13. 南湖case
如果想要得到该例子的hlo/mhlo，可以修改/home/ken/virtualEnv/jaxEnv/lib/python3.7/site-packages/brainpy/dyn/runners.py，在其中加入如下输出代码：（422 ~ 428）为修改后新加入的代码。
Python
418 # running
419 if eval_time: t0 = time.time()
420 outputs, hists = self._predict(xs=(times, indices, xs), shared_args=shared_args)
421
422 ################################################
423 hlo_string = jax.jit(self._predict).lower(xs=(times, indices, xs)).as_text(dialect='hlo') # `mhlo`则输出mhlo
424 with open('/home/ken/log/hlo_string.txt','w') as f:
425 f.write(str(hlo_string))
426 f.close()
427
428 ################################################
429
430 if eval_time: running_time = time.time() - t0
192.168.53.138:/home/ken/workspace/test/brainpy/case2
Python
import brainpy as bp
bp.math.set_platform('cpu')
class EINet(bp.dyn.Network):
def __init__(self, scale=1.0, method='exp_auto'):
# network size
num_exc = int(10000 * scale)
num_inh = int(10000 * scale)
# neurons
pars = dict(V_rest=-60., V_th=-50., V_reset=-60., tau=20., tau_ref=5.)
E = bp.dyn.LIF(num_exc, **pars, method=method)
I = bp.dyn.LIF(num_inh, **pars, method=method)
E.V[:] = bp.math.random.randn(num_exc) * 2 - 55. # 10000 float32
I.V[:] = bp.math.random.randn(num_inh) * 2 - 55.
# synapses（突触）
we = 0.6 / scale # excitatory(兴奋) synaptic weight (voltage)
wi = 6.7 / scale # inhibitory(抑制) synaptic weight
E2E = bp.dyn.ExpCOBA(E, E, bp.conn.FixedProb(0.02),
E=0., g_max=we, tau=5., method=method)
E2I = bp.dyn.ExpCOBA(E, I, bp.conn.FixedProb(0.02),
E=0., g_max=we, tau=5., method=method)
I2E = bp.dyn.ExpCOBA(I, E, bp.conn.FixedProb(0.02),
E=-80., g_max=wi, tau=10., method=method)
I2I = bp.dyn.ExpCOBA(I, I, bp.conn.FixedProb(0.02),
E=-80., g_max=wi, tau=10., method=method)
super(EINet, self).__init__(E2E, E2I, I2E, I2I, E=E, I=I)
net = EINet(scale=1., method='exp_auto')
# simulation
runner = bp.dyn.DSRunner(net,
monitors=['E.spike'],
inputs=[('E.input', 20.), ('I.input', 20.)],
jit=True)
t = runner.run(1000.)
#t = runner.run(1000.)
#print(t)
brainpy.dyn.neurons.LIF doc https://brainpy.readthedocs.io/en/latest/apis/auto/dyn/generated/brainpy.dyn.neurons.LIF.html?highlight=LIF
V是膜电位
Vth是spke threshold电位，
Vrest是静息电位
Vreset是reset膜电位，到达spike threshold后进行重置的值，
tao是时间常量
Tref：不应期时间，为了使模型更精准
上述代码中LIF表示脑仿真领域中的1种经典的神经元模型，LIF的参数num_exc等表示LIF神经元的个数，bp.dyn.LIF（num_exc,……）表示num_exc个LIF神经元组成的神经元簇，ExpCOBA表示用于连接不同神经元的突触模型，EINet表示由一个兴奋性（E）神经元簇和一个抑制性（I）神经元簇通过突触连接而成的一个简单脑仿真网络。建模过程的说明图片如下所示：
class LIF定义在下面这个文件中。
/home/ken/virtualEnv/jaxEnv/lib/python3.7/site-packages/brainpy/dyn/neurons/reduced_models.py
上述代码中构建runner时jit参数设置为True，进行即时编译。设置环境变量JAX_DUMP_IR_TO为某个文件路径，然后运行上述模型。生成的IR文件会储存到环境变量对应的路径中。
Python
├── ir
│ ├── jax_ir0_jit_prim_fun.mlir
│ ├── jax_ir10_jit_prim_fun.mlir
│ ├── jax_ir11_jit_prim_fun.mlir
│ ├── jax_ir12_jit_fn.mlir
│ ├── jax_ir13_jit_fn.mlir
│ ├── jax_ir14_jit_prim_fun.mlir
│ ├── jax_ir15_jit__lambda_.mlir
│ ├── jax_ir1_jit_prim_fun.mlir
│ ├── jax_ir2_jit_fn.mlir
│ ├── jax_ir3_jit__threefry_split.mlir
│ ├── jax_ir4_jit_prim_fun.mlir
│ ├── jax_ir5_jit_prim_fun.mlir
│ ├── jax_ir6_jit_prim_fun.mlir
│ ├── jax_ir7_jit__normal.mlir
│ ├── jax_ir8_jit_fn.mlir
│ └── jax_ir9_jit__lambda_.mlir
├── test.py
└── tmp
└── foo
├── module_0000.jit_prim_fun.before_optimizations.dot
├── module_0000.jit_prim_fun.before_optimizations.hlo.pb
├── module_0000.jit_prim_fun.before_optimizations.html
├── module_0000.jit_prim_fun.before_optimizations.txt
├── module_0000.jit_prim_fun.cpu_after_optimizations-buffer-assignment.txt
├── module_0000.jit_prim_fun.cpu_after_optimizations.dot
├── module_0000.jit_prim_fun.cpu_after_optimizations.hlo.pb
├── module_0000.jit_prim_fun.cpu_after_optimizations.html
├── module_0000.jit_prim_fun.cpu_after_optimizations.txt
├── module_0000.jit_prim_fun.ir-no-opt.ll
├── module_0000.jit_prim_fun.ir-no-opt-noconst.ll
├── module_0000.jit_prim_fun.ir-with-opt.ll
├── module_0000.jit_prim_fun.ir-with-opt-noconst.ll
├── module_0000.jit_prim_fun.o
├── module_0001.jit_prim_fun.before_optimizations.dot
├── module_0001.jit_prim_fun.before_optimizations.hlo.pb
├── module_0001.jit_prim_fun.before_optimizations.html
├── module_0001.jit_prim_fun.before_optimizations.txt
├── module_0001.jit_prim_fun.cpu_after_optimizations-buffer-assignment.txt
├── module_0001.jit_prim_fun.cpu_after_optimizations.dot
├── module_0001.jit_prim_fun.cpu_after_optimizations.hlo.pb
├── module_0001.jit_prim_fun.cpu_after_optimizations.html
├── module_0001.jit_prim_fun.cpu_after_optimizations.txt
├── module_0001.jit_prim_fun.ir-no-opt.ll
├── module_0001.jit_prim_fun.ir-no-opt-noconst.ll
├── module_0001.jit_prim_fun.ir-with-opt.ll
├── module_0001.jit_prim_fun.ir-with-opt-noconst.ll
├── module_0001.jit_prim_fun.o
├── module_0002.jit_fn.before_optimizations.dot
├── module_0002.jit_fn.before_optimizations.hlo.pb
├── module_0002.jit_fn.before_optimizations.html
├── module_0002.jit_fn.before_optimizations.txt
├── module_0002.jit_fn.cpu_after_optimizations-buffer-assignment.txt
├── module_0002.jit_fn.cpu_after_optimizations.dot
├── module_0002.jit_fn.cpu_after_optimizations.hlo.pb
├── module_0002.jit_fn.cpu_after_optimizations.html
├── module_0002.jit_fn.cpu_after_optimizations.txt
├── module_0002.jit_fn.ir-no-opt.ll
├── module_0002.jit_fn.ir-no-opt-noconst.ll
├── module_0002.jit_fn.ir-with-opt.ll
├── module_0002.jit_fn.ir-with-opt-noconst.ll
├── module_0002.jit_fn.o
├── module_0003.jit__threefry_split.before_optimizations.dot
├── module_0003.jit__threefry_split.before_optimizations.hlo.pb
├── module_0003.jit__threefry_split.before_optimizations.html
├── module_0003.jit__threefry_split.before_optimizations.txt
├── module_0003.jit__threefry_split.cpu_after_optimizations-buffer-assignment.txt
├── module_0003.jit__threefry_split.cpu_after_optimizations.dot
├── module_0003.jit__threefry_split.cpu_after_optimizations.hlo.pb
├── module_0003.jit__threefry_split.cpu_after_optimizations.html
├── module_0003.jit__threefry_split.cpu_after_optimizations.txt
├── module_0003.jit__threefry_split.ir-no-opt.ll
├── module_0003.jit__threefry_split.ir-no-opt-noconst.ll
├── module_0003.jit__threefry_split.ir-with-opt.ll
├── module_0003.jit__threefry_split.ir-with-opt-noconst.ll
├── module_0003.jit__threefry_split.o
├── module_0004.jit_prim_fun.before_optimizations.dot
├── module_0004.jit_prim_fun.before_optimizations.hlo.pb
├── module_0004.jit_prim_fun.before_optimizations.html
├── module_0004.jit_prim_fun.before_optimizations.txt
├── module_0004.jit_prim_fun.cpu_after_optimizations-buffer-assignment.txt
├── module_0004.jit_prim_fun.cpu_after_optimizations.dot
├── module_0004.jit_prim_fun.cpu_after_optimizations.hlo.pb
├── module_0004.jit_prim_fun.cpu_after_optimizations.html
├── module_0004.jit_prim_fun.cpu_after_optimizations.txt
├── module_0004.jit_prim_fun.ir-no-opt.ll
├── module_0004.jit_prim_fun.ir-no-opt-noconst.ll
├── module_0004.jit_prim_fun.ir-with-opt.ll
├── module_0004.jit_prim_fun.ir-with-opt-noconst.ll
├── module_0004.jit_prim_fun.o
├── module_0005.jit_prim_fun.before_optimizations.dot
├── module_0005.jit_prim_fun.before_optimizations.hlo.pb
├── module_0005.jit_prim_fun.before_optimizations.html
├── module_0005.jit_prim_fun.before_optimizations.txt
├── module_0005.jit_prim_fun.cpu_after_optimizations-buffer-assignment.txt
├── module_0005.jit_prim_fun.cpu_after_optimizations.dot
├── module_0005.jit_prim_fun.cpu_after_optimizations.hlo.pb
├── module_0005.jit_prim_fun.cpu_after_optimizations.html
├── module_0005.jit_prim_fun.cpu_after_optimizations.txt
├── module_0005.jit_prim_fun.ir-no-opt.ll
├── module_0005.jit_prim_fun.ir-no-opt-noconst.ll
├── module_0005.jit_prim_fun.ir-with-opt.ll
├── module_0005.jit_prim_fun.ir-with-opt-noconst.ll
├── module_0005.jit_prim_fun.o
├── module_0006.jit_prim_fun.before_optimizations.dot
├── module_0006.jit_prim_fun.before_optimizations.hlo.pb
├── module_0006.jit_prim_fun.before_optimizations.html
├── module_0006.jit_prim_fun.before_optimizations.txt
├── module_0006.jit_prim_fun.cpu_after_optimizations-buffer-assignment.txt
├── module_0006.jit_prim_fun.cpu_after_optimizations.dot
├── module_0006.jit_prim_fun.cpu_after_optimizations.hlo.pb
├── module_0006.jit_prim_fun.cpu_after_optimizations.html
├── module_0006.jit_prim_fun.cpu_after_optimizations.txt
├── module_0006.jit_prim_fun.ir-no-opt.ll
├── module_0006.jit_prim_fun.ir-no-opt-noconst.ll
├── module_0006.jit_prim_fun.ir-with-opt.ll
├── module_0006.jit_prim_fun.ir-with-opt-noconst.ll
├── module_0006.jit_prim_fun.o
├── module_0008.jit__normal.before_optimizations.dot
├── module_0008.jit__normal.before_optimizations.hlo.pb
├── module_0008.jit__normal.before_optimizations.html
├── module_0008.jit__normal.before_optimizations.txt
├── module_0008.jit__normal.cpu_after_optimizations-buffer-assignment.txt
├── module_0008.jit__normal.cpu_after_optimizations.dot
├── module_0008.jit__normal.cpu_after_optimizations.hlo.pb
├── module_0008.jit__normal.cpu_after_optimizations.html
├── module_0008.jit__normal.cpu_after_optimizations.txt
├── module_0008.jit__normal.ir-no-opt.ll
├── module_0008.jit__normal.ir-no-opt-noconst.ll
├── module_0008.jit__normal.ir-with-opt.ll
├── module_0008.jit__normal.ir-with-opt-noconst.ll
├── module_0008.jit__normal.o
├── module_0009.jit_fn.before_optimizations.dot
├── module_0009.jit_fn.before_optimizations.hlo.pb
├── module_0009.jit_fn.before_optimizations.html
├── module_0009.jit_fn.before_optimizations.txt
├── module_0009.jit_fn.cpu_after_optimizations-buffer-assignment.txt
├── module_0009.jit_fn.cpu_after_optimizations.dot
├── module_0009.jit_fn.cpu_after_optimizations.hlo.pb
├── module_0009.jit_fn.cpu_after_optimizations.html
├── module_0009.jit_fn.cpu_after_optimizations.txt
├── module_0009.jit_fn.ir-no-opt.ll
├── module_0009.jit_fn.ir-no-opt-noconst.ll
├── module_0009.jit_fn.ir-with-opt.ll
├── module_0009.jit_fn.ir-with-opt-noconst.ll
├── module_0009.jit_fn.o
├── module_0010.jit__lambda_.before_optimizations.dot
├── module_0010.jit__lambda_.before_optimizations.hlo.pb
├── module_0010.jit__lambda_.before_optimizations.html
├── module_0010.jit__lambda_.before_optimizations.txt
├── module_0010.jit__lambda_.cpu_after_optimizations-buffer-assignment.txt
├── module_0010.jit__lambda_.cpu_after_optimizations.dot
├── module_0010.jit__lambda_.cpu_after_optimizations.hlo.pb
├── module_0010.jit__lambda_.cpu_after_optimizations.html
├── module_0010.jit__lambda_.cpu_after_optimizations.txt
├── module_0010.jit__lambda_.ir-no-opt.ll
├── module_0010.jit__lambda_.ir-no-opt-noconst.ll
├── module_0010.jit__lambda_.ir-with-opt.ll
├── module_0010.jit__lambda_.ir-with-opt-noconst.ll
├── module_0010.jit__lambda_.o
├── module_0011.jit_prim_fun.before_optimizations.dot
├── module_0011.jit_prim_fun.before_optimizations.hlo.pb
├── module_0011.jit_prim_fun.before_optimizations.html
├── module_0011.jit_prim_fun.before_optimizations.txt
├── module_0011.jit_prim_fun.cpu_after_optimizations-buffer-assignment.txt
├── module_0011.jit_prim_fun.cpu_after_optimizations.dot
├── module_0011.jit_prim_fun.cpu_after_optimizations.hlo.pb
├── module_0011.jit_prim_fun.cpu_after_optimizations.html
├── module_0011.jit_prim_fun.cpu_after_optimizations.txt
├── module_0011.jit_prim_fun.ir-no-opt.ll
├── module_0011.jit_prim_fun.ir-no-opt-noconst.ll
├── module_0011.jit_prim_fun.ir-with-opt.ll
├── module_0011.jit_prim_fun.ir-with-opt-noconst.ll
├── module_0011.jit_prim_fun.o
├── module_0012.jit_prim_fun.before_optimizations.dot
├── module_0012.jit_prim_fun.before_optimizations.hlo.pb
├── module_0012.jit_prim_fun.before_optimizations.html
├── module_0012.jit_prim_fun.before_optimizations.txt
├── module_0012.jit_prim_fun.cpu_after_optimizations-buffer-assignment.txt
├── module_0012.jit_prim_fun.cpu_after_optimizations.dot
├── module_0012.jit_prim_fun.cpu_after_optimizations.hlo.pb
├── module_0012.jit_prim_fun.cpu_after_optimizations.html
├── module_0012.jit_prim_fun.cpu_after_optimizations.txt
├── module_0012.jit_prim_fun.ir-no-opt.ll
├── module_0012.jit_prim_fun.ir-no-opt-noconst.ll
├── module_0012.jit_prim_fun.ir-with-opt.ll
├── module_0012.jit_prim_fun.ir-with-opt-noconst.ll
├── module_0012.jit_prim_fun.o
├── module_0013.jit_fn.before_optimizations.dot
├── module_0013.jit_fn.before_optimizations.hlo.pb
├── module_0013.jit_fn.before_optimizations.html
├── module_0013.jit_fn.before_optimizations.txt
├── module_0013.jit_fn.cpu_after_optimizations-buffer-assignment.txt
├── module_0013.jit_fn.cpu_after_optimizations.dot
├── module_0013.jit_fn.cpu_after_optimizations.hlo.pb
├── module_0013.jit_fn.cpu_after_optimizations.html
├── module_0013.jit_fn.cpu_after_optimizations.txt
├── module_0013.jit_fn.ir-no-opt.ll
├── module_0013.jit_fn.ir-no-opt-noconst.ll
├── module_0013.jit_fn.ir-with-opt.ll
├── module_0013.jit_fn.ir-with-opt-noconst.ll
├── module_0013.jit_fn.o
├── module_0014.jit_fn.before_optimizations.dot
├── module_0014.jit_fn.before_optimizations.hlo.pb
├── module_0014.jit_fn.before_optimizations.html
├── module_0014.jit_fn.before_optimizations.txt
├── module_0014.jit_fn.cpu_after_optimizations-buffer-assignment.txt
├── module_0014.jit_fn.cpu_after_optimizations.dot
├── module_0014.jit_fn.cpu_after_optimizations.hlo.pb
├── module_0014.jit_fn.cpu_after_optimizations.html
├── module_0014.jit_fn.cpu_after_optimizations.txt
├── module_0014.jit_fn.ir-no-opt.ll
├── module_0014.jit_fn.ir-no-opt-noconst.ll
├── module_0014.jit_fn.ir-with-opt.ll
├── module_0014.jit_fn.ir-with-opt-noconst.ll
├── module_0014.jit_fn.o
├── module_0019.jit_prim_fun.before_optimizations.dot
├── module_0019.jit_prim_fun.before_optimizations.hlo.pb
├── module_0019.jit_prim_fun.before_optimizations.html
├── module_0019.jit_prim_fun.before_optimizations.txt
├── module_0019.jit_prim_fun.cpu_after_optimizations-buffer-assignment.txt
├── module_0019.jit_prim_fun.cpu_after_optimizations.dot
├── module_0019.jit_prim_fun.cpu_after_optimizations.hlo.pb
├── module_0019.jit_prim_fun.cpu_after_optimizations.html
├── module_0019.jit_prim_fun.cpu_after_optimizations.txt
├── module_0019.jit_prim_fun.ir-no-opt.ll
├── module_0019.jit_prim_fun.ir-no-opt-noconst.ll
├── module_0019.jit_prim_fun.ir-with-opt.ll
├── module_0019.jit_prim_fun.ir-with-opt-noconst.ll
├── module_0019.jit_prim_fun.o
├── module_0025.jit__lambda_.before_optimizations.dot
├── module_0025.jit__lambda_.before_optimizations.hlo.pb
├── module_0025.jit__lambda_.before_optimizations.html
├── module_0025.jit__lambda_.before_optimizations.txt
├── module_0025.jit__lambda_.cpu_after_optimizations-buffer-assignment.txt
├── module_0025.jit__lambda_.cpu_after_optimizations.dot
├── module_0025.jit__lambda_.cpu_after_optimizations.hlo.pb
├── module_0025.jit__lambda_.cpu_after_optimizations.html
├── module_0025.jit__lambda_.cpu_after_optimizations.txt
├── module_0025.jit__lambda_.ir-no-opt.ll
├── module_0025.jit__lambda_.ir-no-opt-noconst.ll
├── module_0025.jit__lambda_.ir-with-opt.ll
├── module_0025.jit__lambda_.ir-with-opt-noconst.ll
└── module_0025.jit__lambda_.o
3 directories, 241 files
https://neuronaldynamics.epfl.ch/online/Ch13.S6.html
https://uci-carl.github.io/CARLsim3/ch3_neurons_synapses_groups.html#ch3s2_synapses
Text Only
基于电流 (CUBA) 的描述使用单个突触电流项，而基于电导 (COBA) 的描述使用更复杂的电导方程为每个突触受体类型计算突触电流。 CUBA 和 COBA 电流贡献都受到突触的突触权重的影响。
因为 CUBA 模式的突触电流会持续一个时间步长，而 COBA 模式的突触电流会随着时间衰减，所以从 CUBA 模式切换到 COBA 模式通常需要用户减少突触权重，因为在COBA 模式。
以下两个图是tmp/foo下最大的两个dot文件转换得到的图。
module_0003.jit__threefry_split.cpu_after_optimizations.dot
module_0008.jit__normal.cpu_after_optimizations.dot
