If testing on a custom symbol, then cloud agents fall away. So the agents are their own, and you can pull their logs and see in the logs how the history was synced
Custom, proprietary Agents, error-free synchronisation on real ticks. Expert Advisor for replay.
// Критерий оптимизации - сумма входных параметров + сумма нецелых частей секунды каждого тика (в миллисекундах).inputint inFakeRange1 = 0;
inputint inFakeRange2 = 0;
inputint inFakeRange3 = 0;
inputint inFakeRange4 = 0;
inputint inFakeRange5 = 0;
long Sum = inFakeRange1 + inFakeRange2 + inFakeRange3 + inFakeRange4 + inFakeRange5;
voidOnTick()
{
MqlTick Tick;
if (SymbolInfoTick(_Symbol, Tick))
Sum += Tick.time_msc % 1000;
}
doubleOnTester()
{
return((double)Sum);
}
Result.
By the value of the first line, it's immediately clear that the value is nonsense: it can't be that different from the values in the other lines.
And running a single.
99% rubbish in ticks during GA optimisation. Let me know if you were able to reproduce it or not.
You need to look at the optimisation log from the agent on which the nonsense value was issued.
And a test log of the corresponding single pass.
Look at the single value. It is correct and very different from all Optimizer values, not just one. I.e. the error is literally in every line of the Optimizer.
I'll look at the logs. It would be good for you to run it on your own.
2020.02.2010:47:18.839 Core 6 genetic pass (5, 312) returned result 10563025086.000000in0:00:05.4752020.02.2010:47:20.023 Core 2 genetic pass (5, 209) returned result 8885031952.000000in0:00:04.9472020.02.2010:47:23.048 Core 3 genetic pass (5, 240) returned result 10373435790.000000in0:00:05.9842020.02.2010:47:23.366 Core 1 genetic pass (5, 181) returned result 10236426953.000000in0:00:05.6542020.02.2010:47:24.069 Core 4 genetic pass (5, 271) returned result 10597154696.000000in0:00:05.7682020.02.2010:47:24.509 Core 5 genetic pass (5, 296) returned result 10580660389.000000in0:00:05.8532020.02.2010:47:24.613 Core 6 genetic pass (5, 313) returned result 10563025078.000000in0:00:05.7742020.02.2010:47:25.113 Core 2 genetic pass (5, 218) returned result 8885031945.000000in0:00:05.089
Highlighted the correct value. Ran through the log. Only Core1 has the correct value.
Slava: What's in the agent logs? That's where the history sync is written.
Tried different characters as it doesn't always reproduce. Agent Logs at Optimisation.
Core1.
EL 012:57:18.438127.0.0.1 login (build 2323)
ND 012:57:18.455 Tester account info found with currency USD
FO 012:57:18.456 Network 1482 bytes of tester parameters loaded
FF 012:57:18.456 Network 2748 bytes of input parameters loaded
DS 012:57:18.456 Tester expert file added: Experts\Test123.ex5. 12221 bytes loaded
NK 012:57:18.456 Network 1032 bytes of optimized inputs info loaded
PM 012:57:18.527 Tester successfully initialized
GE 012:57:18.527 Network 12 Kb of total initialization data received
JR 012:57:18.528 Tester Intel Core i7-2700 K @ 3.50 GHz, 16301 MB
OE 012:57:18.528 Tester optimization pass 35025 started (batch of 53 tasks)
QQ 012:57:18.529 Symbols CHFJPY.rann_RannForex: symbol to be synchronized
RP 012:57:18.529 Symbols CHFJPY.rann_RannForex: symbol synchronized, 3720 bytes of symbol info received
LF 012:57:18.534 History CHFJPY.rann_RannForex: load 444 bytes of history data to synchronize in0:00:00.001
EK 012:57:18.534 History CHFJPY.rann_RannForex: history synchronized from2018.02.26 to 2020.02.14
FL 012:57:18.537 Ticks CHFJPY.rann_RannForex: ticks synchronization started
ML 012:57:18.538 Ticks CHFJPY.rann_RannForex: load 64 bytes of tick data to synchronize in0:00:00.000
MK 012:57:18.538 Ticks CHFJPY.rann_RannForex: history ticks synchronized from2019.06.03 to 2020.02.14
MG 012:57:18.882 History CHFJPY.rann_RannForex,M1: history cache allocated for740490 bars and contains 469972 bars from2018.02.2600:08 to 2019.05.3123:54
CD 012:57:18.882 History CHFJPY.rann_RannForex,M1: history begins from2018.02.2600:08
QG 012:57:26.327 Tester 35025 OnTester result 6963974890 : passed in0:00:07.795
JQ 012:57:31.481 Tester 5593 OnTester result 10077649365 : passed in0:00:05.153
CE 012:57:32.683127.0.0.1 tester forced to stop
DS 012:57:32.684 Tester 2 of 53 passes processed (2 successfully finished) in0:00:12.948
CS 012:57:32.684 optimize Experts\Test123.ex5 on CHFJPY.rann_RannForex,M1 thread finished
HE 012:57:32.684127.0.0.1 prepare for shutdown
Core2.
OS 012:57:18.438127.0.0.1 login (build 2323)
IG 012:57:18.456 Tester account info found with currency USD
OL 012:57:18.460 Network 1482 bytes of tester parameters loaded
OE 012:57:18.460 Network 2748 bytes of input parameters loaded
EP 012:57:18.460 Tester expert file added: Experts\Test123.ex5. 12221 bytes loaded
GJ 012:57:18.460 Network 1032 bytes of optimized inputs info loaded
FM 012:57:18.532 Tester successfully initialized
QJ 012:57:18.532 Network 12 Kb of total initialization data received
IS 012:57:18.532 Tester Intel Core i7-2700 K @ 3.50 GHz, 16301 MB
OG 012:57:18.532 Tester optimization pass 115298 started (batch of 18 tasks)
IQ 012:57:18.533 Symbols CHFJPY.rann_RannForex: symbol to be synchronized
KP 012:57:18.533 Symbols CHFJPY.rann_RannForex: symbol synchronized, 3720 bytes of symbol info received
KE 012:57:18.545 History CHFJPY.rann_RannForex: load 444 bytes of history data to synchronize in0:00:00.002
FD 012:57:18.545 History CHFJPY.rann_RannForex: history synchronized from2018.02.26 to 2020.02.14
DK 012:57:18.546 Ticks CHFJPY.rann_RannForex: ticks synchronization started
PO 012:57:18.547 Ticks CHFJPY.rann_RannForex: load 64 bytes of tick data to synchronize in0:00:00.000
HJ 012:57:18.547 Ticks CHFJPY.rann_RannForex: history ticks synchronized from2019.06.03 to 2020.02.14
JH 012:57:18.856 History CHFJPY.rann_RannForex,M1: history cache allocated for740490 bars and contains 469972 bars from2018.02.2600:08 to 2019.05.3123:54
PS 012:57:18.856 History CHFJPY.rann_RannForex,M1: history begins from2018.02.2600:08
EF 012:57:25.613 Tester 115298 OnTester result 14612465132 : passed in0:00:07.073
GP 012:57:30.797 Tester 26858 OnTester result 9858144627 : passed in0:00:05.184
DR 012:57:32.684127.0.0.1 tester forced to stop
NR 012:57:32.684 Tester 2 of 18 passes processed (2 successfully finished) in0:00:12.257
CP 012:57:32.684 optimize Experts\Test123.ex5 on CHFJPY.rann_RannForex,M1 thread finished
GF 012:57:32.685127.0.0.1 prepare for shutdown
Slava: And on one of these Agents did you get the wrong results?
I didn't have any logs of that situation, as I had forbidden to write logs. So I had to start another copy of the Terminal from scratch and repeat the experiment, where the logs are all written.
Above are the Optimization logs. There each Agent counted two passes, after which the Stop button was pressed, because OnTester-results are rambling for each of the Agents.
The logs of the two Agents above. I will try to create a situation that you can reproduce for yourself.
I didn't have any logs of that situation, as I had forbidden to write logs. So I had to start another copy of the Terminal from scratch and repeat the experiment, where the logs are all written.
Above are the Optimization logs. There each Agent counted two passes, after which the Stop button was pressed, because OnTester-results are rambling for each of the Agents.
The logs of the two Agents above. I'll try to create a situation that you can reproduce for yourself.
That's exactly what I'm asking about the Agents logs. Were the results wrong for both agents?
I need to see agent logs with correct results and agent logs with incorrect results
Testing/optimisation on real ticks?
Own Agents or from the cloud?
If testing on a custom symbol, then cloud agents fall away. So the agents are their own, and you can pull their logs and see in the logs how the history was synced
Custom, proprietary Agents, error-free synchronisation on real ticks. Expert Advisor for replay.
Result.
By the value of the first line, it's immediately clear that the value is nonsense: it can't be that different from the values in the other lines.
And running a single.
99% rubbish in ticks during GA optimisation. Let me know if you were able to reproduce it or not.
Search string: Oshibka 005.Custom, own Agents, error free synchronisation on real ticks. EA for playback.
Result.
By the value of the first line, it's immediately clear that the value is nonsense: it can't be that different from the values in the other lines.
And running a single.
99% rubbish in ticks during GA optimisation. Let me know if you were able to reproduce it or not.
Search string: Oshibka 005.Have to look at the optimization log from the agent on which the nonsense value was issued.
And the test log of the corresponding single pass.
You need to look at the optimisation log from the agent on which the nonsense value was issued.
And a test log of the corresponding single pass.
Look at the single value. It is correct and very different from all Optimizer values, not just one. I.e. the error is literally in every line of the Optimizer.
I'll look at the logs. It would be good for you to run it on your own.
Highlighted the correct value. Ran through the log. Only Core1 has the correct value.Ran through the log. Only Core1 has the correct value.
Completely localised. Single pass on Core1 does not match single pass on Core2. So I can't tell which one is correct yet.
Completely localised. The single pass on Core1 is not the same as the single pass on Core2. So I can't say which one is correct yet.
What's in the agent logs? That's where the history sync is written.
Tried different characters as it doesn't always reproduce. Agent Logs at Optimisation.
Core1.
Core2.
Tried different characters as it doesn't always reproduce. Agent Logs at Optimisation.
Core1.
Core2.
And on one of these Agents did you get the wrong results?
I didn't have any logs of that situation, as I had forbidden to write logs. So I had to start another copy of the Terminal from scratch and repeat the experiment, where the logs are all written.
Above are the Optimization logs. There each Agent counted two passes, after which the Stop button was pressed, because OnTester-results are rambling for each of the Agents.
The logs of the two Agents above. I will try to create a situation that you can reproduce for yourself.
I didn't have any logs of that situation, as I had forbidden to write logs. So I had to start another copy of the Terminal from scratch and repeat the experiment, where the logs are all written.
Above are the Optimization logs. There each Agent counted two passes, after which the Stop button was pressed, because OnTester-results are rambling for each of the Agents.
The logs of the two Agents above. I'll try to create a situation that you can reproduce for yourself.
That's exactly what I'm asking about the Agents logs. Were the results wrong for both agents?
I need to see agent logs with correct results and agent logs with incorrect results
I'm asking specifically about the agents' logs. Were the results wrong for both agents?
I need to see agent logs with correct results and agent logs with incorrect results
Both Agents had the wrong results for the optimisation passes. I almost managed to reproduce it on a clean Terminal... I'll post it in a while.
By the way, what's the reason for the Agents still hanging in memory after closing the Terminal?