Skip to content

Commit

Permalink
Improved performance of intermod table
Browse files Browse the repository at this point in the history
  • Loading branch information
Will Boxx committed Aug 28, 2023
1 parent 7e80f91 commit 2f24a85
Showing 1 changed file with 49 additions and 131 deletions.
180 changes: 49 additions & 131 deletions src/intermod_library/intermod_library.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -37,150 +37,73 @@ def intermod_table(signals, order, maximum_single_order=None, bandwidths=None):
>>> table = il.intermod_table(signals, order)
>>> table.head()
========= ========= ========= ========= =========
<index> Frequency Signal 1 Signal 2 Order
========= ========= ========= ========= =========
0 1000.0 1.0 0.0 1.0
1 1000.0 -1.0 1.0 2.0
2 2000.0 0.0 1.0 1.0
3 2000.0 2.0 0.0 2.0
4 3000.0 1.0 1.0 2.0
========= ========= ========= ========= =========
============ ============ ============ ============ ============
<index> Frequency coefficients tx_indexes Order
============ ============ ============ ============ ============
0 1000.0 (-1, 1) (0, 1) 1
1 3000.0 (-1, 2) (0, 1) 2
2 3000.0 (1, 1) (0, 1) 1
3 4000.0 (2, 1) (0, 1) 2
4 5000.0 (1, 2) (0, 1) 2
============ ============ ============ ============ ============
"""

M = np.size(signals) # Number of signals
# N = order + 1

# A = np.arange(0, N)

# coefmat = np.zeros([N**M, M])

# ind = 0

# for i in range(M, 0, -1):
# m = N**(M-i)
# B = np.ones([N**(i-1), m])
# C = np.kron(B, A)
# coefmat[:, ind] = C.T.ravel()
# ind += 1

# B = np.ones(2**M)
# coefmat = np.reshape(np.kron(B, coefmat), [-1, M])
if M < 2:
print("Number of signals must be 2 or greater.")
return None

# # Make sign array
# A = np.array([1, -1])
# signmat = np.zeros([2**M, M])
# ind = 0

# for i in range(M, 0, -1):
# m = 2**(M-i)
# B = np.ones([2**(i - 1), m])
# C = np.kron(B, A)
# signmat[:, ind] = C.T.ravel()
# ind += 1
columns = ["frequency", "coefficients", "tx_indexes", "intermod_order"]
T = pd.DataFrame(columns=columns)

# j = np.size(coefmat)/M
# firstblock = signmat
coefficients = [x for x in range(-order+1, order) if x != 0]

# for i in range(1, int(j / 2**M)):
# signmat = np.vstack([signmat, firstblock])
num_of_sigs_to_combine = [2, 3] if M > 2 else [2]

# finalmat = coefmat * signmat
for i in num_of_sigs_to_combine:
cart_prod = itertools.product(coefficients, repeat=i)
coef_list = list(
filter(
lambda x: np.sum([abs(y) for y in np.array(x)]) <= order, cart_prod
)
)
coef_array = np.array(coef_list)
coef_tuple_array = pd.Index(coef_list).values
idx_combs = pd.Index(list(itertools.combinations(range(M), i))).values

if maximum_single_order is None:
mso = order
else:
mso = maximum_single_order
if i == 2:
sigs = [[signals[x], signals[y]] for x, y in idx_combs]
intermod_order = [np.sum(abs(np.array(z))) for z in coef_array]
else:
sigs = [[signals[x], signals[y], signals[z]] for x, y, z in idx_combs]
intermod_order = [np.sum(abs(np.array(x))) for x in coef_array]

# finalmat = np.array(
# list(
# itertools.product(range(-1*mso, mso + 1, 1), repeat=M)
# )
# )
intermods = np.dot(coef_array, np.array(sigs, dtype=float).T).flatten(order="C")

intermods = np.empty
tx_indexes = np.empty
coefficients = np.empty
num_repeat = len(idx_combs)
num_tile = len(coef_tuple_array)

columns = ["Frequency", "coefficients", "tx_indexes", "intermod_order"]
T = pd.DataFrame(columns=columns)
final_coefs = np.repeat(coef_tuple_array, num_repeat)
final_idxs = np.tile(idx_combs, num_tile)

coefficients = [x for x in range(-order+1, order) if x != 0]
final_order = np.repeat(intermod_order, num_repeat)

for i in range(2, order+1):
cart_prod = list(itertools.product(coefficients, repeat=i))
filtered = list(filter(lambda x: np.sum([abs(y) for y in np.array(x)]) <= order, cart_prod))
indxs = range(len(signals))
idx_combs = list(itertools.combinations(indxs, i))

intermods = []
final_sigs = []
final_idxs = []
final_coefs = []
final_order = []
for idxs in idx_combs:
for coefs in filtered:
sigs = [signals[x] for x in idxs]
intermods = [*intermods, np.dot(np.array(sigs), np.array(coefs))]
final_sigs = [*final_sigs, sigs]
final_idxs = [*final_idxs, idxs]
final_coefs = [*final_coefs, coefs]
final_order = [*final_order, np.sum([abs(x) for x in coefs])]

df = pd.DataFrame({"Frequency": intermods, "tx_indexes": final_idxs, "coefficients": final_coefs, "intermod_order": final_order})
df = pd.DataFrame(
{
"frequency": intermods,
"tx_indexes": final_idxs,
"coefficients": final_coefs,
"intermod_order": final_order
}
)
T = pd.concat([T, df], ignore_index=True)


# combs = list(itertools.combinations(range(len(freqs)), order))
# for comb in combs:
# fun = lambda x: (sum(map(abs,x)) <= order) & (abs(sum(x)) > 1) & (np.count_nonzero(x) > 1)
# coef_mat = list(filter(fun, itertools.product(range(-1*mso, mso + 1, 1), repeat=order)))
# foi = [freqs[x] for x in comb]
# intermods = np.dot(coef_mat, foi)
# coefficients = [x[0] for x in coef_mat]
# tx_indexes = list(itertools.repeat(comb, len(coef_mat)))
# intermod_order = np.sum(abs(np.array(coef_mat)), 1)
# # final = np.column_stack((intermods, coefficients, tx_indexes, intermod_order))
# final = {"Frequency": intermods, "tx_indexes": tx_indexes, "coefficients": coef_mat, "intermod_order": intermod_order}
# test = pd.DataFrame(final)
# df = pd.concat([df, test], ignore_index=True)

print(T)

T.query('Frequency > 0', inplace=True)
# df.query('(Order > 0) & (Order <= @order)', inplace=True)
T.query('frequency > 0.01', inplace=True)
T.drop_duplicates(inplace=True)
T.sort_values(by=['Frequency'], inplace=True)
T.sort_values(by=['frequency'], inplace=True)
T.reset_index(drop=True, inplace=True)

# print(df.head)
# finalmat = list(itertools.product(range(-1*mso, mso + 1, 1), repeat=M))

# intermods = np.dot(finalmat, signals)
# intermod_order = np.sum(abs(finalmat), 1)
# final = np.column_stack((intermods, finalmat, intermod_order))

# header = ['Frequency']
# for i in np.arange(M):
# header = header + ['Signal ' + str(i+1)]

# header = header + ['Order']

# T = pd.DataFrame(final, columns=header)

# T.query('Frequency > 0', inplace=True)
# T.query('(Order > 0) & (Order <= @order)', inplace=True)
# T.drop_duplicates(inplace=True)
# T.sort_values(by=['Frequency'], inplace=True)
# T.reset_index(drop=True, inplace=True)

def find_bw(row, bws):
return max(row[1:-1]*bws)

if bandwidths is not None:
T['bandwidth'] = T.apply(find_bw, 1, True, None, ([bandwidths]))

return (T)
return T


def harmonic_toi(frqs, order, band_of_interest=[]):
Expand Down Expand Up @@ -352,8 +275,3 @@ def intermod_locate(soi, pivot, order):
T.reset_index(drop=True, inplace=True)

return (T)


txrs = pd.read_csv("C:\\Users\\SCYM\\OneDrive\Documents\\Temp\cosite-study\\luke-txrs.csv")
freqs = txrs.rf
intermod_table = intermod_table(freqs, 5)

0 comments on commit 2f24a85

Please sign in to comment.