MATLAB to NumPy/SciPy Cheat Sheet: Every Common Function Mapped

This is a practical reference for engineers migrating MATLAB code to Python. Every entry shows the MATLAB syntax on the left and the exact Python equivalent on the right, including the required import.

Bookmark this page. You will come back to it.

For automated conversion of entire files, paste your MATLAB code into our converter — it applies all of these mappings automatically and flags anything that needs manual review.

zeros(3,4) → np.zeros((3, 4)) ones(3,4) → np.ones((3, 4)) eye(5) → np.eye(5) rand(3,4) → np.random.rand(3, 4) randn(3,4) → np.random.randn(3, 4) linspace(0,1,100) → np.linspace(0, 1, 100) logspace(1,3,50) → np.logspace(1, 3, 50) meshgrid(x,y) → np.meshgrid(x, y) diag(v) → np.diag(v) repmat(A,m,n) → np.tile(A, (m, n)) [1 2 3; 4 5 6] → np.array([[1, 2, 3], [4, 5, 6]]) 1:10 → np.arange(1, 11) (note: Python excludes the end value) 1:0.5:10 → np.arange(1, 10.5, 0.5)

All of these require import numpy as np.

size(A) → A.shape size(A,1) → A.shape[0] (note: 0-indexed in Python) size(A,2) → A.shape[1] length(A) → max(A.shape) or len(A) for 1D numel(A) → A.size ndims(A) → A.ndim isempty(A) → len(A) == 0 or A.size == 0

The #1 gotcha: MATLAB is 1-based, Python is 0-based.

A(1) → A[0] (first element) A(end) → A[-1] (last element) A(end-1) → A[-2] A(2:5) → A[1:5] (elements 2 through 5) A(1:end-1) → A[:-1] (all but last) A(:) → A.flatten() A(i,j) → A[i-1, j-1] A(:,j) → A[:, j-1] A(i,:) → A[i-1, :] A(A>5) → A[A > 5] (logical indexing — no shift needed) A(mask) → A[mask]

Element-wise operations: A .* B → A * B A ./ B → A / B A .^ 2 → A ** 2

Matrix operations: A * B (matrix multiply) → A @ B A' (conjugate transpose) → A.conj().T or A.T for real A.' (transpose) → A.T A \ b (solve Ax=b) → np.linalg.solve(A, b)

Comparison and logical: ~= → != && → and || → or & → & (element-wise) | → | (element-wise) ~ → not or ~ (for arrays)

abs(x) → np.abs(x) sqrt(x) → np.sqrt(x) exp(x) → np.exp(x) log(x) → np.log(x) (natural log) log2(x) → np.log2(x) log10(x) → np.log10(x) sin(x) → np.sin(x) cos(x) → np.cos(x) tan(x) → np.tan(x) asin(x) → np.arcsin(x) atan2(y,x) → np.arctan2(y, x) ceil(x) → np.ceil(x) floor(x) → np.floor(x) round(x) → np.round(x) mod(a,b) → np.mod(a, b) or a % b sign(x) → np.sign(x) real(z) → np.real(z) imag(z) → np.imag(z) conj(z) → np.conj(z) angle(z) → np.angle(z)

mean(x) → np.mean(x) median(x) → np.median(x) std(x) → np.std(x, ddof=1) (ddof=1 to match MATLAB N-1) var(x) → np.var(x, ddof=1) max(x) → np.max(x) min(x) → np.min(x) [val,idx] = max(x) → val = np.max(x); idx = np.argmax(x) sum(x) → np.sum(x) prod(x) → np.prod(x) cumsum(x) → np.cumsum(x) cumprod(x) → np.cumprod(x) sort(x) → np.sort(x) [sorted,idx] = sort(x) → idx = np.argsort(x); sorted = x[idx] unique(x) → np.unique(x) histc(x,edges) → np.histogram(x, edges) prctile(x,p) → np.percentile(x, p) corrcoef(x) → np.corrcoef(x.T) (note: row convention differs) cov(x) → np.cov(x.T) (note: row convention differs)

inv(A) → np.linalg.inv(A) pinv(A) → np.linalg.pinv(A) det(A) → np.linalg.det(A) rank(A) → np.linalg.matrix_rank(A) norm(A) → np.linalg.norm(A) trace(A) → np.trace(A) [V,D] = eig(A) → D, V = np.linalg.eig(A) (note: unsorted in NumPy) [U,S,V] = svd(A) → U, S, V = np.linalg.svd(A) [Q,R] = qr(A) → Q, R = np.linalg.qr(A) [L,U,P] = lu(A) → P, L, U = scipy.linalg.lu(A) (note: different return order) chol(A) → np.linalg.cholesky(A) cond(A) → np.linalg.cond(A) kron(A,B) → np.kron(A, B) cross(a,b) → np.cross(a, b) dot(a,b) → np.dot(a, b)

Requires import numpy as np and import scipy.linalg for lu.

fft(x) → np.fft.fft(x) ifft(x) → np.fft.ifft(x) fft2(x) → np.fft.fft2(x) fftshift(x) → np.fft.fftshift(x)

figure → plt.figure() plot(x,y) → plt.plot(x, y) subplot(m,n,p) → plt.subplot(m, n, p) title('text') → plt.title('text') xlabel('text') → plt.xlabel('text') ylabel('text') → plt.ylabel('text') legend('a','b') → plt.legend(['a', 'b']) grid on → plt.grid(True) hold on → not needed (matplotlib accumulates by default) axis equal → plt.axis('equal') close all → plt.close('all') scatter(x,y) → plt.scatter(x, y) histogram(x,30) → plt.hist(x, 30) bar(x,y) → plt.bar(x, y) stem(x,y) → plt.stem(x, y) semilogx(x,y) → plt.semilogx(x, y) loglog(x,y) → plt.loglog(x, y) colorbar → plt.colorbar() savefig('file.png') → plt.savefig('file.png')

Named arguments: 'LineWidth', 2 → linewidth=2 'MarkerSize', 8 → markersize=8 'FontSize', 14 → fontsize=14 'Location', 'Best' → loc='best'

Requires import matplotlib.pyplot as plt.

strcmp(a,b) → a == b strcmpi(a,b) → a.lower() == b.lower() strcat(a,b) → a + b num2str(x) → str(x) str2num(s) → float(s) sprintf('format',x) → 'format' % (x,) or f'...{x}' strsplit(s,',') → s.split(',') strtrim(s) → s.strip() upper(s) → s.upper() lower(s) → s.lower() contains(s,'sub') → 'sub' in s startsWith(s,'pre') → s.startswith('pre') strrep(s,'old','new') → s.replace('old', 'new')

load('file.mat') → scipy.io.loadmat('file.mat') save('file.mat','var') → scipy.io.savemat('file.mat', {'var': var}) csvread('file.csv') → np.loadtxt('file.csv', delimiter=',') csvwrite('file.csv',A) → np.savetxt('file.csv', A, delimiter=',') xlsread('file.xlsx') → pd.read_excel('file.xlsx') fopen(name,'w') → open(name, 'w') fclose(fid) → fid.close() fprintf(fid,'format',x) → fid.write('format' % (x,)) fullfile('a','b','c') → os.path.join('a', 'b', 'c') exist('file') → os.path.exists('file')

Requires import scipy.io, import pandas as pd, or import os as needed.

for i = 1:10 ... end → for i in range(10): for i = 5:20 ... end → for i in range(5, 21): for i = 1:2:10 ... end → for i in range(1, 11, 2): while cond ... end → while cond: if ... elseif ... else ... end → if ... elif ... else: switch x / case val → if x == val: / elif x == val2: try ... catch ME ... end → try: ... except Exception as ME: break → break continue → continue return → return

pi → np.pi NaN → np.nan Inf → np.inf eps → np.finfo(float).eps true → True false → False 1i → 1j (imaginary unit)

This cheat sheet covers the most common mappings. For a complete file conversion with automatic import injection, index shifting, toolbox detection, and a compatibility report, paste your MATLAB code into the converter. Free for 50 lines.