fix: fixing testing infrastructure fail (#74)

* fix: fixing testing infrastructure fail including tests

* feat: working matrix inverse example (not integrated)

Author: jcabrero

examples/linear_regression/nada-project.toml

@@ -2,6 +2,9 @@ name = "linear-regression"
 version = "0.1.0"
 authors = [""]
 
+[test_framework.nada-test]
+command = "nada-test ./tests"
+
 [[programs]]
 path = "src/linear_regression.py"
 prime_size = 64

examples/linear_regression/src/determinant.py

@@ -38,7 +38,7 @@ def nada_main():
     parties = na.parties(3)
 
     X = na.array([3, 3], parties[0], "A", SecretInteger)
-    X = X.reveal()
+    X = X.to_public()
     detX = determinant(X)
 
     return na.output(detX, parties[2], "my_output")

examples/linear_regression/src/gauss_jordan.py

@@ -51,7 +51,7 @@ def nada_main():
 
     A = na.array([3, 3], parties[0], "A", nada_type=SecretInteger)
 
-    A = A.reveal()
+    A = A.to_public()
     A_inv = gauss_jordan_zn(A, PRIME)
     outputs = na.output(A_inv, parties[2], "my_output")
 

examples/linear_regression/src/linear_regression.py

@@ -91,7 +91,7 @@ def linsol(A: NadaArray, b: NadaArray, modulo: int):
     )  # (n, n) random matrix R with inverse determinant detR_inv
 
     # Revealing matrix RA
-    RA = (R @ A).reveal()  # (n, n) revealed matrix
+    RA = (R @ A).to_public()  # (n, n) revealed matrix
 
     # Computing Rb as a secret matrix multiplication of R and b
     Rb = R @ b  # (n, n) @ (n,) = (n,)

examples/linear_regression/src/linear_regression_256.py

@@ -56,7 +56,7 @@ def private_modular_inverse(secret: SecretInteger, modulo: int) -> SecretInteger
     r = SecretInteger.random()
 
     ra = r * secret  # Masking our secret
-    ra_revealed = ra.reveal()  # Revealing the masked secret
+    ra_revealed = ra.to_public()  # Revealing the masked secret
 
     ra_inv = public_modular_inverse(
         ra_revealed, modulo
@@ -144,7 +144,7 @@ def matrix_inverse(matrix: np.ndarray, modulo: int):
     n = matrix.shape[0]
     R, detR = random_lu_matrix(n)  # n by n random matrix R with determinant detR
     # Revealing matrix RA
-    RA = (R @ matrix).reveal()
+    RA = (R @ matrix).to_public()
     # # Concatenating RA and R
     RAR = RA.hstack(R)
     # Performing Gauss-Jordan elimination
@@ -241,7 +241,7 @@ def linsol(A: NadaArray, b: NadaArray, modulo: int):
     )  # (n, n) random matrix R with inverse determinant detR_inv
 
     # Revealing matrix RA
-    RA = (R @ A).reveal()  # (n, n) revealed matrix
+    RA = (R @ A).to_public()  # (n, n) revealed matrix
 
     # Computing Rb as a secret matrix multiplication of R and b
     Rb = R @ b  # (n, n) @ (n,) = (n,)

examples/linear_regression/src/matrix_inverse.py

@@ -40,7 +40,7 @@ def public_modular_inverse(
     power = value ** Integer(
         mod - 1
     )  # value ** modulo = value ** (modulo // 2)  * modulo ** (modulo // 2)
-    power = power * power * value if rem else Integer(1)  # value ** mo
+    power = power * power * (value if rem else Integer(1))  # value ** mo
     return power
 
 
@@ -133,14 +133,7 @@ def gauss_jordan_zn(mat: na.NadaArray, modulo: int):
 
     # Forward elimination
     for i in range(rows):
-        # # Find pivot row
-        # pivot_row = i
-        # while pivot_row < rows and (mat[pivot_row][i] == Integer(0)) is Boolean(True):
-        #     pivot_row += 1
-
-        # # Swap pivot row with current row
-        # mat[[i, pivot_row]] = mat[[pivot_row, i]]
-
+        
         # Scale pivot row to have leading 1
         diagonal_element = mat[i][i]
         pivot_inv = public_modular_inverse(diagonal_element, modulo)
@@ -168,7 +161,7 @@ def matrix_inverse(matrix: np.ndarray, modulo: int):
     R, detR = random_lu_matrix(n)  # n by n random matrix R with determinant detR
 
     # Revealing matrix RA
-    RA = (R @ matrix).reveal()
+    RA = (R @ matrix).to_public()
     # # Concatenating RA and R
     RAR = RA.hstack(R)
     # Performing Gauss-Jordan elimination
@@ -186,7 +179,7 @@ def matrix_inverse(matrix: np.ndarray, modulo: int):
 def nada_main():
     parties = na.parties(3)
 
-    A = na.array([4, 4], parties[0], "A", nada_type=SecretInteger)
+    A = na.array([10, 10], parties[0], "A", nada_type=SecretInteger)
     A_inv = matrix_inverse(A, PRIME)
 
     result = A @ A_inv

examples/linear_regression/src/modular_inverse.py

@@ -55,7 +55,7 @@ def private_modular_inverse(secret: SecretInteger, modulo: int) -> SecretInteger
     r = SecretInteger.random()
 
     ra = r * secret  # Masking our secret
-    ra_revealed = ra.reveal()  # Revealing the masked secret
+    ra_revealed = ra.to_public()  # Revealing the masked secret
 
     ra_inv = public_modular_inverse(
         ra_revealed, modulo

examples/linear_regression/tests/determinant-3.yaml renamed to examples/linear_regression/tests/determinant_3.yaml

@@ -0,0 +1,24 @@
+from nada_test import nada_test, NadaTest
+import sys
+import nada_numpy.client as  na
+import numpy as np
+
+# Functional style test
+@nada_test(program="matrix_inverse")
+def my_test():
+    n = 10
+    m = np.random.rand(n, n)
+    mx = np.sum(np.abs(m), axis=1)
+    np.fill_diagonal(m, mx)
+    A = na.array(m * (1 << 16), "A", nada_type=int)
+    print("INPUTS:", A, file=sys.stderr)
+    outputs = yield A
+    print(outputs, file=sys.stderr)
+    for output, value in outputs.items():
+        output = output.split("_")
+        if output[-1] == output[-2]:
+            assert value == 1, f"Expected 1 {output}, got {value}"
+        else:
+            assert value == 0, f"Expected 0 {output}, got {value}"
+    
+    #assert outputs["my_output"] == a + b