示例#1
0
def is_undetermined(x):
    num_samples, ndim = x.shape
    if num_samples <= ndim:
        err.warn('regression is underdetermined: {}'.format(
            x.shape))  #: {}'.format(x))
        return True
    else:
        return False
示例#2
0
def overapprox_x0(num_dims, prop, pwa_trace, solver=gopts.opt_engine):
    cons, Vars, vars_grouped_by_state = pwatrace2cons(pwa_trace, num_dims,
                                                      prop)
    cons = list(cons)

    num_opt_vars = len(Vars)
    nvars = num_dims.x + num_dims.pi

    #directions = [(1, 0), (-1, 0), (0, 1), (0, -1)]
    I = np.eye(nvars)
    directions = np.vstack((I, -I))
    left_over_vars = num_opt_vars - nvars
    directions_ext = np.pad(directions, [(0, 0), (0, left_over_vars)],
                            'constant')

    var_array = np.array(Vars)

    for direction in directions_ext:
        print(np.dot(direction, var_array))

    #lambdafied = tuple(
    #        sym.lambdify(Vars, np.dot(direction, var_array), str('numpy')) for direction in directions_ext)
    #obj_f = tuple(lambda x, l=l: l(*x) for l in lambdafied)
    objs = tuple(np.dot(direction, var_array) for direction in directions_ext)

    x_arr = np.array(sym.symbols(['x{}'.format(i) for i in range(nvars)]))

    res = solve_mult_opt(nlpfun(solver), objs, cons, Vars)

    l = len(res)
    assert (l % 2 == 0)
    min_res, max_res = res[:l // 2], res[l // 2:]

    ranges = []

    for di, rmin, rmax in zip(directions, min_res, max_res):
        if (rmin.success and rmax.success):
            print('{} \in [{}, {}]'.format(np.dot(di, x_arr), rmin.fun,
                                           -rmax.fun))
            ranges.append([rmin.fun, -rmax.fun])
        else:
            if settings.debug:
                print('LP failed')
                print('rmin status:', rmin.status)
                print('rmax status:', rmax.status)
            return None

    r = np.asarray(ranges)
    # due to numerical errors, the interval can be malformed
    try:
        ret_val = IntervalCons(r[:, 0], r[:, 1])
    except ValueError:
        err.warn('linprog fp failure: Malformed Interval! Please fix.')
        return None

    print(ret_val)
    U.pause()
    return ret_val
示例#3
0
文件: plotMP.py 项目: zutshi/S3CAMR
 def plot_rect(self, r, edgecolor="k"):
     if len(r[0]) > 2:
         err.warn("dim>2, projecting before plottting the rectangle on the first 2 dims.")
         c = r[0][0:2]
         rng = r[1][0:2]
     else:
         c = r[0]
         rng = r[1]
     self.ax.add_patch(
         patches.Rectangle(c, *rng, fill=False, edgecolor=edgecolor)
         # patches.Rectangle((0.1, 0.1), 0.5, 0.5, fill=False)
     )
示例#4
0
def cell_affine_models(cell, step_sim, ntrain, ntest, tol, include_err):
    """cell_affine_models

    Parameters
    ----------
    cell : cell
    step_sim : 1 time step (delta_t) simulator
    tol : each abs state is split further into num_splits cells
    in order to meet: modeling error < tol (module ntests samples)

    Returns
    -------
    pwa.SubModel()

    Notes
    ------
    """
    # XXX: Generate different samples for each time step or reuse?
    # Not clear!
    sub_models = []

    X, Y = getxy_ignoramous(cell, ntrain, step_sim)
    rm = RegressionModel(X, Y)
    X, Y = getxy_ignoramous(cell, ntest, step_sim)
    e_pc = rm.error_pc(X, Y) # error %
    if __debug__:
        print('error%:', e_pc)
    #error = np.linalg.norm(e_pc, 2)
    # error exceeds tol in error_dims
    error_dims = np.arange(len(e_pc))[np.where(e_pc >= tol)]

    if len(error_dims) > 0:
        err.warn('splitting on e%:{}, |e%|:{}'.format(
            e_pc, np.linalg.norm(e_pc, 2)))
        for split_cell in cell.split(axes=error_dims):
            sub_models_ = cell_affine_models(
                    split_cell, step_sim, ntrain, ntest, tol, include_err)
            sub_models.extend(sub_models_)
        return sub_models
    else:
        #print('error%:', rm.error_pc(X, Y))
        A, b = rm.Ab
        C, d = cell.ival_constraints.poly()
        e = rm.error(X, Y) if include_err else None
        dmap = pwa.DiscreteAffineMap(A, b, e)
        part = pwa.Partition(C, d, cell)
        sub_model = pwa.SubModel(part, dmap)
        if __debug__:
            print('----------------Finalized------------------')
    return [sub_model]
    def verify_tree_ref(self):
        node = self.root

        # The parent of the root node should always be root
        if node.p is not node:
            err.warn("Root's (" + str(node) + ") parent isn't root")
            return False

        # Verify the parent and children pointers are valid for each subtree
        left = self.verify_tree_ref_helper(node, node.l, True)
        right = self.verify_tree_ref_helper(node, node.r, False)

        err.log("Left of Root: " + str(left))
        err.log("Left of Root: " + str(right))

        return left and right
示例#6
0
文件: plotMP.py 项目: zutshi/S3CAMR
 def plot_rect(self, r, edgecolor='k'):
     if len(r[0]) > 2:
         err.warn(
             'dim>2, projecting before plottting the rectangle on the first 2 dims.'
         )
         c = r[0][0:2]
         rng = r[1][0:2]
     else:
         c = r[0]
         rng = r[1]
     self.ax().add_patch(
         patches.Rectangle(c,
                           *rng,
                           fill=False,
                           edgecolor=edgecolor,
                           linewidth=2)
         #patches.Rectangle((0.1, 0.1), 0.5, 0.5, fill=False)
     )
示例#7
0
    def wraped_call(tcd):
        err.warn('wraped_call being used!')
        tcd.input_array = np.array(
            [int(round(i * cf)) for i in tcd.input_array])
        tcd.state_array = np.array(
            [int(round(i * cf)) for i in tcd.state_array])
        tcd.x_array = np.array([int(round(i * cf)) for i in tcd.x_array])

        #        tcd.state_array = (tcd.state_array * cf).astype(int)
        #        tcd.x_array = (tcd.x_array * cf).astype(int)

        fcd = controller_call_fun(tcd)

        # WHY ROUND here???
        # fcd.state_array = list((np.round(fcd.state_array / cf).astype(float)))
        # fcd.output_array = list((np.round(fcd.output_array / cf).astype(float)))

        fcd.state_array = list(fcd.state_array.astype(float) / cf)
        fcd.output_array = list(fcd.output_array.astype(float) / cf)
        return fcd
示例#8
0
 def trace_generator(self, depth):
     # TODO: make the graph search depth bounded?
     err.warn('ignoring depth')
     path_gen = self.pwa_model.get_all_path_generator(
         self.sources, self.targets)
     path_ctr = 0
     for path in path_gen:
         path_ctr += 1
         ptrace = [self.pwa_model.node_p(qi) for qi in path]
         mtrace = [
             self.pwa_model.edge_m((qi, qj)) for qi, qj in U.pairwise(path)
         ]
         pwa_trace = PWATRACE(partitions=ptrace, models=mtrace)
         x_array = azp.feasible(self.num_dims, self.prop, pwa_trace)
         if x_array is not None:
             concrete_trace = ConcreteTrace(x_array, pwa_trace)
             print('Model Found')
             yield concrete_trace, pwa_trace
     print('Total paths checked: {}'.format(path_ctr))
     return
示例#9
0
def build_pwa_dt_model(AA, abs_states, sp, sys_sim):
    """build_pwa_dt_model

    Parameters
    ----------
    AA :
        AA is
    abs_states :
        abs_states is
    sp :
        sp is
    sys_sim :
        sys_sim is

    Returns
    -------

    Notes
    ------
    Builds a model with time as a discrete variable.
    i.e., models the behaviors resulting from several chosen time
    steps and not only the one specified in .tst as delta_t.
    """

    dt_steps = [0.01, 0.1, AA.plant_abs.delta_t]
    err.warn('using time steps: {}'.format(dt_steps))
    step_sims = [simsys.get_step_simulator(sp.controller_sim, sp.plant_sim, dt)
                 for dt in dt_steps]

    pwa_models = {}
    for dt, step_sim in zip(dt_steps, step_sims):
        pwa_model = pwa.PWA()
        for abs_state in abs_states:
            sub_model = affine_model(abs_state, AA, sp, step_sim)
            pwa_model.add(sub_model)

        pwa_models[dt] = pwa_model
    return pwa_models
示例#10
0
def getxy_rel_ignoramous(cell1, cell2, force, N, sim, t0=0):
    """getxy_rel_ignoramous

    Parameters
    ----------
    force : force to return non-zero samples. Will loop for infinitiy
            if none exists.
    """
    xl = []
    yl = []
    sat_count = 0
    iter_count = itertools.count()
    print(cell1.ival_constraints)
    print(cell2.ival_constraints)
    if __debug__:
        obs_cells = set()
    while next(iter_count) <= MAX_ITER:
        x_array, y_array = getxy_ignoramous(cell1, N, sim, t0=0)
        if __debug__:
            for i in y_array:
                obs_cells.add(CM.cell_from_concrete(i, cell1.eps))
            print('reachable cells:', obs_cells)
        # satisfying indexes
        sat_array = cell2.ival_constraints.sat(y_array)
        sat_count += np.sum(sat_array)
        xl.append(x_array[sat_array])
        yl.append(y_array[sat_array])
        if sat_count >= MIN_TRAIN:
            break
        # If no sample is found and force is True, must keep sampling till
        # satisfying samples are found
    if __debug__:
        if sat_count < MIN_TRAIN:
            err.warn('Fewer than MIN_TRAIN samples found!')
    print('found samples: ', sat_count)
    return np.vstack(xl), np.vstack(yl)
elif algarg in {'avl', 'avltree'}:
    algo = algs[3]

elif algarg in {'wavl', 'wavltree', 'weakavl', 'weakavltree'}:
    algo = algs[4]

elif algarg in {'tango', 'tangotree'}:
    algo = algs[5]

elif algarg in {
        'static', 'osbst', 'optimalstatic', 'opt', 'optbst', 'optimalstaticbst'
}:
    algo = algs[6]

else:
    err.err("Algorithm not recognized.")

debug = args.debug_level

if args.debug and debug == 0:
    debug = 1

if debug == 1:
    err.warn("Debug mode is set to: SIMPLE")
elif debug == 2:
    err.warn("Debug mode is set to: VERBOSE")
elif debug == 3:
    err.warn("Debug mode is set to: VERIFY")

ops.exec_ops(algo, args.pages, args.graphs, args.clean_off, debug)
示例#12
0
    def plot(self, X, Y, tol, title='unknown'):
        from matplotlib import pyplot as plt
        from mpl_toolkits.mplot3d import Axes3D

        dimX = X.shape[1]
        assert (dimX == 2)

        # if 1 or less rows
        if X.shape[0] <= 1:
            err.warn('cant plot, only 1 value!!')
            return

        div = 50
        X_min, X_max = np.min(X, axis=0), np.max(X, axis=0)
        step0 = (X_max[0] - X_min[0]) / div
        step1 = (X_max[1] - X_min[1]) / div
        step = min(step0, step1)
        #print('x0 range:', X_min[0], X_max[0])
        #print('x1 range:', X_min[1], X_max[1])

        # Predict data of estimated models
        xx1, xx2 = np.mgrid[X_min[0]:X_max[0]:step, X_min[1]:X_max[1]:step]
        xx = np.vstack([xx1.ravel(), xx2.ravel()]).T
        yy = self.predict(xx)

        yy0 = yy[:, 0]
        yy1 = yy[:, 1]
        Y0 = Y[:, 0]
        Y1 = Y[:, 1]

        error_pc = self.error_pc(X, Y)
        outlier0_idx = error_pc[:, 0] > tol
        outlier1_idx = error_pc[:, 1] > tol

        if any(outlier0_idx):
            #print('X:\n', X[outlier0_idx, :])
            #print('Y0_pred:', self.predict(X[outlier0_idx, :])[:, 0])
            #print('Y0_true', Y0[outlier0_idx])
            Y0_pred_ = self.predict(X[outlier0_idx, :])[:, 0]
            Y0_pred = np.reshape(Y0_pred_, (Y0_pred_.size, 1))
            # make it 2-dim to match dim of X and Y0_pred
            Y0_true_ = Y0[outlier0_idx]
            Y0_true = np.reshape(Y0_true_, (Y0_true_.size, 1))
            epc0_ = error_pc[outlier0_idx, 0]
            epc0 = np.reshape(epc0_, (epc0_.size, 1))

            logger.debug('X - Y0_pred - Y0_true - error_pc')
            logger.debug(
                np.hstack((X[outlier0_idx, :], Y0_pred, Y0_true, epc0)))

        if any(outlier1_idx):
            Y1_pred_ = self.predict(X[outlier1_idx, :])[:, 1]
            Y1_pred = np.reshape(Y1_pred_, (Y1_pred_.size, 1))

            # make it 2-dim to match dim of X and Y1_pred
            Y1_true_ = Y1[outlier1_idx]
            Y1_true = np.reshape(Y1_true_, (Y1_true_.size, 1))
            epc1_ = error_pc[outlier1_idx, 1]
            epc1 = np.reshape(epc1_, (epc1_.size, 1))
            logger.debug('X - Y1_pred - Y1_true - error_pc')
            logger.debug(
                np.hstack((X[outlier1_idx, :], Y1_pred, Y1_true, epc1)))

        # plot the surface
        #print(xx)
        #print(xx2)
        #print(yy[:, 0])
        #embed()

        ################
        # First subplot
        ################

        fig = plt.figure()  #figsize=plt.figaspect(2.))
        fig.suptitle(title)

        ax = fig.add_subplot(2, 1, 1, projection='3d')
        ax.plot_surface(xx1, xx2, np.reshape(yy0, (xx1.shape)))
        ax.plot(X[:, 0], X[:, 1], Y0, 'y.')
        ax.plot(X[outlier0_idx, 0], X[outlier0_idx, 1], Y0[outlier0_idx], 'r.')
        ax.set_title('y0 vs x')
        ax.set_xlabel('x0')
        ax.set_ylabel('x1')
        ax.set_zlabel('y0')

        #################
        # Second subplot
        #################
        ax = fig.add_subplot(2, 1, 2, projection='3d')
        ax.plot_surface(xx1, xx2, np.reshape(yy1, (xx1.shape)))
        ax.plot(X[:, 0], X[:, 1], Y1, 'y.')
        ax.plot(X[outlier1_idx, 0], X[outlier1_idx, 1], Y1[outlier1_idx], 'r.')
        ax.set_title('y1 vs x')
        ax.set_xlabel('x0')
        ax.set_ylabel('x1')
        ax.set_zlabel('y1')
示例#13
0
def cell_rel_affine_models(cell1, cell2, force, step_sim, ntrain, ntest, tol, include_err):
    """cell_affine_models

    Parameters
    ----------
    cell1 : source cell
    cell2 : target cell
    step_sim : 1 time step (delta_t) simulator
    tol : each abs state is split further into num_splits cells
    in order to meet: modeling error < tol (module ntests samples)

    Returns
    -------
    pwa.SubModel()

    Notes
    ------
    """
    # XXX: Generate different samples for each time step or reuse?
    # Not clear!
    sub_models = []

    X, Y = getxy_rel_ignoramous(cell1, cell2, force, ntrain, step_sim)
    # No samples found => no model
    training_samples_found = len(X) != 0
    if not training_samples_found:
        return [None]
    rm = RegressionModel(X, Y)
    X, Y = getxy_rel_ignoramous(cell1, cell2, True, ntest, step_sim)
    testing_samples_found = len(X) != 0
    # If valid samples are found, compute e_pc (error %) and dims
    # where error % >= given tol
    if testing_samples_found:
        e_pc = rm.error_pc(X, Y)
        error_dims = np.arange(len(e_pc))[np.where(e_pc >= tol)]
    # Otherwise, forget it!
    else:
        e_pc = None
        error_dims = []

    if __debug__:
        print('error%:', e_pc)

    if len(error_dims) > 0:
        err.warn('splitting on e%:{}, |e%|:{}'.format(
            e_pc, np.linalg.norm(e_pc, 2)))
        for split_cell1 in cell1.split(axes=error_dims):
            sub_models_ = cell_rel_affine_models(
                    split_cell1, cell2, False, step_sim, ntrain, ntest, tol, include_err)
            sub_models.extend(sub_models_)
        return sub_models
    else:
        A, b = rm.Ab
        C1, d1 = cell1.ival_constraints.poly()
        C2, d2 = cell2.ival_constraints.poly()

        e = rm.error(X, Y) if (include_err and testing_samples_found) else None

        dmap = rel.DiscreteAffineMap(A, b, e)
        part1 = rel.Partition(C1, d1, cell1)
        part2 = rel.Partition(C2, d2, cell2)
        sub_model = rel.Relation(part1, part2, dmap)
        if __debug__:
            print('----------------Finalized------------------')
    return [sub_model]
示例#14
0
def warn_if_small_data_set(x):
    if len(x) <= 2:
        err.warn_severe('Less than 2 training samples!: {}'.format(x))
    elif len(x) <= 10:
        err.warn('Less than 10 training samples!: {}'.format(x))
示例#15
0
    def get_all_path_generator(
            G,
            source_list,
            sink_list,
            max_depth,
            max_paths
            ):
        # Define super source and sink nodes
        # A Super source node has a directed edge to each source node in the
        # source_list
        # Similarily, a Super sink node has a directed edge from each sink node
        # in the sink_list

        dummy_super_source_node = 'source'
        dummy_super_sink_node = 'sink'
        num_source_nodes = len(source_list)
        num_sink_nodes = len(sink_list)

        # increment max_depth by 2 to accommodate edges from 'super source' and
        # to 'super sink'
        max_depth += 2

        # Add edges:
        #   \forall source \in source_list. super source node -> source

        edge_list = list(zip([dummy_super_source_node] * num_source_nodes,
                        source_list))
        G.add_edges_from(edge_list)

        if settings.debug:
            U.eprint('source -> list')
            for e in edge_list:
                U.eprint(e)

        # Add edges:
        #   \forall sink \in sink_list. sink -> super sink node

        edge_list = list(zip(sink_list, [dummy_super_sink_node] *
                             num_sink_nodes))
        G.add_edges_from(edge_list)

        if settings.debug:
            U.eprint('sink -> list')
            for e in edge_list:
                U.eprint(e)

        if settings.debug:
            #print(the graph first)
            U.eprint('Printing graph...')
            for e in G.G.edges():
                s, t = e
                U.eprint('{}, {}'.format(G.v_attr[s], G.v_attr[t]))
            U.eprint('Printing graph...done')

        path_it = G.all_shortest_paths(dummy_super_source_node,
                                       dummy_super_sink_node,
                                       )

        if settings.debug:
            U.eprint('path list')
            paths = list(path_it)
            for path in paths[0:max_paths]:
                p = [G.v_attr[v] for v in path]
                U.eprint(p)
            path_it = (i for i in paths)

#############################################################
        # TODO: Remove the extra step to count paths
        # It is there just as a debug print
        paths = list(U.bounded_iter(path_it, max_paths))
        num_paths = len(paths)
        err.warn('counting paths...found: {}'.format(num_paths))

        def path_gen():
            for path in paths:
                p = [G.v_attr[v] for v in path]
                yield p[1:-1]

        # END: CODE TO BE REMOVED
        # CORRECT CODE BELOW
#         def path_gen():
#             for path in U.bounded_iter(path_it, max_paths):
#                 p = [G.v_attr[v] for v in path]
#                 yield p[1:-1]
#############################################################
        return path_gen()
示例#16
0
def init(
        A,
        init_cons_list_plant,
        final_cons,
        init_d,
        controller_init_state,
        ):

    PA = A.plant_abs
    d, pvt, n = init_d, (0, ), 0

    plant_initial_state_set = set()

    def f(init_cons_): return PA.get_abs_state_set_from_ival_constraints(init_cons_, n, d, pvt)

    for init_cons in init_cons_list_plant:
        init_abs_states = f(init_cons)

        # filters away zero measure sets
        def fnzm(as_):
            intsec = PA.get_ival_cons_abs_state(as_) & init_cons
            return (intsec is not None) and not intsec.zero_measure

        err.warn('what is going on?')
        #filt_init_abs_states = filter(fnzm, init_abs_states)
        filt_init_abs_states = init_abs_states

        plant_initial_state_set.update(filt_init_abs_states)


# Old code to compute plant_initial_state_set
#     plant_initial_state_list = []
#     for init_cons in init_cons_list_plant:
#         plant_initial_state_list += \
#             PA.get_abs_state_set_from_ival_constraints(init_cons, n, d, pvt)
#     plant_initial_state_set = set(plant_initial_state_list)


    # The below can be very very expensive in time and memory for large final
    # sets!
    # plant_final_state_set = \
    #    set(PA.get_abs_state_set_from_ival_constraints(final_cons, 0, 0, 0))

    # ##!!##logger.debug('{0}initial plant states{0}\n{1}'.format('=' * 10, plant_initial_state_set))

    # set control states for initial states
    # TODO: ideally the initial control states should be supplied by the
    # user and the below initialization should be agnostic to the type


    initial_state_list = []
    for plant_init_state in plant_initial_state_set:
        initial_state_list.append(AA.TopLevelAbs.get_abs_state(plant_state=plant_init_state))

    #final_state_list = []

#    for plant_final_state in plant_final_state_set:
#        final_state_list.append(AA.TopLevelAbs.get_abs_state(
#            plant_state=plant_final_state,
#            controller_state=controller_init_abs_state))

    # print('='*80)
    # print('all final states')
    # for ps in plant_final_state_set:
    #    print(ps)
    # print('='*80)

    def is_final(_A, abs_state):

        #print('----------isfinal-----------')
        #print(abs_state.plant_state.cell_id == ps.cell_id)
        #print(abs_state.plant_state in plant_final_state_set)
        #print(hash(abs_state.plant_state), hash(ps))
        #print(abs_state.plant_state == ps)
        #if abs_state.plant_state.cell_id == ps.cell_id:
            #exit()

        # return abs_state.plant_state in plant_final_state_set

        #print('---------------------------------------------')
        #print(_A.plant_abs.get_ival_constraints(abs_state.ps))
        #print('---------------------------------------------')
        pabs_ic = _A.plant_abs.get_ival_cons_abs_state(abs_state.plant_state)
        intersection = pabs_ic & final_cons
        return not(intersection is None or intersection.zero_measure)

    # ##!!##logger.debug('{0}initial{0}\n{1}'.format('=' * 10, plant_initial_state_set))

    return (set(initial_state_list), is_final)
示例#17
0
def overapprox_x0(num_dims, prop, pwa_trace, solver=gopts.opt_engine):
    A_ub, b_ub = pwatrace2lincons(pwa_trace, num_dims, prop)

    num_opt_vars = A_ub.shape[1]

    nvars = num_dims.x + num_dims.pi

    #directions = [(1, 0), (-1, 0), (0, 1), (0, -1)]
    I = np.eye(nvars)
    directions = np.vstack((I, -I))
    left_over_vars = num_opt_vars - nvars
    directions_ext = np.pad(directions, [(0, 0), (0, left_over_vars)],
                            'constant')

    # LP structure: A_ub [x x'] <= b_ub

    x_arr = np.array(sym.symbols(['x{}'.format(i) for i in range(nvars)]))

    A_ub, b_ub = truncate(A_ub, b_ub)

    res = solve_mult_lp(lpfun(solver), directions_ext, A_ub, b_ub)

    l = len(res)
    assert (l % 2 == 0)
    #min_directions, max_directions = np.split(directions, l/2, axis=1)
    min_res, max_res = res[:l // 2], res[l // 2:]

    ranges = []

    for di, rmin, rmax in zip(directions, min_res, max_res):
        if (rmin.success and rmax.success):
            print('{} \in [{}, {}]'.format(np.dot(di, x_arr), rmin.fun,
                                           -rmax.fun))
            ranges.append([rmin.fun, -rmax.fun])
        else:
            if settings.debug:
                print('LP failed')
                print('rmin status:', rmin.status)
                print('rmax status:', rmax.status)
            return None

            #raise e

    # For python2/3 compatibility
#     try:
#         input = raw_input
#     except NameError:
#         pass
#     prompt = input('load the prompt? (y/Y)')
#     if prompt.lower() == 'y':
#         import IPython
#         IPython.embed()

#ranges = [[0.00416187, 0.00416187],[3.47047152,3.47047152],[9.98626028,9.98626028],[4.98715449,4.98715449]]
    r = np.asarray(ranges)
    # due to numerical errors, the interval can be malformed
    try:
        ret_val = cons.IntervalCons(r[:, 0], r[:, 1])
    except:
        #from IPython import embed
        err.warn('linprog fp failure: Malformed Interval! Please fix.')
        #embed()
        return None

    return ret_val
示例#18
0
文件: graphGT.py 项目: zutshi/S3CAMR
    def get_all_path_generator(
            G,
            source_list,
            sink_list,
            max_depth,
            max_paths
            ):
        # Define super source and sink nodes
        # A Super source node has a directed edge to each source node in the
        # source_list
        # Similarily, a Super sink node has a directed edge from each sink node
        # in the sink_list

        dummy_super_source_node = 'source'
        dummy_super_sink_node = 'sink'
        num_source_nodes = len(source_list)
        num_sink_nodes = len(sink_list)

        # increment max_depth by 2 to accommodate edges from 'super source' and
        # to 'super sink'
        max_depth += 2

        # Add edges:
        #   \forall source \in source_list. super source node -> source

        edge_list = zip([dummy_super_source_node] * num_source_nodes,
                        source_list)
        G.add_edges_from(edge_list)

        if __debug__:
            print >> sys.stderr, 'source -> list'
            for e in edge_list:
                print >> sys.stderr, e

        # Add edges:
        #   \forall sink \in sink_list. sink -> super sink node

        edge_list = zip(sink_list, [dummy_super_sink_node] * num_sink_nodes)
        G.add_edges_from(edge_list)

        if __debug__:
            print >> sys.stderr, 'sink -> list'
            for e in edge_list:
                print >> sys.stderr, e

        if __debug__:
            #print the graph first
            print >>sys.stderr,  'Printing graph...'
            for e in G.G.edges():
                s, t = e
                print >>sys.stderr, '{}, {}'.format(G.v_attr[s], G.v_attr[t])
            print >>sys.stderr,  'Printing graph...done'

        path_it = G.all_shortest_paths(dummy_super_source_node,
                                       dummy_super_sink_node,
                                       )

        if __debug__:
            print >> sys.stderr, 'path list'
            paths = list(path_it)
            for path in paths[0:max_paths]:
                p = [G.v_attr[v] for v in path]
                print >> sys.stderr, p
            path_it = (i for i in paths)

#############################################################
        # TODO: Remove the extra step to count paths
        # It is there just as a debug print
        paths = list(U.bounded_iter(path_it, max_paths))
        num_paths = len(paths)
        err.warn('counting paths...found: {}'.format(num_paths))
        def path_gen():
            for path in paths:
                p = [G.v_attr[v] for v in path]
                yield p[1:-1]

        # END: CODE TO BE REMOVED
        # CORRECT CODE BELOW
#         def path_gen():
#             for path in U.bounded_iter(path_it, max_paths):
#                 p = [G.v_attr[v] for v in path]
#                 yield p[1:-1]
#############################################################
        return path_gen()
示例#19
0
    def exec_ops(self, algo, pages, gen_graphs, no_clean, debug):
        logt = []
        logn = []
        opst = []
        opsn = []

        graphs = []

        api = API(logn, logt, gen_graphs, graphs, debug)
        if algo == 'simple':
            tree = SimpleBST(api)
        elif algo == 'rb':
            tree = RedBlackBST(api)
        elif algo == 'splay':
            tree = SplayBST(api)
        elif algo == 'avl':
            err.err("Algorithm not yet implemented")
        elif algo == 'wavl':
            err.err("Algorithm not yet implemented")
        elif algo == 'tango':
            err.err("Algorithm not yet implemented")
        elif algo == 'static':
            err.err("Algorithm not yet implemented")

        time = 1
        for op in self.ops:
            if debug > 2:
                tree_str = str(tree)
            if debug > 1:
                err.log("operation #" + str(time) + ": " + str(op))
            api.reset()
            if op.op == 'ins':
                opst.append(time)
                opsn.append(op.arg)
                api.set_time(time)
                api.set_log_on()
                tree.insert(op.arg)
                api.set_log_off()
                time += 1
            elif op.op == 'sea':
                opst.append(time)
                opsn.append(op.arg)
                api.set_time(time)
                api.set_log_on()
                tree.search(op.arg)
                api.set_log_off()
                time += 1
            elif op.op == 'del':
                opst.append(time)
                opsn.append(op.arg)
                api.set_time(time)
                api.set_log_on()
                tree.delete(op.arg)
                api.set_log_off()
                time += 1
            if debug == 2:
                err.warn(tree)
            if debug > 2:
                err.log("Before op, tree was:")
                err.warn(tree_str)
                err.log("After op, tree is:")
                err.warn(tree)
                err.warn("Beginning tree verification:")
                if tree.verify_tree():
                    err.warn("Verified")
                else:
                    err.err("Issue found when verifying tree!")
            if gen_graphs and pages:
                api.viz()

        if gen_graphs and not pages:
            api.viz()

        if debug == 1 and len(self.ops) < 50:
            err.warn(tree)

        plot.plot(logn, logt, opsn, opst, pages, graphs, no_clean, debug)