diff --git a/qsymm/json_serialization/convert_func.py b/qsymm/json_serialization/convert_func.py
new file mode 100644
index 0000000000000000000000000000000000000000..326eb23f35186db4ba68bd2a24609f76cfba7584
--- /dev/null
+++ b/qsymm/json_serialization/convert_func.py
@@ -0,0 +1,74 @@
+import numpy as np
+import sympy
+import re
+import json
+from astropy.utils.misc import JsonCustomEncoder
+import qsymm
+from icecream import ic
+
+def bloch_model_to_json(model, name):
+ name = name + '.json'
+ model_dict = {str(key) : model[key].tolist() for key in model.keys()}
+ with open(name, 'w') as f:
+ json.dump(model_dict, f, cls = JsonCustomEncoder)
+
+def bloch_model_from_json(name):
+ with open("json_models/" + name + ".json", "r") as f:
+ model = json.load(f, cls=JsonCustomDecoderExp)
+ return qsymm.Model(model, momenta=["k_x", "k_y"])
+
+
+class JsonCustomDecoderExp(json.JSONDecoder):
+ """
+ A decoder function tailored specifically to our purposes: take a dictionary of strings, output a dictionary
+ of symbolic keys (Mul types etc.) and arrays.
+ """
+
+ def __init__(self, *args, **kwargs):
+ json.JSONDecoder.__init__(self, object_hook=self.object_hook, *args, **kwargs)
+
+ def object_hook(self, obj):
+ new_dict = dict()
+ for key in obj.keys():
+ # back-convert the key from string to symbols
+ symbolic = []
+ # find exponents, works for arbitrary number of exponents
+ exp_full = re.findall('\*?e\*\*\(-?I\*k_\w\)\*?', key) # gives list with exponential terms in key
+ if exp_full:
+ coef = key
+ for exp in exp_full:
+ coef = coef.replace(exp, '') # constant coefficient
+
+ exp_power_pl = re.findall('\*?e\*\*\(I\*(k_\w)\)\*?', key) # gives list of positive exponents
+ exp_power_mn = re.findall('\*?e\*\*\(-I\*(k_\w)\)\*?', key) # gives list of negative exponents
+
+ if len(exp_power_pl)>=1:
+ sym_exp_pls = []
+ for exp in exp_power_pl:
+ sym_exp_pls.append(sympy.Pow(sympy.Symbol('e'), sympy.I*sympy.Symbol(exp)))
+ if len(exp_power_mn)==0: # no negative powers
+ symbolic_key = sympy.Mul(*sym_exp_pls, sympy.Symbol(coef))
+
+ if len(exp_power_mn)>=1:
+ sym_exp_mns = []
+ for exp in exp_power_mn:
+ sym_exp_mns.append(sympy.Pow(sympy.Symbol('e'), -sympy.I*sympy.Symbol(exp)))
+ if len(exp_power_pl)==0: # no positive powers
+ symbolic_key = sympy.Mul(*sym_exp_mns, sympy.Symbol(coef))
+ else: # both positive and negative powers
+ symbolic_key = sympy.Mul(*sym_exp_mns, *sym_exp_pls, sympy.Symbol(coef))
+ else:
+ symbolic_key = sympy.Symbol(key)
+
+ assert str(key)==str(symbolic_key) # make sure converter works
+ array = []
+ for line in obj[key]:
+ lineski = []
+ for entry in line:
+ lineski.append(entry[0] + 1j * entry[1])
+ array.append(lineski)
+ new_dict[symbolic_key] = np.array(array)
+ return new_dict
+
+
+
diff --git a/qsymm/json_serialization/minimal_model.ipynb b/qsymm/json_serialization/minimal_model.ipynb
new file mode 100644
index 0000000000000000000000000000000000000000..36dfa66aee317a8a6900f7f145f868f8932add59
--- /dev/null
+++ b/qsymm/json_serialization/minimal_model.ipynb
@@ -0,0 +1,151 @@
+{
+ "cells": [
+ {
+ "cell_type": "code",
+ "execution_count": 6,
+ "id": "ec96970a-359f-4e6f-b2d9-529749253dab",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "import numpy as np\n",
+ "import tinyarray\n",
+ "import qsymm\n",
+ "import sympy\n",
+ "from sympy import Matrix\n",
+ "import string\n",
+ "from convert_func import bloch_model_from_json, bloch_model_to_json"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 3,
+ "id": "d7832d0c-1b99-47aa-9a40-00091eda980e",
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/latex": [
+ "$\\displaystyle \\left[\\begin{matrix}0.5 e^{i k_{x}} - 0.5 i e^{i k_{y}} + 0.5 i e^{- i k_{y}} + 0.5 e^{- i k_{x}} & 0\\\\0 & - 0.5 e^{i k_{x}} - 0.5 i e^{i k_{y}} + 0.5 i e^{- i k_{y}} - 0.5 e^{- i k_{x}}\\end{matrix}\\right]$"
+ ],
+ "text/plain": [
+ "Matrix([\n",
+ "[0.5*e**(I*k_x) - 0.5*I*e**(I*k_y) + 0.5*I*e**(-I*k_y) + 0.5*e**(-I*k_x), 0],\n",
+ "[ 0, -0.5*e**(I*k_x) - 0.5*I*e**(I*k_y) + 0.5*I*e**(-I*k_y) - 0.5*e**(-I*k_x)]])"
+ ]
+ },
+ "execution_count": 3,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "sigma_x = tinyarray.array([[0, 1], [1, 0]])\n",
+ "sigma_y = tinyarray.array([[0, -1j], [1j, 0]])\n",
+ "sigma_z = tinyarray.array([[1, 0], [0, -1]])\n",
+ "\n",
+ "#create qsymm model\n",
+ "ham = \" cos(k_x) * sigma_z + sin(k_y) * eye(2) \"\n",
+ "H = qsymm.Model(ham, momenta=[\"k_x\", \"k_y\"])\n",
+ "H.tosympy()"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 4,
+ "id": "309fb778-7db3-4a02-a163-ab6479aa7711",
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "8 1\n"
+ ]
+ },
+ {
+ "data": {
+ "text/latex": [
+ "$\\displaystyle \\left[\\begin{matrix}d + e^{i k_{x}} i + \\frac{e^{i k_{x}}}{2} + e^{i k_{y}} m + i e^{i k_{y}} q - \\frac{i e^{i k_{y}}}{2} + e^{- i k_{y}} m - i e^{- i k_{y}} q + \\frac{i e^{- i k_{y}}}{2} + e^{- i k_{x}} i + \\frac{e^{- i k_{x}}}{2} & e^{i k_{x}} j + i e^{i k_{x}} l + e^{i k_{y}} n + i e^{i k_{y}} r + f + i h + e^{- i k_{y}} o - i e^{- i k_{y}} s + e^{- i k_{x}} j + i e^{- i k_{x}} l\\\\e^{i k_{x}} j - i e^{i k_{x}} l + e^{i k_{y}} o + i e^{i k_{y}} s + f - i h + e^{- i k_{y}} n - i e^{- i k_{y}} r + e^{- i k_{x}} j - i e^{- i k_{x}} l & e^{i k_{x}} k - \\frac{e^{i k_{x}}}{2} + e^{i k_{y}} p + i e^{i k_{y}} t - \\frac{i e^{i k_{y}}}{2} + g + e^{- i k_{y}} p - i e^{- i k_{y}} t + \\frac{i e^{- i k_{y}}}{2} + e^{- i k_{x}} k - \\frac{e^{- i k_{x}}}{2}\\end{matrix}\\right]$"
+ ],
+ "text/plain": [
+ "Matrix([\n",
+ "[d + e**(I*k_x)*i + e**(I*k_x)/2 + e**(I*k_y)*m + I*e**(I*k_y)*q - I*e**(I*k_y)/2 + e**(-I*k_y)*m - I*e**(-I*k_y)*q + I*e**(-I*k_y)/2 + e**(-I*k_x)*i + e**(-I*k_x)/2, e**(I*k_x)*j + I*e**(I*k_x)*l + e**(I*k_y)*n + I*e**(I*k_y)*r + f + I*h + e**(-I*k_y)*o - I*e**(-I*k_y)*s + e**(-I*k_x)*j + I*e**(-I*k_x)*l],\n",
+ "[ e**(I*k_x)*j - I*e**(I*k_x)*l + e**(I*k_y)*o + I*e**(I*k_y)*s + f - I*h + e**(-I*k_y)*n - I*e**(-I*k_y)*r + e**(-I*k_x)*j - I*e**(-I*k_x)*l, e**(I*k_x)*k - e**(I*k_x)/2 + e**(I*k_y)*p + I*e**(I*k_y)*t - I*e**(I*k_y)/2 + g + e**(-I*k_y)*p - I*e**(-I*k_y)*t + I*e**(-I*k_y)/2 + e**(-I*k_x)*k - e**(-I*k_x)/2]])"
+ ]
+ },
+ "execution_count": 4,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "#find symmetries\n",
+ "candidates = qsymm.groups.square()\n",
+ "discrete_symm, continuous_symm = qsymm.symmetries(H, candidates)\n",
+ "print(len(discrete_symm), len(continuous_symm))\n",
+ "\n",
+ "abcd_sympy = list(sympy.symbols('a:z'))\n",
+ "abcd_list = list(string.ascii_lowercase)\n",
+ "del abcd_sympy[4]\n",
+ "del abcd_list[4]\n",
+ "\n",
+ "norbs = [('A', 2)] # A atom per unit cell, 4 orbitals each\n",
+ "hopping_vectors = [('A', 'A', [0, 1]), ('A', 'A', [1, 0])] \n",
+ "\n",
+ "#find all temrs compatible with 1 specific symmetry\n",
+ "family= qsymm.bloch_family(hopping_vectors, [discrete_symm[1]], norbs)\n",
+ "\n",
+ "#add all the terms compatible with the symmetry\n",
+ "H = (H + sum([symbol * term for symbol, term in zip(abcd_sympy[3:], family)]))\n",
+ "H.tosympy(nsimplify=True)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 5,
+ "id": "387f8212-1861-4875-8fc5-798cb5d83220",
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "bloch_model_to_json(H, 'model')"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "de61fbf0-f6a6-43db-b32c-300e408fcbd3",
+ "metadata": {},
+ "outputs": [],
+ "source": []
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "id": "f5cb9c98-cc14-448d-8a03-41df556ce3d2",
+ "metadata": {},
+ "outputs": [],
+ "source": []
+ }
+ ],
+ "metadata": {
+ "kernelspec": {
+ "display_name": "Python 3 (ipykernel)",
+ "language": "python",
+ "name": "python3"
+ },
+ "language_info": {
+ "codemirror_mode": {
+ "name": "ipython",
+ "version": 3
+ },
+ "file_extension": ".py",
+ "mimetype": "text/x-python",
+ "name": "python",
+ "nbconvert_exporter": "python",
+ "pygments_lexer": "ipython3",
+ "version": "3.9.10"
+ }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/qsymm/json_serialization/model.json b/qsymm/json_serialization/model.json
new file mode 100644
index 0000000000000000000000000000000000000000..8f80422ce4459d12508738506308cc1ded303a96
--- /dev/null
+++ b/qsymm/json_serialization/model.json
@@ -0,0 +1 @@
+{"e**(-I*k_x)": [[[0.5, 0.0], [0.0, 0.0]], [[0.0, 0.0], [-0.5, 0.0]]], "e**(-I*k_y)": [[[0.0, 0.5], [0.0, 0.0]], [[0.0, 0.0], [0.0, 0.5]]], "e**(I*k_x)": [[[0.5, 0.0], [0.0, 0.0]], [[0.0, 0.0], [-0.5, 0.0]]], "e**(I*k_y)": [[[0.0, -0.5], [0.0, 0.0]], [[0.0, 0.0], [0.0, -0.5]]], "e**(-I*k_x)*l": [[[-0.0, 0.0], [0.0, 1.0]], [[0.0, -1.0], [-0.0, 0.0]]], "e**(-I*k_x)*j": [[[0.0, 0.0], [1.0, 0.0]], [[1.0, 0.0], [0.0, 0.0]]], "e**(I*k_y)*s": [[[0.0, 0.0], [0.0, 0.0]], [[0.0, 1.0], [0.0, 0.0]]], "e**(-I*k_y)*o": [[[0.0, 0.0], [1.0, 0.0]], [[0.0, 0.0], [0.0, 0.0]]], "e**(-I*k_y)*t": [[[0.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [0.0, -1.0]]], "e**(-I*k_y)*m": [[[1.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [0.0, 0.0]]], "e**(I*k_y)*o": [[[0.0, 0.0], [0.0, 0.0]], [[1.0, 0.0], [0.0, 0.0]]], "e**(I*k_y)*t": [[[0.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [0.0, 1.0]]], "e**(I*k_y)*m": [[[1.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [0.0, 0.0]]], "e**(-I*k_y)*r": [[[0.0, 0.0], [0.0, 0.0]], [[0.0, -1.0], [0.0, 0.0]]], "e**(-I*k_y)*n": [[[0.0, 0.0], [0.0, 0.0]], [[1.0, 0.0], [0.0, 0.0]]], "e**(-I*k_x)*i": [[[1.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [0.0, 0.0]]], "e**(I*k_x)*j": [[[0.0, 0.0], [1.0, 0.0]], [[1.0, 0.0], [0.0, 0.0]]], "e**(I*k_x)*l": [[[-0.0, 0.0], [0.0, 1.0]], [[0.0, -1.0], [-0.0, 0.0]]], "e**(-I*k_x)*k": [[[0.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [1.0, 0.0]]], "e**(-I*k_y)*p": [[[0.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [1.0, 0.0]]], "d": [[[1.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [0.0, 0.0]]], "f": [[[0.0, 0.0], [1.0, 0.0]], [[1.0, 0.0], [0.0, 0.0]]], "e**(I*k_y)*n": [[[0.0, 0.0], [1.0, 0.0]], [[0.0, 0.0], [0.0, 0.0]]], "h": [[[-0.0, 0.0], [0.0, 1.0]], [[0.0, -1.0], [-0.0, 0.0]]], "e**(I*k_y)*p": [[[0.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [1.0, 0.0]]], "e**(-I*k_y)*q": [[[0.0, -1.0], [0.0, 0.0]], [[0.0, 0.0], [0.0, 0.0]]], "e**(I*k_y)*r": [[[0.0, 0.0], [0.0, 1.0]], [[0.0, 0.0], [0.0, 0.0]]], "g": [[[0.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [1.0, 0.0]]], "e**(I*k_y)*q": [[[0.0, 1.0], [0.0, 0.0]], [[0.0, 0.0], [0.0, 0.0]]], "e**(I*k_x)*i": [[[1.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [0.0, 0.0]]], "e**(I*k_x)*k": [[[0.0, 0.0], [0.0, 0.0]], [[0.0, 0.0], [1.0, 0.0]]], "e**(-I*k_y)*s": [[[0.0, 0.0], [0.0, -1.0]], [[0.0, 0.0], [0.0, 0.0]]]}
\ No newline at end of file