowscale(M, row, 1 / M[row, col])
+
+        # (c) make everything below zero using row_replacement function
+        for r in range(row + 1, nrow):
+            if not float_equality(M[r, col], 0):
+                mul = - M[r, col] / M[row, col]
+                rowreplacement(M, r, row, 1.0, mul)
+
+        row += 1
+
+    # scale the last row separately:
+    last = nrow - 1
+    for col in range(ncol):
+        if not float_equality(M[last, col], 0.0):
+            rowscale(M, last, 1 / M[last, col])
+
+    # make really smal numbers resulting due to floating point operations zero
+    # before returning the matrix
+    M[torch.abs(M) < 1e-5] = 0.0
+
+    return M
+
+# main function for testing the code and analyzing the reuslts:
+if __name__ == "__main__":
+    M = torch.tensor([[0, 3, -6, 6, 4], [3, -7, 8, -5, 8], [3, -9, 12, -9, 6]], dtype = torch.float32)
+
+    # testing phase of the rref function:
+    M = rref(M)
+    print(M)

{blindscrambler-0.1.0.dist-info → blindscrambler-0.1.2.dist-info}/METADATA

@@ -1,6 +1,8 @@
 Metadata-Version: 2.4
 Name: blindscrambler
-Version: 0.1.0
+Version: 0.1.2
+Requires-Dist: numpy>=2.3.3
+Requires-Dist: torch>=2.8.0
 Requires-Dist: twine>=6.1.0
 Summary: Add your description here
 Author-email: blindscramblergh <blindscrambler@gmail.com>

blindscrambler-0.1.2.dist-info/RECORD

@@ -0,0 +1,13 @@
+blindscrambler-0.1.2.dist-info/METADATA,sha256=h_1FCVyxGU3DcB6M7H66MnCHTj8zvzimG2Se37yAtl4,373
+blindscrambler-0.1.2.dist-info/WHEEL,sha256=DLqF2HZq4W_umZdP6RnfAuqhmtX_UrV4mkqrSIMhipE,102
+blindscrambler/__init__.py,sha256=N6o-PTyGSlQ4ny1UA4ByeNenVF-wCTALnyP4WJ8PGas,98
+blindscrambler/_core.abi3.so,sha256=4uKUtCwAO1Hbvzv0FXAt38rEHYbg-Quio8CdkJ_UMrk,440112
+blindscrambler/_core.pyi,sha256=b6oJaUXUzEzqUE5rpqefV06hl8o_JCU8pgKgIIzQgmc,33
+blindscrambler/differential/__init__.py,sha256=INnk5rX2ae6mG5yynAQYKzpQ0BYsHquUhA9ZzbPVLm8,45
+blindscrambler/differential/discrete.py,sha256=mPJg6YrDVuXK-dLXgb_VDqKl1IvKfSKahMA_rRTVKQY,369
+blindscrambler/distributions/__init__.py,sha256=8O4VQvymecRFRP1njwAfbD4yUACA25RcLqn0QtZEjaE,58
+blindscrambler/distributions/cvdistributions.py,sha256=_IgaPSbN5emA9J8daF9l12r1rsDKrscD8jLERuYWKY4,378
+blindscrambler/matrix/__init__.py,sha256=qlItVU8AVj_mP2NUJ3gor-lsovxk3Wxf5tUfKynoUbg,157
+blindscrambler/matrix/elementary.py,sha256=hArZLiBTA_vW1EZ0RniECf6ybJiJxO7KNuVHb_TZFQU,3987
+blindscrambler/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+blindscrambler-0.1.2.dist-info/RECORD,,

blindscrambler-0.1.0.dist-info/RECORD

@@ -1,9 +0,0 @@
-blindscrambler-0.1.0.dist-info/METADATA,sha256=KxZRLgUTtGWxx_i4NS1ibY__4Crkn9geJzY_2eAIj5E,317
-blindscrambler-0.1.0.dist-info/WHEEL,sha256=DLqF2HZq4W_umZdP6RnfAuqhmtX_UrV4mkqrSIMhipE,102
-blindscrambler/__init__.py,sha256=N6o-PTyGSlQ4ny1UA4ByeNenVF-wCTALnyP4WJ8PGas,98
-blindscrambler/_core.abi3.so,sha256=OG585R2Nb-fp_sH7C6qCpxzKDFG6hKWqIkwWvB6n0II,440336
-blindscrambler/_core.pyi,sha256=b6oJaUXUzEzqUE5rpqefV06hl8o_JCU8pgKgIIzQgmc,33
-blindscrambler/differential/__init__.py,sha256=wthVEgnKH3fg_e8fjcWu5DRIJsy6LSAfd8Jntn-JiJ4,46
-blindscrambler/differential/discrete.py,sha256=mPJg6YrDVuXK-dLXgb_VDqKl1IvKfSKahMA_rRTVKQY,369
-blindscrambler/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-blindscrambler-0.1.0.dist-info/RECORD,,