There's more rubbish put out on this debate than on any other heating-related topic I know!
A few facts (please excuse some repetition):
- a building loses heat in DIRECT proportion to the difference in temperature inside to outside (disregard issues like evaporative cooling off damp walls, wind-related effects,etc.)
- a CH system can put heat into a building at a rate determined by radiator sizes, the difference in temperature between water in the radiators and the air in the room. Automatic operation of TRVs will affect heat flows from individual radiators. So will manual adjusment of TRVs and other types of valve.
- the heat source (boiler) raises the temperature of the water in direct proportion to the rate of fuel burn AND will also waste heat through the flue at a rate depending on the fluegas temperature, whether there is airflow through the boiler, condensing or not, and other factors. Depending on the boiler, it may modulate the burn-rate according to setpoint(s), Return temperature or other factors. If it's 'all or nothing', chances are it will cycle on, off and on again at some rate, depending on its own thermostat setpoint and other factors, even though the call-for-heat from the roomstat may be continuous.
Considering all these related and unrelated variables, it's virtually IMPOSSIBLE to predict with any accuracy the EXACT energy consumption of a system.
BUT it should also be obvious that if the burner in the boiler is OFF, then no energy is used at all.
The only issue then is the difference in total energy use between continuous operation at a low roomstat setting and heating from cold (say 10 degrees) to comfortable (say 20 degrees). Considering all the above factors, I suggest that more energy MUST be used, in total, in continuous than in cyclic operation of any given system. The only issue is how much more. The energy lost up the flue ONLY when the burner and fan are operating (the case with most modern boilers) will be a significant cost which will ALWAYS swing costs strongly in favour of two warm-ups from 'cold' per day if there is no actual requirement for heat during the majority of the day in between (say) 0900 and 1700. I believe this would still be true ignoring boiler losses and in all cases, considering only heat-loss from the building as a whole.