使用大量内存来绘制越来越多的点，从而解决问题

Question

我一次拍摄一张订单簿的快照，并在3D散点图上绘制每个快照。我的代码可以工作，但是我遇到了使用大量内存的问题。

如何优化它，以便仅绘制添加到数据框中的新点，从而避免出现此问题？

这是相关的代码段。

def orderbook_slice(market, xmin=20, xmax=20, ymax=20, graph=False):
    orderbook = polo.returnOrderBook(str(market))
    bid_price = list()
    bid_amount = list()
    ask_price = list()
    ask_amount = list()
    for i in range(len(orderbook['bids'])):
        bid_price.append(orderbook['bids'][i][0])
        bid_amount.append(orderbook['bids'][i][1])
        ask_price.append(orderbook['asks'][i][0])
        ask_amount.append(orderbook['asks'][i][1])
    market_orderbook = pd.DataFrame({
        'bid_price': bid_price,
        'bid_amount': bid_amount,
        'ask_price': ask_price,
        'ask_amount': ask_amount,
        }, dtype=np.float64)
    market_orderbook['bid cum sum'] =np.cumsum(
        market_orderbook['bid_price'].multiply(market_orderbook['bid_amount']))
    market_orderbook['ask cum sum'] = np.cumsum(
        market_orderbook['ask_price'].multiply(market_orderbook['ask_amount']))
    market_orderbook['time'] = '{:%H:%M:%S}'.format(datetime.datetime.now())
    x_min_lim = float(ask_price[xmin])
    x_max_lim = float(bid_price[xmax])
    if market_orderbook['bid cum sum'][ymax] > market_orderbook['ask cum sum'][ymax]:
        y_max_lim = market_orderbook['bid cum sum'][ymax]
    else:
        y_max_lim = market_orderbook['ask cum sum'][ymax]
    if (graph == True):
        ax = market_orderbook.plot(kind='scatter', x='bid_price', y='bid cum sum', color='Green',
                                   xlim=(x_min_lim, x_max_lim), ylim=(0, y_max_lim))
        market_orderbook.plot(kind='scatter', x='ask_price', y='ask cum sum', color='Red', ax=ax)
        plt.xticks(rotation=45)
    market_orderbook['bid_price'] =     market_orderbook['bid_price'].map('{:,.8f}'.format)
    market_orderbook['ask_price'] =     market_orderbook['ask_price'].map('{:,.8f}'.format)
    market_orderbook['time_s'] = hms_to_s_converter(market_orderbook)
    return market_orderbook

def hms_to_s_converter(df):
        h, m, s = df['time'][0].split(':')
        time = (int(h) * 60 * 60 + int(m) * 60 + int(s))
        return time

fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')

orderbook = orderbook_slice(market)
for i in range(0,900):
    ax.scatter(list(orderbook['bid_price'].astype(float)), list(orderbook['time_s'].astype(float)), list(orderbook['bid cum sum'].astype(float)), color='Green')
    ax.scatter(list(orderbook['ask_price'].astype(float)), list(orderbook['time_s'].astype(float)), list(orderbook['ask cum sum'].astype(float)), color='Red')
    orderbook = orderbook.append(orderbook_slice(market), ignore_index=True)
    plt.pause(1)

plt.show()