mathai 0.4.0__py3-none-any.whl → 0.6.9__py3-none-any.whl

mathai/bivariate_inequality.py

@@ -0,0 +1,317 @@
+import math
+from .linear import linear_or
+from functools import reduce
+import operator
+from .base import *
+from .simplify import simplify
+from .expand import expand
+from .logic import logic0
+
+def shoelace_area(vertices):
+    n = len(vertices)
+    area = 0.0
+    for i in range(n):
+        j = (i + 1) % n
+        area += vertices[i][0] * vertices[j][1]
+        area -= vertices[j][0] * vertices[i][1]
+    return abs(area) / 2.0
+
+def triangle_area(p1, p2, p3):
+    area = 0.0
+    area += p1[0] * (p2[1] - p3[1])
+    area += p2[0] * (p3[1] - p1[1])
+    area += p3[0] * (p1[1] - p2[1])
+    return abs(area) / 2.0
+
+def is_point_inside_polygon(point, vertices):
+    if len(vertices) < 3:
+        return False
+
+    polygon_area = shoelace_area(vertices)
+    
+    total_triangle_area = 0.0
+    n = len(vertices)
+    for i in range(n):
+        j = (i + 1) % n
+        total_triangle_area += triangle_area(point, vertices[i], vertices[j])
+    
+    tolerance = 1e-5
+    return abs(total_triangle_area - polygon_area) < tolerance
+
+def distance_point_to_segment(px, py, x1, y1, x2, y2):
+    dx, dy = x2 - x1, y2 - y1
+    if dx == dy == 0:
+        return ((px - x1)**2 + (py - y1)**2)**0.5
+    t = max(0, min(1, ((px - x1) * dx + (py - y1) * dy) / (dx*dx + dy*dy)))
+    proj_x = x1 + t * dx
+    proj_y = y1 + t * dy
+    return ((px - proj_x)**2 + (py - proj_y)**2)**0.5
+
+def deterministic_middle_point(vertices, grid_resolution=100):
+    xs = [v[0] for v in vertices]
+    ys = [v[1] for v in vertices]
+    xmin, xmax = min(xs), max(xs)
+    ymin, ymax = min(ys), max(ys)
+    
+    best_point = None
+    max_dist = -1
+    
+    for i in range(grid_resolution + 1):
+        for j in range(grid_resolution + 1):
+            px = xmin + (xmax - xmin) * i / grid_resolution
+            py = ymin + (ymax - ymin) * j / grid_resolution
+            if not is_point_inside_polygon((px, py), vertices):
+                continue
+            min_edge_dist = float('inf')
+            n = len(vertices)
+            for k in range(n):
+                x1, y1 = vertices[k]
+                x2, y2 = vertices[(k + 1) % n]
+                d = distance_point_to_segment(px, py, x1, y1, x2, y2)
+                min_edge_dist = min(min_edge_dist, d)
+            if min_edge_dist > max_dist:
+                max_dist = min_edge_dist
+                best_point = (px, py)
+    
+    return best_point
+
+def build(eq):
+    eq = TreeNode("f_or", eq)
+    eq = flatten_tree(eq)
+    orig = eq.copy_tree()
+    def fxhelper3(eq):
+        if eq.name[2:] in "le ge lt gt".split(" "):
+            return TreeNode("f_eq", [child.copy_tree() for child in eq.children])
+        return TreeNode(eq.name, [fxhelper3(child) for child in eq.children])
+    eq = fxhelper3(eq)
+    
+    result = linear_or(eq)
+    
+    if result is None:
+        return None
+    
+    maxnum = tree_form("d_2")
+    if len(result[1]) != 0:
+        maxnum = max([max([simplify(item2.fx("abs")) for item2 in item], key=lambda x: compute(x)) for item in result[1]], key=lambda x: compute(x))
+        maxnum += 1
+        maxnum = simplify(maxnum)
+    eq = flatten_tree(eq | simplify(TreeNode("f_or", [TreeNode("f_eq", [tree_form(item)+maxnum, tree_form("d_0")])|\
+                                                      TreeNode("f_eq", [tree_form(item)-maxnum, tree_form("d_0")]) for item in ["v_0","v_1"]])))
+    result2 = linear_or(eq)
+    if result2 is None:
+        return None
+    
+    point_lst = result2[2]
+
+    def gen(point):
+        nonlocal point_lst
+        out = []
+        for item in point_lst:
+            p = None
+            if point in item:
+                p = item.index(point)
+            else:
+                continue
+            if p < len(item)-1:
+                out.append(item[p+1])
+            if p > 0:
+                out.append(item[p-1])
+        return list(set(out))
+    start = list(range(len(result2[1])))
+    graph= {}
+    for item in start:
+        graph[item] = gen(item)
+
+    points = {}
+    for index, item in enumerate(result2[1]):
+        points[index] = [compute(item2) for item2 in item]
+
+    res = []
+    for index, item in enumerate(result2[1]):
+        if any(simplify(item2.fx("abs")-maxnum)!=0 and abs(compute(item2))>compute(maxnum) for item2 in item):
+            res.append(index)
+            
+    graph = {k: sorted(v) for k, v in graph.items()}
+
+    def dfs(current, parent, path, visited, cycles):
+        path.append(current)
+        visited.add(current)
+        for neighbor in graph[current]:
+            if neighbor == parent:
+                continue
+            if neighbor in visited:
+                idx = path.index(neighbor)
+                cycle = path[idx:]
+                cycles.append(cycle)
+            else:
+                dfs(neighbor, current, path, visited, cycles)
+        path.pop()
+        visited.remove(current)
+
+    cycles = []
+    for start in sorted(graph.keys()):
+        path = []
+        visited = set()
+        dfs(start, -1, path, visited, cycles)
+
+    def normalize(cycle):
+        k = len(cycle)
+        if k < 3:
+            return None
+        candidates = []
+        for direction in [cycle, list(reversed(cycle))]:
+            doubled = direction + direction[:-1]
+            for i in range(k):
+                rot = tuple(doubled[i:i + k])
+                candidates.append(rot)
+        return min(candidates)
+
+    unique = set()
+    for c in cycles:
+        norm = normalize(c)
+        if norm:
+            unique.add(norm)
+
+    cycles = sorted(list(unique), key=lambda x: (len(x), x))
+
+    start = list(range(len(result2[1])))
+    for i in range(len(cycles)-1,-1,-1):
+        if any(item in cycles[i] for item in res) or\
+           any(is_point_inside_polygon([compute(item2) for item2 in list(result2[1][p])], [[compute(item2) for item2 in result2[1][item]] for item in cycles[i]]) for p in list(set(start) - set(cycles[i]))) or\
+           any(len(set(graph[item]) & set(cycles[i]))>2 for item in cycles[i]):
+            cycles.pop(i)
+            
+    point_lst = [index for index, item in enumerate(result2[1]) if item in result[1]]
+
+    border = []
+    for item in start:
+        for item2 in graph[item]:
+            a = result2[1][item]
+            b = result2[1][item2]
+            
+            if a[0] == b[0] and simplify(a[0].fx("abs") - maxnum) == 0:
+                continue
+            if a[1] == b[1] and simplify(a[1].fx("abs") - maxnum) == 0:
+                continue
+            
+            border.append(tuple(sorted([item, item2])))
+
+    line = []
+    for key in graph.keys():
+        for item in list(set(point_lst)&set(graph[key])):
+            line.append(tuple(sorted([item, key])))
+    line = list(set(line+border))
+    point_in = [deterministic_middle_point([[compute(item3) for item3 in result2[1][item2]] for item2 in item]) for item in cycles]
+    def work(eq, point):
+        nonlocal result2
+        if eq.name[:2] == "d_":
+            return float(eq.name[2:])
+        if eq.name in result2[0]:
+            return point[result2[0].index(eq.name)]
+        if eq.name == "f_add":
+            return sum(work(item, point) for item in eq.children)
+        if eq.name == "f_mul":
+            return math.prod(work(item, point) for item in eq.children)
+        if eq.name == "f_sub":
+            return work(eq.children[0], point) - work(eq.children[1], point)
+        return {"eq": lambda a,b: abs(a-b)<0.001, "gt":lambda a,b: False if abs(a-b)<0.001 else a>b, "lt":lambda a,b: False if abs(a-b)<0.001 else a<b}[eq.name[2:]](work(eq.children[0], point), work(eq.children[1], point))
+
+    data = []
+    for index, item in enumerate(result2[2][:-4]):
+        a = tuple([item for item in point_lst if work(orig.children[index], [compute(item2) for item2 in result2[1][item]])])
+        #a = tuple(set(item) & set(point_lst))
+        #b = tuple(set([tuple(sorted([item[i], item[i+1]])) for i in range(len(item)-1)]) & set(line))
+        b = None
+        if orig.children[index] == "f_eq":
+            b = tuple([tuple(item) for item in line if work(orig.children[index], [compute(item2) for item2 in result2[1][item[1]]]) and work(orig.children[index], [compute(item2) for item2 in result2[1][item[0]]])])
+        else:
+            b = tuple([tuple(item) for item in line if work(orig.children[index], [compute(item2) for item2 in result2[1][item[1]]]) or work(orig.children[index], [compute(item2) for item2 in result2[1][item[0]]])])
+        c = tuple([tuple(item) for index2, item in enumerate(cycles) if work(orig.children[index], point_in[index2])])
+        data.append((a,b,c))
+        
+    total = tuple([tuple(point_lst), tuple(line), tuple(cycles)])
+    final = {}
+    for index, item in enumerate(orig.children):
+        final[item] = tuple(data[index])
+    return final, total, result2[1]
+
+def inequality_solve(eq):
+    
+    eq = logic0(eq)
+    element = []
+    def helper(eq):
+        nonlocal element
+        
+        if eq.name[2:] in "le ge lt gt eq".split(" ") and "v_" in str_form(eq):
+            element.append(eq)
+        return TreeNode(eq.name, [helper(child) for child in eq.children])
+    helper(eq)
+    
+    out = build(list(set(element)))
+
+    if out is None:
+        return eq
+    
+    def helper2(eq):
+        nonlocal out
+        if eq == tree_form("s_true"):
+            return [set(item) for item in out[1]]
+        if eq == tree_form("s_false"):
+            return [set(), set(), set()]
+        if eq in out[0].keys():
+            return [set(item) for item in out[0][eq]]
+        if eq.name == "f_or":
+            result = [helper2(child) for child in eq.children]
+            a = []
+            b = []
+            c = []
+            for item in result:
+                a += [item[0]]
+                b += [item[1]]
+                c += [item[2]]
+            x = a[0]
+            for item in a[1:]:
+                x |= item
+            y = b[0]
+            for item in b[1:]:
+                y |= item
+            z = c[0]
+            for item in c[1:]:
+                z |= item
+            return [x, y, z]
+        if eq.name == "f_and":
+            result = [helper2(child) for child in eq.children]
+            a = []
+            b = []
+            c = []
+            for item in result:
+                a += [item[0]]
+                b += [item[1]]
+                c += [item[2]]
+            x = a[0]
+            for item in a[1:]:
+                x &= item
+            y = b[0]
+            for item in b[1:]:
+                y &= item
+            z = c[0]
+            for item in c[1:]:
+                z &= item
+            return [x, y, z]
+        if eq.name == "f_not":
+            eq2 = helper2(eq.children[0])
+            a,b,c= eq2
+            d,e,f= [set(item) for item in out[1]]
+            return [d-a,e-b,f-c]
+        return helper2(dowhile(eq, lambda x: logic0(expand(simplify(eq)))))
+    out2 = helper2(eq)
+    
+    out = list(out)
+    out[1] = [set(item) for item in out[1]]
+    if tuple(out[1]) == (set(), set(), set()):
+        return eq
+    if tuple(out[1]) == tuple(out2):
+        return tree_form("s_true")
+    if tuple(out2) == (set(), set(), set()):
+        return tree_form("s_false")
+    return eq

mathai/diff.py

@@ -1,9 +1,9 @@
-from .simplify import solve
+from .simplify import simplify
 from .base import *
 from .trig import trig0
 def diff(equation, var="v_0"):
     def diffeq(eq):
-        eq = solve(eq)
+        eq = simplify(eq)
         if "v_" not in str_form(eq):
             return tree_form("d_0")
         if eq.name == "f_add":
@@ -65,4 +65,4 @@ def diff(equation, var="v_0"):
         return TreeNode(equation.name, [helper(child, var) for child in equation.children])
     equation = diffeq(trig0(equation))
     equation = helper(equation, var)
-    return solve(equation)
+    return simplify(equation)

mathai/expand.py

@@ -1,96 +1,95 @@
-import itertools
 from .base import *
-from .simplify import solve, simplify
+from .simplify import simplify
+import itertools
+
+def eliminate_powers(node):
+    if not node.children:
+        return node
+
+    node.children = [eliminate_powers(c) for c in node.children]
+
+    if node.name == "f_pow":
+        base, exp = node.children
+        n = frac(exp)
+
+        # Only expand positive integer powers
+        if not (n and n.denominator == 1 and n.numerator > 1):
+            return node
+
+        n = n.numerator
+
+        # ---- Multinomial expansion ----
+        if base.name == "f_add":
+            terms = []
+            for combo in itertools.product(base.children, repeat=n):
+                prod = combo[0]
+                for c in combo[1:]:
+                    prod = prod * c
+                terms.append(prod)
+            return simplify(TreeNode("f_add", terms))
+
+        # ---- Fallback: simple power ----
+        return TreeNode("f_mul", [base] * n)
+
+    return node
 
-def expand(eq):
-    if eq is None:
-        return None
-
-    stack = [(eq, 0)]  # (node, stage)
-    result_map = {}     # id(node) -> expanded TreeNode
-
-    while stack:
-        node, stage = stack.pop()
-        node_id = id(node)
-
-        # Leaf node
-        if not node.children and stage == 0:
-            result_map[node_id] = TreeNode(node.name, [])
-            continue
-
-        if stage == 0:
-            # Stage 0: push node back for stage 1 after children
-            stack.append((node, 1))
-            # Push children to stack
-            for child in reversed(node.children):
-                if id(child) not in result_map:
-                    stack.append((child, 0))
-        else:
-            # Stage 1: all children processed
-            children_expanded = [result_map[id(child)] for child in node.children]
-
-            # Only f_mul or f_pow need special expansion
-            if node.name in ["f_mul", "f_pow"]:
-                current_eq = TreeNode(node.name, children_expanded)
-
-                if node.name == "f_pow":
-                    current_eq = TreeNode("f_pow", [current_eq])
-
-                ac = []
-                addchild = []
-
-                for child in current_eq.children:
-                    tmp5 = [solve(x) for x in factor_generation(child)]
-                    ac += tmp5
-
-                tmp3 = []
-                for child in ac:
-                    tmp2 = []
-                    if child.name == "f_add":
-                        if child.children != []:
-                            tmp2.extend(child.children)
-                        else:
-                            tmp2 = [child]
-                    else:
-                        tmp3.append(child)
-                    if tmp2 != []:
-                        addchild.append(tmp2)
-
-                tmp4 = 1
-                for item in tmp3:
-                    tmp4 = tmp4 * item
-                addchild.append([tmp4])
-
-                def flatten(lst):
-                    flat_list = []
-                    for item in lst:
-                        if isinstance(item, list) and item == []:
-                            continue
-                        if isinstance(item, list):
-                            flat_list.extend(flatten(item))
-                        else:
-                            flat_list.append(item)
-                    return flat_list
-
-                if len(flatten(addchild)) > 0:
-                    add = 0
-                    for prod_items in itertools.product(*addchild):
-                        mul = 1
-                        for item2 in prod_items:
-                            mul = mul * item2
-                            mul = simplify(mul)
-                        add = add + mul
-                        add = simplify(add)
-                    current_eq = simplify(add)
-                else:
-                    current_eq = simplify(current_eq)
-
-                # Store expanded result
-                result_map[node_id] = current_eq
-            else:
-                # Default: reconstruct node with children
-                result_map[node_id] = TreeNode(node.name, children_expanded)
-
-    # Return final expanded eq
-    return result_map[id(eq)]
 
+
+# =====================================================
+# Phase 2: Single distributive rewrite (DEEPEST FIRST)
+# =====================================================
+
+def expand_once(node):
+    """
+    Performs exactly ONE distributive expansion.
+    Deepest-first (post-order).
+    """
+
+    # ---- recurse FIRST (this is the fix) ----
+    for i, c in enumerate(node.children):
+        new, changed = expand_once(c)
+        if changed:
+            node.children[i] = new
+            return node, True
+
+    # ---- now try expanding at this node ----
+    if node.name == "f_mul":
+        for i, child in enumerate(node.children):
+            if child.name == "f_add":
+                left = node.children[:i]
+                right = node.children[i+1:]
+
+                terms = []
+                for t in child.children:
+                    prod = t
+                    for r in right:
+                        prod = prod * r
+                    for l in reversed(left):
+                        prod = l * prod
+                    terms.append(prod)
+
+                return TreeNode("f_add", terms), True
+
+    return node, False
+
+
+# =====================================================
+# Phase 3: Global fixed-point driver
+# =====================================================
+
+def expand(eq):
+    orig = TreeNode.matmul
+    eq = simplify(eq)
+    if TreeNode.matmul is not None:
+        TreeNode.matmul = True
+        eq = tree_form(str_form(eq).replace("f_wmul", "f_mul"))
+        eq = flatten_tree(eq)
+    eq = eliminate_powers(eq)
+    while True:        
+        eq = flatten_tree(eq)        
+        eq, changed = expand_once(eq)
+        if not changed:
+            break
+    eq =simplify(eq)
+    TreeNode.matmul = orig
+    return eq